Packaging method and device for optical fiber array

Abstract

The application discloses a packaging method of an optical fiber array, and relates to the technical field of optical packaging. The application reversely calculates and reversely superimposes the optical fiber center drift amount caused by resin curing shrinkage, thermal expansion mismatch and other factors into the design profile of a mold cavity, so that the optical fiber array after molding and cooling is self-converged to a target interval in a free state. Meanwhile, the optical fiber positioning groove and the polarization maintaining stress rod positioning groove are integrated by means of a complex cavity mold, and the optical fiber and the stress element are fixed by embedding and fixing the optical fiber and the stress element by using ultraviolet curing resin at one time, so that the complicated assembly process of splicing and bonding of multiple ceramic ferrules is replaced. The application eliminates the splicing positioning deviation and the interval drift risk caused by the mismatch of the thermal expansion coefficients of materials under thermal cycling, greatly reduces the single-piece manufacturing cost by using the reusable metal mold and the low-cost liquid resin, and meets the requirements of high precision, high reliability, low cost and batch production.

Description

Technical Field

[0001] This invention relates to the field of optical packaging technology, specifically to a method for packaging fiber arrays. Background Technology

[0002] An optical fiber array is an optical coupling device that arranges and fixes multiple optical fibers in a precise spatial relationship. It is widely used in optical circuit switches, photonic integrated chip packaging and wavelength division multiplexing systems as an input / output coupling interface for micro-mirror arrays or optical waveguide arrays.

[0003] A typical manufacturing process for fiber optic arrays includes: first, fabricating precision channel structures on a substrate material to accommodate and position the optical fibers; then, placing bare optical fibers stripped of their coatings one by one into the corresponding channels; finally, fixing the fibers in the channels with adhesives or encapsulating materials; and finally, grinding and polishing the array end faces to obtain an optical-grade exit surface. In this process, the relative spatial position accuracy between the optical fibers is the core indicator that determines the optical coupling efficiency of the device. In particular, the consistency of the center spacing between adjacent optical fibers directly affects the alignment accuracy of the light spot with subsequent optical path nodes and the long-term operational reliability of the link.

[0004] As stated in the published patent "A fiber array manufacturing method and manufacturing equipment" with announcement number CN121741941B, "In the prior art, a single substrate V-groove processing scheme is adopted, but due to the limitations of photolithography process resolution, substrate material stress release and etching uniformity, the cumulative error increases linearly with the number of channels, and the pitch error of a 128-channel substrate is difficult to meet the technical requirements."

[0005] In the existing technology, the ceramic ferrule combination scheme is adopted. High-density arrangement is achieved by splicing multiple small-channel ceramic ferrules. Although it can reduce the processing error of a single ferrule, there is a positioning deviation during the splicing process. In addition, the thermal expansion coefficients of ceramic materials and optical fibers do not match, which can easily lead to the drift of the center-to-center distance between adjacent optical fibers after temperature cycling. At the same time, the assembly process of this scheme is complicated and the cost remains high.

[0006] For medium- and low-density fiber arrays such as 8-channel, 16-channel, and 32-channel arrays, single-substrate V-groove etching or small ceramic ferrule assembly can basically meet the accuracy requirements. However, when the number of channels expands to 64 channels and above, existing technologies reveal shortcomings in terms of accuracy, reliability, and mass production feasibility. On the one hand, the single-substrate solution is limited by photolithography resolution, substrate material stress release, and etching uniformity. The cumulative error of the positioning groove increases linearly with the number of channels, leading to unacceptable positional deviations of edge channels. On the other hand, while the ceramic ferrule assembly solution can avoid the bottleneck of large-size single-piece processing, the positioning deviation of multiple pieces splicing and the mismatch of thermal expansion coefficients between ceramic and fiber materials can easily cause spacing drift after temperature cycling. Moreover, the assembly process is cumbersome and the manufacturing cost is high. In addition, the positioning of stress bars for polarization-maintaining fibers still relies on manual or semi-automatic operation, resulting in low alignment efficiency and poor consistency.

[0007] In summary, the existing technology has the following pressing technical problems: how to simultaneously meet the high precision requirements of the center spacing between adjacent optical fibers in the manufacturing of high-density fiber arrays and ensure long-term reliability under temperature cycling conditions. Summary of the Invention

[0008] To overcome the shortcomings mentioned above, this invention aims to provide a packaging method for fiber optic arrays that can solve the aforementioned problems.

[0009] To achieve the above objectives, the present invention provides the following technical solution: A method for packaging fiber optic arrays, used in the manufacture of fiber optic arrays, includes the following steps: S1: Based on the target number of fiber channels N and the fiber core spacing P, determine the arrangement array of the mold cavity, wherein each unit cavity includes a main fiber receiving slot and positioning slots symmetrically arranged on both sides of the main receiving slot. The main receiving slot and the positioning slot are connected by a current limiting neck with a gradually changing width. An isolation boss is reserved between the positioning slots of adjacent unit cavities. S2: Perform thermo-mechanical coupling pre-distortion compensation on the nominal dimensions of the unit cavity, including obtaining the curing shrinkage rate and thermal expansion coefficient of the photocurable molding resin used, as well as the thermal mismatch between the resin and the optical fiber material, calculating the drift of the center spacing between adjacent optical fibers after molding through finite element simulation, and superimposing the compensation obtained by inverse calculation onto the main accommodating slot spacing and the positioning slot spacing to generate the compensated three-dimensional curved surface profile of the cavity. S3: Coating multiple layers of positive and negative photoresist on a silicon substrate, and forming a multi-depth stepped prototype with the three-dimensional curved surface contour of the cavity by grayscale photolithography exposure and alternating dry and wet etching, wherein the depth of the main accommodating groove is greater than the depth of the positioning groove, and the bottom of the main accommodating groove is provided with a capillary guiding micro-groove extending along the optical fiber axis. S4: The original mold is made into a metal mold core by electroforming, and an organic fluorine hydrophobic self-assembled monolayer is deposited on the cavity surface of the metal mold core as a release layer to obtain a complex cavity mold for ultraviolet embossing. S5: Fill the complex cavity mold with photocurable resin, place the bare optical fiber with the coating removed and the optical fiber array in parallel into the corresponding main receiving groove and positioning groove, and then cure by ultraviolet light to obtain a high-density optical fiber array that combines optical fiber positioning and stress element embedding in one step.

[0010] A fiber optic array encapsulation device, wherein the encapsulation device is operated using the method described above.

[0011] Compared with the prior art, the beneficial effects of the present invention are as follows: By introducing a cavity pre-distortion compensation mechanism based on thermo-mechanical coupling finite element simulation, this invention actively calculates and superimposes the fiber center drift caused by factors such as resin curing shrinkage and thermal expansion mismatch into the design contour of the mold cavity. This allows the fiber array after molding and cooling to converge to the target spacing in a free state. At the same time, the fiber positioning groove and the polarization-maintaining stress rod positioning groove are integrated into one piece using a complex cavity mold, and the fiber and stress element are embedded and fixed at one time using UV-cured resin. This replaces the cumbersome assembly process of splicing and bonding multiple ceramic ferrules. It eliminates the risk of splicing positioning deviation and spacing drift caused by the mismatch of material thermal expansion coefficients under thermal cycling. Furthermore, the use of reusable metal molds and low-cost liquid resin significantly reduces the unit manufacturing cost, meeting the requirements of high precision, high reliability, low cost, and mass production. Attached Figure Description

[0012] Figure 1 This is a cross-sectional structural diagram of a fiber optic array in the prior art; Figure 2 This is a three-dimensional view of the nominal structure of the mold cavity and the structure of the bare optical fiber and stress bar in this invention; Figure 3 This is a cross-sectional view of the nominal structure of the mold cavity in this invention, along with the bare optical fiber and stress bar; Figure 4 This is another cross-sectional view of the nominal structure of the mold cavity in this invention, along with the bare optical fiber and stress bar; Figure 5 yes Figure 4 A partial view at point A in the middle; Figure 6 This is a top view of the nominal structure of the mold cavity in this invention; Figure 7 yes Figure 6 A partial view at point B in the middle; Figure 8 This is a schematic diagram of the mold cavity structure in this invention; The reference numerals and names in the figure are as follows: Unit cavity-000, bare optical fiber-001, stress bar-002, closed-loop overflow groove-005, overflow weir-006, main receiving groove-100, positioning groove-200, flow-limiting neck-300. Detailed Implementation

[0013] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0014] Please see Figure 1-8 A method for packaging fiber optic arrays, used in the manufacture of fiber optic arrays, includes the following steps: S1: Determine the arrangement array of mold cavities according to the target number of fiber channels N and fiber core spacing P, wherein each unit cavity includes a main fiber receiving slot and stress bar positioning slots symmetrically arranged on both sides of the main receiving slot. The main receiving slot and the positioning slot are connected by a current limiting neck with gradually changing width. An isolation boss is reserved between the positioning slots of adjacent unit cavities. S2: Perform thermo-mechanical coupling pre-distortion compensation on the nominal (preset) dimensions of the unit cavity, including obtaining the curing shrinkage rate and thermal expansion coefficient of the photocurable molding resin used, as well as the thermal mismatch between the resin and the optical fiber material, calculating the drift of the center spacing between adjacent optical fibers after molding through finite element simulation, and superimposing the compensation obtained by inverse calculation onto the main accommodating slot spacing and the positioning slot spacing to generate the compensated three-dimensional curved surface profile of the cavity. S3: Coating multiple layers of positive and negative photoresist on a silicon substrate, and forming a multi-depth stepped prototype with the three-dimensional curved surface contour of the cavity by grayscale photolithography exposure and alternating dry and wet etching, wherein the depth of the main accommodating groove is greater than the depth of the positioning groove, and the bottom of the main accommodating groove is provided with a capillary guiding micro-groove extending along the optical fiber axis. S4: The original mold is made into a metal mold core by electroforming, and an organic fluorine hydrophobic self-assembled monolayer is deposited on the cavity surface of the metal mold core as a release layer to obtain a complex cavity mold for ultraviolet embossing. S5: Fill the complex cavity mold with photocurable resin, place the bare optical fiber 001 with the coating removed and the stress bar 002 in parallel into the corresponding main receiving groove and positioning groove, and then cure by ultraviolet light to obtain a high-density optical fiber array that combines optical fiber positioning and stress element embedding in one step. This invention uses ultraviolet embossing to form a high-density fiber array in one step, while simultaneously achieving precise fiber positioning and embedding of stress elements. This invention abandons the traditional approach of V-groove etching and stacking or ceramic ferrule splicing, and instead integrates precision mechanical positioning, thermo-mechanical coupling pre-deformation compensation, and self-alignment structure into a reusable mold cavity, thereby systematically solving the contradiction between accuracy, reliability, and mass production under high channel count. The working principle is explained in detail step by step: like Figure 2 As shown, in S1, a cavity unit configuration is designed, with an independent positioning and embedding unit cavity for each optical fiber. This unit is not a simple V-groove, but includes a "main receiving groove" to accommodate the bare optical fiber, and "stress bar positioning grooves" symmetrically located on both sides of the "main receiving groove". The two are connected by a "gradually wide current-limiting neck". The main receiving groove and the positioning groove realize the precise relative positional relationship between the optical fiber and the stress bar, making the stress applied by the stress bar to the fiber core symmetrical and controllable. During the subsequent UV imprinting and resin filling, the narrow neck can generate a capillary resistance effect, controlling the resin to first fill and surround the optical fiber from the main receiving groove, and then slowly wet the positioning groove, avoiding air bubbles and stress bar displacement. The isolation boss is a boss left between the positioning grooves of adjacent channels, which serves as a physical isolation to prevent resin overflow from causing crosstalk between adjacent channels, while also increasing the local stiffness of the mold and reducing deformation during imprinting. It should be noted that "imprinting" here refers to the molding process (i.e., step S5) in ultraviolet nanoimprint lithography. This means using a pre-made complex cavity mold (mold core) as a "stamp" to press the desired shape onto liquid photocurable resin, and then using ultraviolet light to instantly cure and set it. First, resin filling is performed: liquid photocurable resin is dripped or coated into the mold cavity where the optical fiber has already been placed. This resin is liquid at room temperature, has good fluidity, and can penetrate every micro-corner of the mold (such as capillary microgrooves and flow-limiting necks). Then, imprinting (the key action) is performed: a transparent cover plate (such as quartz glass) is pressed onto the liquid resin. This "pressing" action has two core functions: first, shaping: the extrusion pressure causes the liquid resin to completely fill the complex cavity of the mold, precisely shaping all the micro- and nano-structures on the mold (V-grooves, flow-limiting necks, stepped depths, etc.). The first step is accurate replication; the second is positioning and fixing: under pressure, the pre-placed optical fibers and stress bars are firmly pressed into their respective designed positions in the mold (main receiving groove and positioning groove) to prevent them from floating or misaligning during subsequent curing; at this time, the liquid resin completely encapsulates the optical fibers and stress bars; UV curing and shaping refers to irradiating the liquid resin with ultraviolet light through a transparent cover plate or the back of the mold while maintaining pressure; the resin undergoes a photochemical reaction, changing from liquid to solid within seconds to tens of seconds; demolding refers to releasing the pressure and removing the cured solid resin part containing the optical fiber array from the metal mold core; this is where the "organic fluorine hydrophobic self-assembled monolayer as a demolding layer" mentioned in step S4 comes into play; it ensures that the solid resin does not adhere to the mold, allowing for clean and complete demolding, and the mold cavity is not contaminated by residues, making it immediately usable for the next stamping; It's important to clarify why this technical solution uses "imprinting" instead of "injection molding" or "casting." Firstly, imprinting offers high precision: it doesn't rely on resin self-leveling but rather uses mechanical pressure to force bonding, achieving nanometer-level precision in replicating the mold structure—crucial for micrometer-pitch fiber arrays. Secondly, it's a low-temperature process: the entire process is conducted at or slightly above room temperature, without high-temperature melting and cooling, thus minimizing thermal shrinkage and ensuring dimensional accuracy (thermo-mechanical coupling compensation primarily addresses the resin's chemical shrinkage and slight thermal mismatch, rather than drastic temperature changes). Thirdly, it offers structural inclusion: it perfectly encapsulates pre-placed bare optical fibers and stress bars that cannot withstand high-temperature, high-pressure injection molding, preventing damage to the fibers. Finally, it ensures mold durability: the low pressure and low temperature result in an extremely long lifespan for the metal mold core, allowing for tens of thousands of repeated imprints, enabling low-cost, mass production. Therefore, the mold manufactured in step S4 is the tool used for this imprinting process, while step S5 describes the imprinting process itself. In S2, the thermo-mechanical coupling pre-distortion compensation is a core step to ensure high-precision positioning. During UV curing, the resin undergoes volume shrinkage, and the difference in thermal expansion coefficients between the resin, quartz fiber, and stress bar introduces internal stress into the molded array, causing the center-to-center spacing of adjacent fibers to drift. This step does not aim for "zero-shrinkage" materials but instead employs a "pre-reverse compensation" strategy. Specific operations include: first, obtaining the resin's curing shrinkage rate, thermal expansion coefficient, and thermal mismatch; then, using finite element simulation software, performing thermo-mechanical coupling analysis on the entire molding, demolding, post-curing, and temperature cycling process to calculate the final drift vector of the fiber center at each location; next, directly superimposing the inverse vector of the drift onto the nominal design size of the mold cavity; for example, if the simulation predicts that a certain fiber will drift 0.3 μm to the right, then the cavity position at that location will be shifted 0.3 μm to the left during mold design; finally, a three-dimensional model with a compensation surface is generated, ensuring that the array size after molding and cooling converges precisely to the target value. It is important to note that the amount of drift varies at different locations. The fibers at the outermost edge of the array are relatively free, unconstrained by other fibers and resin. When the resin cures and shrinks, the shrinkage resistance at the edge is minimal, resulting in a significant inward drift towards the array center. The fiber at the very center of the array is surrounded by other curing resin and fibers. This high degree of symmetrical constraint restricts the macroscopic shrinkage of the resin, converting it into tensile stress within the resin rather than causing fiber displacement. Therefore, the center-to-center drift in the central region is usually small, but residual stress may exist at the resin-fiber interface. In actual UV imprinting, UV light shines through the cover plate. If the illumination is not completely uniform, or if the resin thickness varies in different areas (affected by the depth of the stepped groove), the resin curing reaction will be asynchronous. When the central region cures and hardens first, the surrounding areas may not yet be cured or are still shrinking. The rigid center that cures first will prevent the free shrinkage of the later-cured areas, resulting in a complex, non-uniform strain distribution. UV irradiation is accompanied by thermal radiation. The heat dissipation rate of the cover plate and the metal mold core edge is faster than that of the center; during the curing and cooling to room temperature process, the center and edge of the array will experience different temperature histories, and the thermal expansion / contraction will naturally be different; the difficulty of the V-groove lies in the fact that "the cumulative error increases linearly with the number of channels". Although this solution uses pre-compensation to solve this problem, from the 1st channel to the Nth channel, the drift is not a linear relationship with equal increments for each channel; it is more like a spring network: assuming the array has 64 channels, starting from the left edge, the 1st fiber drifts Δ1 towards the center, and this force is transmitted through the resin to the 1st fiber... Two optical fibers, the second one drifts by Δ2... to the central region, the contraction stress on the left and right sides cancels each other out, and the drift approaches zero; therefore, the drift amount in spatial distribution usually presents a parabola with large values ​​at both ends and small values ​​in the middle or a more complex gradient curve; it is precisely because the drift amount is different at different positions that the independent application of compensation to the coordinates of each unit cavity in step S2 is creative; it does not assume that the drift is uniform, but rather calculates the exact drift vector of each channel through simulation, and then corrects it point by point in reverse, thereby accurately resolving the non-uniform deformation of the entire array; The principle behind the multi-depth stepped prototype fabrication in S3 is as follows: Traditional V-grooves have a single depth, which cannot achieve differentiated functions for fiber maintenance and stress bar positioning. This step uses grayscale photolithography to create a complex three-dimensional prototype with multiple depth levels on a silicon substrate. First, grayscale photolithography and alternating etching are performed: grayscale photolithography can produce different exposure depths in different areas. Combined with the selection of positive / negative photoresists and alternating dry / wet etching, a stepped contour with "deeper main accommodating grooves and shallower positioning grooves" can be precisely processed. The depth of the main accommodating groove must completely accommodate the bare fiber and leave sufficient resin cladding. The depth of the positioning groove only needs to match the radius of the stress bar to achieve semi-embedded positioning. The capillary guiding microgrooves refer to the microgrooves added at the bottom of the main accommodating groove along the fiber axis. Their function is to guide the resin to spread rapidly along the fiber axis during resin filling using capillary action, expelling air towards the front end, thereby completely eliminating encapsulated air bubbles at the bottom of the fiber and improving the uniformity of the optical interface after curing. In S4, the principle of metal mold core fabrication and demolding is as follows: Silicon-based original molds are fragile and cannot be directly used for mass imprinting; through electroforming processes (such as electroforming nickel), the micro-nano structures on the original mold are replicated in high fidelity onto the metal (nickel) mold core to obtain a durable mold with high hardness and high thermal conductivity; furthermore, an organic fluorine self-assembled monolayer (such as fluorosilane) is deposited on the cavity surface of the metal mold core; this monolayer is only nanometer thick, does not change the cavity size, but has extremely low surface energy, forming a permanent hydrophobic demolding layer; the resin cured after imprinting can be easily and completely separated from the mold, avoiding the problems of traditional demolding agent residue affecting accuracy and contaminating the fiber end face, ensuring the consistency of the dimensions of each mold product; The S5 achieves integrated UV imprinting molding and self-alignment, combining a complex cavity mold as both a "precision fixture" and an "encapsulation mold" to achieve one-step molding: first, liquid photocurable resin is filled into the mold cavity, then the bare optical fiber (with the coating removed) and stress rod are placed parallel to each other into their respective slots; because the positioning slots use a non-circular or semi-enclosed cross-section that matches the cross-section of the stress rod (such as... Figure 5 As shown in the diagram (ellipse plus circle), when the stress bar is pressed in or lifted into the positioning groove by the resin buoyancy, its geometric shape will force it to rotate to a unique and stable orientation, completely replacing manual or semi-automatic monitoring and adjustment; finally, overall curing is performed: the transparent cover is closed and ultraviolet light is used to irradiate it, and the resin cures within a few seconds, forming the optical fiber, stress bar and array structure as one piece; after demolding, a high-density optical fiber array with the center spacing of the optical fibers pre-compensated and each component firmly embedded is obtained. It's important to note that the UV-cured resin undergoes a slight volume shrinkage when exposed to ultraviolet light (the shrinkage that needs to be compensated for in S2). This shrinkage process naturally generates shear stress at the interface between the cover plate and the cured resin, causing microscopic slippage, essentially "loosening" them prematurely. When external force is applied to remove them, they can be separated completely without damaging the embedded fiber array. Secondly, the cover plate is subjected to low contact pressure, preventing strong adhesion. In step S5, the cover plate is "pressed" onto the liquid resin, but this pressure is controlled very low (generally...). The resin is typically applied at a pressure on the order of 0.1 bar. The purpose is to fill the cavity with resin and compact the optical fiber, rather than pressing the resin and cover plate together like strong adhesive. Furthermore, similar to the "deposition of an organic fluorine hydrophobic self-assembled monolayer as a release layer" in the metal mold core in S4, the quartz glass cover plate is also deposited with an organic fluorine self-assembled monolayer on its surface by vapor deposition before imprinting. This monolayer is only a few nanometers thick and does not affect light transmission, but its surface energy becomes extremely low. After the resin cures, there are only very weak van der Waals forces between it and the cover plate, allowing for very easy separation.

[0015] In this embodiment of the invention, step S1 specifically includes the following steps: Based on the target number of fiber channels N, the cavity arrangement of the mold is determined to be a one-dimensional linear array, and the center distance between adjacent unit cavities is strictly equal to the fiber core spacing P. Obtain the bare fiber diameter Df, stress bar diameter Ds, and center distance Ls of the two stress bars of the target optical fiber. Design the bottom width of the main accommodating groove in the unit cavity as Df-δ, where δ=1~3um. In two symmetrically distributed V-shaped positioning grooves with their openings facing the main accommodating groove, make the symmetry plane of the two V-shaped positioning grooves coincide with the central axis of the main accommodating groove, and make the center distance between the two V-shaped positioning grooves equal to Ls. The included angle of the V-shaped grooves is 60°~120°. The depth Hf of the main accommodating groove is set to (0.6~0.8)Df, and the depth Hs of the positioning groove is set to (0.4~0.6)Ds, so that after the optical fiber and stress rod are inserted, a portion of their height protrudes from their respective groove openings. The flow-limiting neck connecting the main receiving groove and the positioning groove is set such that the width gradually decreases from the main receiving groove to the positioning groove, and the width W2 at the end of the neck on the positioning groove side satisfies W2 < Ds. The width Wt of the isolation boss reserved between the positioning slots of adjacent unit cavities is determined to be ≥10μm, and the top surface of the isolation boss is not lower than the top surface of the slot opening of the positioning slot. The sole purpose of the mold cavity is to precisely fix multiple independent optical fibers in a specific spatial position in a reproducible and batch manner to form an optical fiber array; this array ultimately needs to be docked and coupled with a micro-mirror array or photonic integrated circuit. Therefore, in the technical solution of this invention, the center distance between adjacent unit cavities is strictly equal to the fiber core spacing P: In the design of step S1, each bare optical fiber is accommodated in the main receiving slot of the unit cavity; the geometric center of the main receiving slot determines the spatial coordinates of the optical fiber after it is placed and fixed by resin; this is a one-to-one direct mapping relationship; secondly, the target that must be met: the design specifications of the optical circuit switch clearly stipulate that the output spot spacing of adjacent optical fibers must be completely matched with the spacing of the next stage micro-mirror unit; this spacing is the fiber core spacing P; therefore, the center distance between adjacent units of the mold cavity, which serves as its spatial positioning reference, must be strictly equal to the target value P in the initial design; if the design reference deviates from P, the final product will inevitably deviate, and the coupling efficiency will be out of the question; Meanwhile, the center distance between adjacent unit cavities is strictly equal to the fiber core spacing P, providing a unique and clear nominal reference for thermo-mechanical coupling pre-distortion compensation. The most creative core of the entire technical solution of this invention lies in the "pre-distortion compensation" in step S2. The success of this compensation process depends entirely on a clear and unambiguous initial reference system. Compensation is a reverse correction of the deviation: the logic of compensation is to first calculate how much the fiber center will drift from "a certain original position" after molding, and then move the mold cavity in the opposite direction by the same amount. This "original position" is the nominal coordinate; the grid points of the nominal coordinate have a row and column spacing of P. Simultaneously, a closed loop of "design-simulation-compensation" is formed: Initial design: The cavity center is at coordinates (Xi, Yj), satisfying X{i+1}-Xi=P; Simulation prediction: Simulation results show that after molding, the center of the optical fiber located in this cavity actually drifts to (Xi+Δx,Yj+Δy); Compensation implementation: The center of this cavity is modified to (Xi -Δx,Yj -Δy), and simulation is performed again to verify whether the center of the optical fiber after molding has returned to (Xi,Yj); If the initial design is not a regular grid that is strictly equal to P, then the "drift" loses a unified reference benchmark, and the compensation direction and magnitude of each point cannot be systematically reversed, and the entire pre-distortion compensation methodology cannot be established; like Figure 2 As shown, the array arrangement and unit cavity period are first determined. Based on the target number of fiber channels N and the fiber core spacing P, a one-dimensional linear array is preferred. like Figure 3 As shown, the optical fiber geometric parameters are collected and the relative positions of the main receiving slot and the positioning slot are designed. First, the precise parameters of the optical fiber used are obtained: bare fiber diameter Df, stress bar diameter Ds, and center distance Ls between the two stress bars. Based on these parameters, the following design is carried out: Main accommodating slot cross-section design: designed as an isosceles trapezoid or rectangle, with a bottom width Wf=Df-δ, where δ is the accommodation tolerance, ranging from 1 to 3 μm; the side wall inclination angle is 3° to 8° to facilitate demolding and fiber embedding; like Figure 4 As shown, the stress bar positioning groove cross-section design: two positioning grooves 200 are set in each unit cavity 000, and the two positioning grooves 200 are symmetrically distributed on the left and right sides of the main receiving groove 100; as Figure 5 As shown, the cross-section of the positioning groove is non-circular, preferably V-shaped or arc-shaped, with its opening facing the main receiving groove, and the included angle θ of the V-shaped groove satisfies 60°≤θ≤120°; the axis of symmetry of the positioning groove coincides with the vertical central axis of the main receiving groove, such that the center distance Ld between the two positioning grooves is strictly equal to Ls, as shown. Figure 4 As shown, when the stress bar falls into the positioning groove, its outer cylindrical surface forms a double-line contact with the V-shaped side walls. The geometric constraint forces the stress bar to rotate until its axis is aligned with the groove length direction, and automatically maintains the symmetrical distribution of the two stress bars relative to the fiber core. Interconnection: The main receiving slot and each positioning slot are connected by a current-limiting neck 300, which extends from the opening on the side wall of the main receiving slot to the opening of the positioning slot; the unit cavity 000 is designed with multiple levels of slot depth to construct a stepped cavity functional surface. To simultaneously achieve stable accommodation of optical fibers and semi-embedded positioning of stress rods, the slot depth design adopts a differentiated multi-depth scheme: First, the main receiving groove depth Hf = (0.6~0.8) × Df; the groove depth is greater than the fiber radius but less than the diameter, so that about 30%~40% of the fiber height protrudes from the groove after it is embedded, ensuring that the transparent cover can press and position it during subsequent imprinting, and that the resin completely covers the lower half of the fiber. The positioning groove depth Hs = (0.4~0.6) × Ds; after the stress rod is placed, the upper part protrudes from the positioning groove. In the subsequent S5 step, by lightly pressing the protruding part of the stress rod with the cover plate or an external positioning block, the stress rod can be driven to sit completely in the V-shaped groove and complete precise self-alignment. The protruding part can ensure that the stress rod is completely sealed by the resin filled later. The bottom of the main receiving groove is deeper than that of the positioning groove, forming a stepped difference. This depth difference provides the basis for the exposure depth modulation of grayscale lithography in S3. A flow-limiting neck with a gradually changing width is constructed between the main receiving groove and the positioning groove to control the resin flow sequence. The flow-limiting neck connecting the main receiving groove and the positioning groove is designed with a width that decreases non-linearly or linearly monotonically from the main receiving groove to the positioning groove. The width of the neck at the inlet side (near the main receiving groove) is W1 = (0.3~0.6) × Ds; the width of the neck at the outlet side (near the positioning groove) is W2 = (0.1... 5~0.4)×Ds; the neck length is Ln=(0.5~1.5)×D_s; this gradient structure forms a capillary valve effect: the liquid photocurable resin preferentially fills the main accommodating groove with a larger space and surrounds the optical fiber. Because the narrowest part of the neck is much smaller than the diameter of the stress rod, the resin flow resistance increases sharply, thereby slowing down the speed at which the resin enters the positioning groove. Only after the main accommodating groove is completely filled and the optical fiber is fixed can the resin gradually wet the positioning groove and embed the stress rod; this orderly filling process can prevent air bubble trapping and stress rod displacement. At the same time, it is necessary to determine the size of the isolation boss to ensure physical isolation between channels; in adjacent unit cavities, the remaining base material between two relative positioning grooves constitutes the isolation boss; it is necessary to ensure that Wt≥10μm to ensure the mechanical strength of the mold and effectively prevent resin overflow to adjacent channels; the top surface of the isolation boss is slightly higher than the top surface of the positioning groove opening, and serves as a contact support surface during imprinting to prevent the mold microstructure from collapsing under pressure; Through the above steps, the deterministic design of all key geometric features of the complex cavity unit is completed, resulting in a nominal three-dimensional model that can be directly used for thermo-coupling compensation in S2, ensuring the integrity, operability, and uniqueness of the design.

[0016] In this embodiment of the invention, step S2 specifically includes the following steps: Obtain the curing shrinkage rate, coefficient of thermal expansion, Young's modulus and Poisson's ratio of the photocurable resin used, as well as the coefficient of thermal expansion and elastic modulus of the optical fiber and stress rod materials, and calculate the thermal mismatch between the resin and the optical fiber. Based on the nominal cavity profile generated in step S1, a three-dimensional finite element parametric model including mold, resin, optical fiber and stress bar is established. In the model, the optical fiber, stress bar and resin domain are bound together, and the outer surface of the mold is completely fixed. Three load steps are applied sequentially to the model: isothermal curing shrinkage, cooling to room temperature, and demolding release. The curing shrinkage is equivalent to the resin by giving it a virtual thermal shrinkage coefficient, while the cooling process uses the real thermal expansion coefficient. Calculate the center coordinates of each fiber after demolding, extract the actual center distance in the row or column direction of all adjacent fiber pairs, and calculate the drift amount ΔP between them and the nominal fiber core spacing P. Invert the drift amount corresponding to each fiber position, translate the corresponding unit cavity center coordinates, keep the feature size in the slot unchanged, and generate the compensated cavity three-dimensional surface profile. The compensated cavity profile is resubmitted into the finite element model for verification. If the maximum residual drift is greater than the allowable error, the compensation step in the previous step is repeated for iterative compensation until the residual drift converges to the allowable range. Obtain the physicochemical parameters of the UV-curable resin used in all stages—liquid, gel, and solid—as well as the corresponding parameters of the optical fiber: For resins: Measure the volume shrinkage Sv (or linear shrinkage SL) and the coefficients of thermal expansion CTEr1 and CTEr2 above and below the glass transition temperature Tg after complete curing using a dilatometer or thermomechanical analyzer; obtain the Young's modulus Er and Poisson's ratio νr after curing through dynamic mechanical analysis. For optical fibers and stress bars: obtain the thermal expansion coefficient CTEf, Young's modulus Ef, and Poisson's ratio νf of quartz glass; The thermal mismatch between the resin and the optical fiber, ΔCTE = CTE_r - CTEf, is calculated, where CTEr is the equivalent thermal expansion coefficient within the operating temperature range. This provides the basic data for subsequent simulations. The operating temperature range specifically refers to the working temperature range of the high-density fiber array under a predetermined service scenario, which is typically selected as -40°C to +85°C. The equivalent thermal expansion coefficient of the resin within this temperature range is obtained by measuring the thermal strain curve of the resin from the lower limit (-40°C) to the upper limit (+85°C) of the operating temperature range using a thermomechanical analyzer, and calculating the average linear thermal expansion coefficient within this range, which is then used as the CTE_r value for simulation calculations. Simultaneously, a three-dimensional finite element parametric model containing all components is established. Based on the nominal cavity three-dimensional contour generated in step S1, a parametric model of the "mold-resin-fiber-stress rod" assembly is established in the finite element software: Mold part: an array segment containing at least 10 continuous channels is cut to reflect the cumulative effect. The mold material is set as an elastic body, and its outer surface is constrained as a fixed boundary; Resin part: according to the cavity configuration in S1, a resin domain is generated to fill the main receiving groove, positioning groove and current-limiting neck, and the resin domain is set to be in bonded contact with the mold cavity surface; Fiber and stress rod part: cylindrical fiber and stress rod are accurately placed according to nominal coordinates. The fiber, stress rod and resin domain are set to share nodes or bonded constraints to simulate the fully embedded state; Mesh generation: a fine mesh is used in stress concentration areas such as the current-limiting neck and the junction of fiber and resin. The number of circumferential meshes of the fiber cross section is not less than 36 to ensure the accuracy of displacement calculation; Then, a multi-stage thermo-mechanical coupling process simulation load step is set: according to the actual UV imprinting process, the following simulation load steps are defined in sequence: Step 1: Isothermal curing shrinkage simulation, set the initial temperature Tcure (UV curing temperature, e.g., 60°C); through the thermo-strain analogy method, a virtual temperature shrinkage coefficient αv is assigned to the resin material, so that the thermal strain generated when the temperature drops by ΔTv=1°C is equivalent to the linear curing shrinkage strain SL, i.e., αv=SL / ΔTv; in this step, a temperature drop of ΔTv is applied to the resin domain to simulate isothermal chemical shrinkage; Step 2: Cooling to room temperature simulation; based on the results of Step 1, the entire model is uniformly cooled from Tcure to room temperature Troom (e.g., 25°C), the true thermal expansion coefficients of the resin and optical fiber are activated, and the thermal stress and deformation caused by thermal mismatch are calculated; Step 3: Demolding and release simulation; remove the mold constraint (or make the resin-mold contact into a frictionless contact and allow separation), let the molded body spring back freely, and obtain the three-dimensional coordinates of the center of each optical fiber in the final free state; Then, the center-to-center spacing drift of adjacent fibers is extracted. In post-processing, multiple equidistant cross-sections (no less than 5 cross-sections) are taken along the axis of each fiber, and the coordinates of the outer nodes of the fiber on each cross-section are extracted. The center coordinates of the fiber on each cross-section are fitted to obtain the coordinates of the center point of the fiber. The actual center-to-center distance P'ij of all adjacent fiber pairs in the row or column direction is calculated. The drift ΔPij = P'ij - P is calculated, where P is the nominal fiber core spacing. The drift vector direction and magnitude at each position are recorded. Then, the compensation amount is calculated inversely and the compensated cavity 3D surface profile is generated. For each unit cavity position, the inverse vector of the drift amount is superimposed on the nominal coordinates as the geometric compensation amount: if the drift amount of a fiber center in the X direction is Δx, then the coordinates of the unit cavity center corresponding to the fiber are modified to (Xn-Δx); and so on, to obtain the cavity center coordinate matrix after compensation for all channels. Keeping the local features such as slot width, slot depth, and current-limiting neck designed in S1 unchanged, the entire unit cavity is translated and repositioned according to the compensated center coordinates, and the isolation bosses and continuous surfaces between cavities are regenerated to obtain the compensated cavity 3D profile model. Finally, the iterative convergence judgment and final model determination are performed. The compensated cavity model is substituted back into the above simulation process to calculate the residual drift after compensation. If the maximum residual drift ΔPmax ≤ allowable error ε, the compensation model is determined to be converged, and this model is used as the final mold cavity design output. If ΔPmax>ε, then the residual drift is used as the new drift, and the above compensation steps are repeated for two or more iterations until the accuracy requirements are met. Through the above steps, nominal geometric design, material behavior characterization, multiphysics process simulation and geometric reverse compensation are formed into a closed loop, so that the center spacing between adjacent fibers of the high-density fiber array can still converge to the target value after going through the complete process, thus ensuring the repeatability and manufacturing accuracy of the technical solution of this invention.

[0017] In this embodiment of the invention, step S3 includes the following specific steps: Negative photoresist and positive photoresist are spin-coated sequentially on a silicon substrate to form a stacked photoresist structure; The stacked photoresist was subjected to a single ultraviolet exposure using a grayscale photomask with varying transmittance by region. The main accommodating groove region had the highest transmittance, the positioning groove region had medium transmittance, the capillary guide microgroove region located at the bottom of the main accommodating groove had its transmittance modulated separately, and the isolation boss region had zero transmittance. The exposed multilayer photoresist is post-baked and developed stepwise. First, the exposed positive photoresist is removed, and then the uncrosslinked negative photoresist is removed, forming a stepped photoresist profile with multiple levels of depth difference composed of the remaining photoresist layers. Using the stepped adhesive layer outline as a mask, anisotropic dry etching is performed. By utilizing the difference in etching blocking time of adhesive layers of different thicknesses, a stepped silicon-based structure consisting of a main accommodating groove, a positioning groove, and a bottom capillary micro-groove is simultaneously formed. Remove residual photoresist and smooth the tank wall using short-time isotropic wet etching; A multi-depth stepped prototype was obtained, in which the depth of the main accommodating groove is greater than the depth of the positioning groove, and the bottom of the main accommodating groove has capillary guiding microgrooves extending along the optical fiber axis. As described in the published patent CN102608702B, “A method for fabricating a V-groove of a silicon-based fiber array with low surface tension,” etching on a silicon substrate using photoresist is a common technique in the industry. First, a stacked photoresist structure is prepared: a double-sided polished silicon substrate is cleaned and dehydrated by RCA, and then a silicon dioxide film with a thickness of 50-200 nm is grown on the surface by thermal oxidation, serving as a hard mask or stress buffer layer for subsequent etching; a first layer of photoresist is spin-coated onto the silicon dioxide layer. This photoresist is a negative chemical amplification photoresist, and its thickness is determined by the depth of the target main containment trench. After spin-coating, pre-baking is performed to allow the solvent to evaporate and for initial cross-linking; a second layer of photoresist is spin-coated onto the first negative photoresist. This photoresist is a positive photoresist, and its thickness corresponds to the transition height between the positioning trench depth and the capillary micro-groove; after spin-coating, pre-baking is performed again; thus, a four-layer stacked structure of "positive photoresist / negative photoresist / silicon dioxide / silicon" is formed. Design a grayscale photomask and implement regional dose-controlled exposure. Based on the cavity 3D surface contour after compensation in step S2, generate a grayscale photomask. The transmittance of each pixel on the mask is encoded as a grayscale level of 0~100%, corresponding to continuous or quasi-continuous changes in exposure dose. The specific allocation is as follows: Main accommodating groove area: transmittance is set to 100% to allow complete UV light transmission and provide the highest dose exposure for the multilayer adhesive; Positioning groove area: transmittance is set to 30%~60% to provide medium dose exposure; Capillary guide microgroove area... Located at the center of the bottom of the main accommodating groove, extending along the optical fiber axis, the transmittance is finely modulated between 20% and 40% to produce a latent image with a gradual depth of positive adhesive; the isolation boss and non-groove area have a transmittance of 0%, completely blocking light; the silicon substrate is aligned and placed under the mask, and a single exposure is performed using ultraviolet light; the exposure dose is jointly calibrated by the minimum dose Dn required for complete cross-linking of negative adhesive and the dose Dp required for complete photosensitive exposure of positive adhesive, preferably the exposure dose is in the intersection range of Dn and Dp, so as to simultaneously satisfy the cross-linking of negative adhesive and the photosensitive decomposition of positive adhesive; The development process creates a multi-tiered, stepped adhesive layer profile. After exposure, post-baking promotes the acid-catalyzed cross-linking reaction of the negative adhesive and the chemical amplification reaction of the positive adhesive. Subsequent stepwise development involves: first, selectively removing the exposed areas of the positive adhesive with a positive adhesive developer, exposing the surface of the underlying negative adhesive layer. At this stage, due to varying exposure doses received in different areas, the depth of positive adhesive removal also differs: in the high-dose main reservoir region, the positive adhesive is completely removed; in the medium-dose positioning reservoir region, the positive adhesive is partially removed, leaving a thinner layer; and in the low-dose microgroove region, the remaining positive adhesive is even thinner. Then, a negative adhesive developer is used... The gel developer dissolves and removes the negative gel that has not been cross-linked (i.e., not sufficiently exposed to ultraviolet light). The negative gel in the main containment area is highly cross-linked due to sufficient exposure and is retained. The negative gel in the positioning area receives partial exposure, has a slightly lower degree of cross-linking but is still insoluble in the developer and is retained. The negative gel in the zero-exposure boss area is completely dissolved, exposing the underlying silica layer. Finally, a stepped gel layer contour is formed on the silicon substrate, consisting of the retained cross-linked negative gel and the residual positive gel. This contour precisely corresponds to the depth of the main containment area, the shallowness of the positioning area, and the primary depression of the micro-grooves. The first dry etching process is performed to form the main containment trench body and the microtrench preform. Using the stepped adhesive layer outline as a mask, anisotropic reactive ion etching (RIE) is performed using a continuous etching process. During the etching process, the blocking time of ion bombardment varies depending on the thickness of the adhesive layer: the adhesive layer in the main containment trench area is the thinnest (only negative adhesive), and it is etched through first, exposing the silicon substrate and starting to etch the silicon, forming most of the depth of the main containment trench; the positioning trench area has residual positive and negative adhesive, so the etching time is slightly later, and the silicon etching starts later, resulting in a shallower trench depth; the adhesive layer in the microtrench area is even thinner, and even after development, the silicon surface is slightly exposed, so silicon etching starts earliest and the etching depth is the deepest, naturally forming the initial shape of capillary guiding microtrench along the fiber axis. By precisely controlling the etching time, the overall depth of the main accommodating groove can reach the design value Hf, the depth of the positioning groove can reach Hs, and the micro trench can be further etched to 5~15μm below the bottom plane of the main accommodating groove, forming a micro trench with a rectangular or trapezoidal cross section. Wet photoresist removal and short-time isotropic finishing are employed to remove the etched silicon wafer. The wafer is then thoroughly cleaned using oxygen plasma ashing and wet cleaning to remove residual photoresist and any potential polymers. Subsequently, it is immersed in a diluted hydrofluoric acid / nitric acid mixture for short-time isotropic wet etching. This process smooths sidewall ripples generated by dry etching, eliminates microburrs, and rounds the corners of the capillary microchannels, facilitating resin flow. Simultaneously, wet etching slightly widens the channels; this widening is pre-considered in the S2 pre-distortion compensation. Finally, a multi-depth stepped master mold is obtained. After the above steps, a multi-depth stepped structure that fully meets the design requirements is formed on the silicon substrate: the main accommodating groove is the deepest, and there is a capillary guiding micro-groove that extends continuously along the optical fiber axis in the center of the bottom; shallower positioning grooves are symmetrically distributed on both sides; and there are complete isolation bosses between adjacent units; this master mold can be used as a three-dimensional master for the electroformed metal mold core in S4.

[0018] In this embodiment of the invention, step S4 includes the following specific steps: The silicon-based prototype obtained in step S3 is subjected to surface cleaning and hydrogen passivation treatment, and then a metal conductive seed layer with a thickness of 50~200 nm is deposited on the surface of the cavity structure by physical vapor deposition. A sacrificial layer with a thickness of 10~30 nm is formed in situ on the conductive seed layer for subsequent precise separation of the master mold and the metal mold core; Using a silicon master mold with a seed layer and a sacrificial layer as the cathode, pulse electroforming is performed in a nickel sulfamate electroforming solution to form a metal core body with a thickness of 1~5 mm. After planar grinding of the back side of the metal mold core, the sacrificial layer is removed by selective wet etching, so that the silicon original mold and the metal mold core are separated without damage, and a metal mold core with a cavity structure that is complementary to the original mold is obtained. The metal mold core is placed in a plasma chamber for oxygen plasma activation treatment, generating high-density hydroxyl active sites on the surface of the mold cavity; The activated metal core is immersed in an anhydrous organic solvent containing organofluorine silane precursors. Through the condensation reaction between the silane head group and the surface hydroxyl group, an organofluorine hydrophobic monolayer with a thickness of 1~3 nm is self-assembled on the surface of the cavity to form a permanent release layer. The silicon-based multi-depth stepped prototype obtained in step S3 is subjected to a conductive pretreatment. First, it is cleaned by oxygen plasma ashing to remove organic residues, and then it is rinsed in diluted hydrofluoric acid for several seconds to remove the natural oxide layer on the silicon surface, so that the silicon surface is in a hydrophobic hydrogen passivated state. Subsequently, a metal conductive seed layer is deposited on the surface of the cavity structure by physical vapor deposition. The preferred deposition material is nickel or silver, and the deposition thickness is 50~200 nm. This seed layer needs to completely cover the entire three-dimensional surface of the main receiving groove, positioning groove, capillary guiding micro-groove and isolation boss, so as to provide a uniform conductive path for subsequent electroforming. At the same time, its grains are small and can be used as an epitaxial growth template for subsequent electroforming layers. A sacrificial release layer is prepared by forming an oxide sacrificial layer with a thickness of 10-30 nm in situ on a conductive seed layer through electrochemical or chemical oxidation methods. If the seed layer is nickel, it can be anodized in a dilute sulfuric acid solution to form a nickel oxide film by controlling the potential. After subsequent electroforming, this sacrificial layer can be removed by selective wet etching without damaging the metal mold core and the silicon master mold, thus achieving physical separation between the silicon master mold and the metal mold core. The thickness of the sacrificial layer is much smaller than the cavity feature size, so it does not affect the replication accuracy. The electroforming deposition of the metal mold core involves immersing a silicon primary mold with a conductive seed layer and a sacrificial layer as the cathode and a high-purity metal (preferably nickel or a nickel-cobalt alloy) as the anode in an electroforming solution. The electroforming solution is composed of a nickel sulfamate system, supplemented with stress-control additives to reduce deposition internal stress and prevent mold core warping. Pulsed or periodically commutated currents are used for electroforming to improve the uniformity of the plating thickness in deep grooves and micro-grooves, avoiding the "dog bone" effect. Electroforming continues until a continuous metal body with a thickness of 1-5 mm is formed above the cavity surface. During the electroforming process, the plating solution temperature is kept constant (e.g., 50-55°C) and the pH value is kept stable to ensure a dense, pinhole-free deposition layer. The backing is separated from the original mold. After electroforming, the back side (non-cavity surface) of the metal body is planar ground and polished to obtain a flat mounting reference surface. Then, the sacrificial layer of S4 is removed by selective etching: if it is a nickel oxide sacrificial layer, it is soaked in dilute hydrochloric acid or a special nickel oxide removal solution. After the sacrificial layer dissolves, the silicon original mold and the metal core separate naturally. After separation, the cavity surface of the metal core is cleaned with deionized water and dried with nitrogen to obtain a metal core that is completely complementary to the three-dimensional structure of the silicon original mold. Cavity surface plasma activation treatment: The separated metal core is placed in an oxygen plasma or air plasma chamber and treated for 5 to 15 minutes under low-voltage radio frequency discharge conditions. This treatment generates a high-density layer of hydroxyl (-OH) active sites on the metal surface without changing the cavity size, providing anchor points for the subsequent chemical bonding of organofluorosilane molecules. The treated core needs to be temporarily stored in an inert atmosphere or a clean and dry environment to avoid contamination or hydration of the active sites. Finally, an organofluorine hydrophobic self-assembled monolayer is deposited. The plasma-activated metal core is immersed in an anhydrous organic solvent containing organofluorine silane precursors (such as perfluorooctyltrichlorosilane, perfluorodecyltrimethoxysilane, etc.). The solvent is preferably anhydrous n-hexane or anhydrous isooctane, and the precursor concentration is 0.1~5 mmol / L. The core is then immersed in a sealed container filled with dry inert gas for 30 minutes to 2 hours, allowing the organofluorine silane molecules to undergo hydrolytic condensation reactions with the hydroxyl groups on the metal surface through the silane head groups, forming a Me-O-Si covalently bonded, oriented monolayer on the cavity surface. The fluorinated alkyl segments extend outward to form a hydrophobic and oleophobic surface with extremely low surface free energy. After removal, rinse with anhydrous solvent to remove unbonded physical adsorption residues, and then anneal and cure in an oven at 100~120°C for 30~60 minutes to further promote cross-linking and stabilize the monolayer structure. Thus, a permanent release layer with a thickness of only 1~3 nm is obtained on the surface of all microstructures in the cavity, without changing the cavity design size, resulting in a complex cavity mold that can be directly used for UV embossing in S5.

[0019] In this embodiment of the invention, step S5 includes the following specific steps: The mold obtained in step S4 is activated by instantaneous plasma, and low-viscosity UV-curable resin is filled into the cavity area by quantitative dispensing. The resin-filled mold is placed in a vacuum chamber and a vacuum is drawn to ensure that the resin completely wets all microgrooves and flow-limiting necks under bubble-free conditions. The bare optical fiber and stress bar are arranged in parallel according to the design and pre-clamped. The whole thing is lowered so that the optical fiber and stress bar are simultaneously aligned with the main receiving slot and the positioning slot. During the embedding process, the V-shaped cross section of the positioning groove is used to generate geometric constraints on the cylindrical surface of the stress rod, causing the stress rod to spontaneously rotate to the lowest potential energy state. Cover the transparent quartz cover plate with anti-stick treatment and apply controllable pressure to press the optical fiber and stress bar completely into the bottom of the tank, while squeezing out excess resin to control the cladding thickness. With the cover plate pressed in place, ultraviolet light with a wavelength of 365 nm is uniformly irradiated from above the cover plate to cure the resin. Remove the cover plate, use the permanent hydrophobic release layer of the mold to release the formed high-density fiber array without damage, and perform post-curing heat treatment to eliminate internal stress. The metal mold with an organic fluorine hydrophobic self-assembled monolayer on the cavity surface obtained in step S4 is placed in a clean environment, and oxygen plasma is used to perform instantaneous (3-5 seconds) surface activation on the mold. While maintaining the hydrophobic body, the initial wettability of the resin is slightly improved for filling. Then, a micro-dispensing system is used to quantitatively add UV-curable resin at the starting end of the mold cavity area; the resin formulation is preferably a low viscosity (<300 mPa·s) and low curing shrinkage (<3%) acrylate or epoxy-acrylate hybrid system to ensure sufficient penetration into the capillary microgrooves; the amount of resin dispensed is calculated as 1.05 to 1.1 times the total volume of the cavity to ensure complete filling with a slight excess; Vacuum-assisted defoaming and resin pre-impregnation: The mold with resin dripped on it is moved into a vacuum chamber, and a vacuum is drawn to 10~100Pa and maintained for 1~5 minutes. Under vacuum conditions, the residual gas in the microgrooves and flow-limiting necks of the cavity will expand, overflow and be pumped away. The liquid resin will penetrate into all the microstructures by itself under pressure difference and capillary action, especially completely wetting the bottom capillary guide microgrooves. Then, the pressure is slowly released to atmospheric pressure, further pressing the resin into the remaining nanoscale gaps to achieve bubble-free filling. The parallel array of optical fibers and stress bars is pre-assembled to prepare the bare optical fibers and stress bars determined by steps S1 and S2; the optical fibers have been stripped of the coating and cleaned at the preset position; all bare optical fibers and corresponding stress bars are arranged in parallel according to the design order and installed in a pre-assembled fixture with adjustable spacing so that their axes are consistent with the groove direction of the mold cavity, and the relative initial position of each optical fiber and the two stress bars on both sides roughly corresponds to the layout of the main receiving groove and the positioning groove in the cavity. Self-aligned embedding is achieved using the geometric constraints of the positioning slot. Under manual or automatic optical monitoring, the pre-assembled fixture is lowered as a whole, allowing the optical fiber and stress bar to simultaneously enter the corresponding cavity. The specific process is as follows: The bare optical fiber first contacts the inclined walls on both sides of the main receiving groove, and automatically slides into the bottom of the groove under the assistance of gravity and slight external vibration, and is precisely positioned by the main receiving groove; when the stress rod falls into the V-shaped positioning grooves on both sides, its cylindrical surface forms two lines of contact with the two inclined surfaces of the V-shaped grooves; due to the precise design of the groove opening width and included angle, the stress rod will spontaneously rotate after contacting the groove wall until its own axis is consistent with the length direction of the groove, and its center of mass drops to the lowest point; during the entire falling process, due to the capillary effect of the flow-limiting neck in S1, the resin flow is delayed and will not push the stress rod away from the positioning groove; at the same time, the resin displaced by the optical fiber and stress rod slowly overflows to the top surface of the isolation boss, forming a thin layer; A pressure cap is applied and the resin thickness is adjusted. A transparent quartz glass cover plate treated with fluorosilane for non-sticking is placed over the mold. The cover plate is pressed tightly by precision pads around the perimeter or by applying controllable pressure (0.1~0.3 MPa) to achieve the following functions: completely press the optical fiber and stress bar into the bottom of their respective grooves, eliminating any floating or tilting state and locking their position; squeeze excess resin overflowing from the cavity into the overflow grooves around the perimeter, precisely controlling the resin coating thickness above the optical fiber and stress bar; the cover plate pressure further promotes the complete filling of resin in the microgrooves and flow-limiting necks, and removes any remaining microbubbles. Uniform UV irradiation curing: With the cover plate pressed down, a 365 nm UV LED surface light source is used, incident directly from above the cover plate, with a light intensity of 100~200 mW / cm² for 30~120 seconds. The UV light passes through the transparent cover plate, initiating resin polymerization and curing. During the curing process, the uniformity of light irradiation must be controlled (within ±5%) to avoid uneven local curing shrinkage that could cause fiber misalignment. It is advisable to pre-cur at low intensity (10~20 mW / cm²) for 5~10 seconds for "shaping" before full-intensity curing to reduce shrinkage stress. After curing, the cover plate is removed first. Due to the ultra-low surface energy of the permanent hydrophobic release layer on the mold cavity surface, the adhesion between the cured resin-fiber-stress rod composite and the mold is extremely low. Using a vacuum suction head or ejection device, the formed high-density fiber array is completely and undamagedly removed from the mold. After demolding, the array is visually inspected or automatically optically tested to verify the fiber alignment. If necessary, short-time thermal annealing (80~120°C, 30~60 minutes) can be performed to further release internal stress and stabilize the dimensions. Thus, a high-density fiber array with both precise fiber positioning and stress element embedding is obtained.

[0020] In this embodiment of the invention, the following steps are set between steps S2 and S3: Between the compensated three-dimensional curved surface contour of the cavity generated in S2 and the grayscale lithography mask designed in S3, a lithography process proximity effect compensation model is established. The compensated cavity contour data is input into the model, and the transmittance value of each pixel of the grayscale mask is calculated and corrected to compensate for the pattern distortion caused by optical proximity effect and development diffusion during the lithography process. This step incorporates the nonlinear distortion of the photolithography process into the compensation chain. Through inversion calculation using the proximity effect compensation model, the transmittance distribution of the grayscale mask is no longer directly equal to the depth mapping of the target contour, but rather a value processed by "pre-distortion". In this way, the developed contour after photolithography can accurately reproduce the three-dimensional surface after S2 compensation. This ensures consistent precision across the entire chain from design to manufacturing, preventing the iterative compensation work of S2 from being offset by unknown errors in the photolithography process. The cavity in this scheme includes fine features such as bottom capillary microgrooves, flow-limiting necks, and multi-level steps, with dimensions ranging from several micrometers to hundreds of micrometers. These structures have large linewidth / spatial period spans. It is highly susceptible to the proximity effect; the compensation model can apply differentiated dose correction to patterns of different spatial frequencies, so that narrow microgrooves are not widened by overexposure and wide positioning grooves are not narrowed by underexposure, ensuring that the cross-sectional shape of all functional structures is consistent with the design; at the same time, the lithographic proximity effect is also affected by changes in process parameters such as substrate reflectivity and photoresist thickness fluctuations; after modeling, the substrate conditions of different batches can be quickly and adaptively corrected, without the need to re-produce the mask each time, which significantly shortens the process adjustment cycle and ensures a high degree of consistency of the original mold size in mass production, providing a stable and reliable master mold quality for S4 electroforming metal mold cores.

[0021] In this embodiment of the invention, a closed-loop overflow groove 005 connected end to end is provided around the mold cavity area. The depth of the overflow groove is 1.2 to 1.5 times the depth of the main receiving groove. The overflow groove and the unit cavity are separated by an overflow weir with a height of 5 to 20 μm, so that when the capping is performed in S5, the excess resin preferentially crosses the overflow weir 006 and enters the overflow groove instead of being connected between adjacent channels. like Figure 8 As shown, the closed-loop overflow channel surrounds the entire cavity area and is separated from the unit cavity by the overflow weir. The overflow weir is higher than the top surface of the isolation boss but lower than the pressing surface of the cover plate, forming a controllable "overflow threshold". During the pressing process, excess resin must cross the top of the weir to enter the overflow channel, while the resin on the top surface of the isolation boss between adjacent channels will not be laterally connected because there is no such flow driving force. This physically cuts off the possibility of resin bridging between channels from the flow channel topology. The absence of burr residue between channels means that each optical fiber is independently held in its own resin cladding, and there is no rigid mechanical coupling between adjacent optical fibers, maintaining the precise center distance achieved from S1 to S2. The closed-loop overflow groove also acts as an "exhaust ring" during the pressing process. When the resin flows from the center to the periphery, the residual gas in the cavity can be driven to the overflow groove and discharged. Combined with the vacuum pre-impregnation of S5, the probability of microbubble residue is further reduced, especially improving the filling integrity rate of the edge channels of the array.

[0022] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A method for packaging fiber optic arrays, used in the manufacture of fiber optic arrays, characterized in that, Includes the following steps: S1: Based on the target number of fiber channels N and the fiber core spacing P, determine the arrangement array of the mold cavity, wherein each unit cavity includes a main fiber receiving slot and positioning slots symmetrically arranged on both sides of the main receiving slot. The main receiving slot and the positioning slot are connected by a current limiting neck with a gradually changing width. An isolation boss is reserved between the positioning slots of adjacent unit cavities. S2: Perform thermo-mechanical coupling pre-distortion compensation on the nominal dimensions of the unit cavity, including obtaining the curing shrinkage rate and thermal expansion coefficient of the photocurable molding resin used, as well as the thermal mismatch between the resin and the optical fiber material, calculating the drift of the center spacing between adjacent optical fibers after molding through finite element simulation, and superimposing the compensation obtained by inverse calculation onto the main accommodating slot spacing and the positioning slot spacing to generate the compensated three-dimensional curved surface profile of the cavity. S3: Coating multiple layers of positive and negative photoresist on a silicon substrate, and forming a multi-depth stepped prototype with the three-dimensional curved surface contour of the cavity by grayscale photolithography exposure and alternating dry and wet etching, wherein the depth of the main accommodating groove is greater than the depth of the positioning groove, and the bottom of the main accommodating groove is provided with a capillary guiding micro-groove extending along the optical fiber axis. S4: The original mold is made into a metal mold core by electroforming, and an organic fluorine hydrophobic self-assembled monolayer is deposited on the cavity surface of the metal mold core as a release layer to obtain a complex cavity mold for ultraviolet embossing. S5: Fill the complex cavity mold with photocurable resin, place the bare optical fiber with the coating removed and the optical fiber array in parallel into the corresponding main receiving groove and positioning groove, and then cure by ultraviolet light to obtain a high-density optical fiber array that combines optical fiber positioning and stress element embedding in one step. Step S1 specifically includes the following steps: Based on the target number of fiber channels N, the cavity arrangement of the mold is determined to be a one-dimensional linear array, and the center distance between adjacent unit cavities is strictly equal to the fiber core spacing P. Obtain the bare fiber diameter Df, stress bar diameter Ds, and center distance Ls of the two stress bars in the target optical fiber. In two symmetrically distributed V-shaped positioning slots with their openings facing the main receiving slot, make the symmetry plane of the two V-shaped positioning slots coincide with the central axis of the main receiving slot, and make the center distance between the two V-shaped positioning slots equal to Ls.

2. The fiber optic array packaging method according to claim 1, characterized in that, Step S1 specifically includes the following steps: The depth of the main receiving groove is Hf < Df, the depth of the positioning groove is Hs < Ds, and after the optical fiber and stress rod are inserted, a portion of their height protrudes from their respective groove openings.

3. The fiber optic array packaging method according to claim 2, characterized in that, Step S1 specifically includes the following steps: setting the width of the flow-limiting neck connecting the main accommodating slot and the positioning slot to gradually decrease from the main accommodating slot to the positioning slot.

4. The fiber optic array packaging method according to claim 1, characterized in that, Step S2 specifically includes the following steps: Obtain the curing shrinkage rate, coefficient of thermal expansion, Young's modulus and Poisson's ratio of the photocurable resin used, as well as the coefficient of thermal expansion and elastic modulus of the optical fiber and stress rod materials, and calculate the thermal mismatch between the resin and the optical fiber. Based on the nominal cavity profile generated in step S1, a three-dimensional finite element parametric model including mold, resin, optical fiber and stress bar is established. In the model, the optical fiber, stress bar and resin domain are bound together, and the outer surface of the mold is completely fixed. Three load steps are applied sequentially to the model: isothermal curing shrinkage, cooling to room temperature, and demolding release. The curing shrinkage is equivalent to the resin by giving it a virtual thermal shrinkage coefficient, while the cooling process uses the real thermal expansion coefficient. Calculate the center coordinates of each fiber after demolding, extract the actual center distance in the row or column direction of all adjacent fiber pairs, and calculate the drift amount ΔP between them and the nominal fiber core spacing P. Invert the drift amount corresponding to each fiber position, translate the corresponding unit cavity center coordinates, keep the feature size in the slot unchanged, and generate the compensated cavity three-dimensional surface profile. The compensated cavity profile is resubmitted into the finite element model for verification. If the maximum residual drift is greater than the allowable error, the compensation step in the previous step is repeated for iterative compensation until the residual drift converges to the allowable range.

5. The fiber optic array packaging method according to claim 1, characterized in that, Step S3 includes the following specific steps: Negative photoresist and positive photoresist are spin-coated sequentially on a silicon substrate to form a stacked photoresist structure; The stacked photoresist was subjected to a single ultraviolet exposure using a grayscale photomask with varying transmittance by region. The main accommodating groove region had the highest transmittance, the positioning groove region had medium transmittance, the capillary guide microgroove region located at the bottom of the main accommodating groove had its transmittance modulated separately, and the isolation boss region had zero transmittance. The exposed multilayer photoresist is post-baked and developed stepwise. First, the exposed positive photoresist is removed, and then the uncrosslinked negative photoresist is removed, forming a stepped photoresist profile with multiple levels of depth difference composed of the remaining photoresist layers. Using the stepped adhesive layer outline as a mask, anisotropic dry etching is performed. By utilizing the difference in etching blocking time of adhesive layers of different thicknesses, a stepped silicon-based structure consisting of a main accommodating groove, a positioning groove, and a bottom capillary micro-groove is simultaneously formed. Remove residual photoresist and smooth the tank wall using short-time isotropic wet etching; A multi-depth stepped silicon-based prototype was obtained, in which the depth of the main accommodating groove is greater than the depth of the positioning groove, and the bottom of the main accommodating groove has capillary guiding microgrooves extending along the optical fiber axis.

6. The fiber optic array packaging method according to claim 5, characterized in that, Step S4 includes the following specific steps: The silicon-based prototype obtained in step S3 is subjected to surface cleaning and hydrogen passivation treatment, and then a metal conductive seed layer is deposited on the surface of the cavity structure by physical vapor deposition. A sacrificial layer is formed in situ on the conductive seed layer for subsequent precise separation of the master mold and the metal mold core; Using a silicon primary mold with a seed layer and a sacrificial layer as the cathode, pulse electroforming is performed in a nickel sulfamate electroforming solution to form a metal mold core body. After planar grinding of the back side of the metal mold core, the sacrificial layer is removed by selective wet etching, so that the silicon original mold and the metal mold core are separated without damage, and a metal mold core with a cavity structure that is complementary to the original mold is obtained. The metal mold core is placed in a plasma chamber for oxygen plasma activation treatment, generating high-density hydroxyl active sites on the surface of the mold cavity; The activated metal core is immersed in an anhydrous organic solvent containing an organofluorine silane precursor. Through the condensation reaction between the silane head group and the surface hydroxyl group, an organofluorine hydrophobic monolayer is formed on the surface of the cavity, which serves as a permanent release layer.

7. The fiber optic array packaging method according to claim 6, characterized in that, Step S5 includes the following specific steps: The mold obtained in step S4 is activated by instantaneous plasma, and low-viscosity UV-curable resin is filled into the cavity area by quantitative dispensing. The resin-filled mold is placed in a vacuum chamber and a vacuum is drawn to ensure that the resin completely wets all microgrooves and flow-limiting necks under bubble-free conditions. The bare optical fiber and stress bar are arranged in parallel according to the design and pre-clamped. The whole thing is lowered so that the optical fiber and stress bar are simultaneously aligned with the main receiving slot and the positioning slot. During the embedding process, the V-shaped cross section of the positioning groove is used to generate geometric constraints on the cylindrical surface of the stress rod, causing the stress rod to spontaneously rotate to the lowest potential energy state. Cover the transparent quartz cover plate with anti-stick treatment and apply controllable pressure to press the optical fiber and stress bar completely into the bottom of the tank, while squeezing out excess resin to control the cladding thickness. With the cover plate pressed in place, ultraviolet light is evenly irradiated from above the cover plate to cure the resin. Remove the cover plate.

8. The fiber optic array packaging method according to claim 5, characterized in that, The steps between S2 and S3 include the following: Between the compensated three-dimensional curved surface contour of the cavity generated in S2 and the grayscale lithography mask designed in S3, a lithography process proximity effect compensation model is established. The compensated cavity contour data is input into the model, and the transmittance value of each pixel of the grayscale mask is calculated and corrected to compensate for the pattern distortion caused by optical proximity effect and development diffusion during the lithography process.

9. A packaging device for fiber optic arrays, characterized in that, The packaging device is operated using the method described in any one of claims 1-8.

Images