High-precision alignment fitting and on-line metering reworking system and process for flexible optical element

Abstract

The invention discloses a high-precision alignment fitting and online metering reworking system and process of a flexible optical element, which are applied to the field of optical elements, and the scheme establishes a computable error model in the fitting process of a flexible substrate and an optical microstructure, completes estimation in a production takt, fuses visual, electrical and optical metering results, and improves the precision of the flexible substrate and the optical microstructure. According to the method, a unified'pass / rework / degradation / scrap 'decision is output, alignment error statistics and defect reason codes are recharged into manufacturing parameters and design parameters, alignment errors are converted into an estimable, compensable and traceable quantitative closed loop from'empirical debugging', the rejection rate is remarkably reduced through online metering and rework window control, the yield and consistency are improved, and the product quality is improved. And a systematic barrier of a process control loop and a data packet tracing interface is formed.

Description

Technical Field

[0001] This invention relates to the field of optical components, and in particular to a high-precision alignment and bonding system and process for online measurement and rework of flexible optical components. Background Technology

[0002] Adjustable focus / electroactive lenses, as a novel type of vision correction device, can achieve dynamic adjustment of refractive power through the coordinated operation of a flexible electronic backplane and an active optical region (such as Fresnel phase structures and GRIN dielectric layers). However, in the process of moving from the laboratory to mass production, these devices face a key technological bottleneck: the high-precision alignment and bonding of the flexible electronic layer with the optical microstructure.

[0003] Adjustable focusing lenses require high-precision alignment between the flexible electronics / electrode layer and the optical active area; alignment errors can lead to refractive power misalignment and aberrations that are inconsistent with those of both eyes. This problem is more prominent in roll-to-roll mass production and is a key risk point in the transition from laboratory to mass production.

[0004] In the mass production of adjustable-focus / electroactive lenses, geometric deviations between the flexible electronic layer and the optical active region are amplified into diopter eccentricity, prismatic effects, and aberration increases, potentially manifesting as binocular differences, decreased image quality, wearing discomfort, or unstable diopter response. Existing bonding processes often face the following challenges: Visual alignment alone is insufficient to cover electrical defects (broken wires / leakage) and optical defects (wavefront / haze). Lack of quantifiable rework window: Irreversible curing of the adhesive layer leads to a high scrap rate; Lack of design / manufacturing closed loop: Alignment deviation statistics cannot be fed back into the layout / structural design and process window updates. Summary of the Invention

[0005] The core of this invention lies in transforming the alignment error from "experience-based machine adjustment" into a quantifiable closed loop that is estimable, compensable, and traceable. Through online metering and rework window control, it significantly reduces the scrap rate, improves yield and consistency, and forms a systematic barrier of "process control loop + data packet traceability interface". This solves the problems in existing technologies, such as visual alignment alone leading to various defects, lack of quantifiable rework windows, and lack of design / manufacturing closed loop.

[0006] To solve the above problems, the present invention adopts the following technical solution.

[0007] A high-precision alignment and bonding system for flexible optical components, along with online measurement and rework systems, includes: The stage and clamping mechanism are used to clamp the flexible electronic backplane and the optical substrate with optical microstructure respectively, and to apply tension control to the flexible electronic backplane. The alignment reference structure includes visual reference marks and mechanical references set on the flexible electronic backplane and the optical substrate, which are used to establish at least one pair of alignment coordinate systems; The machine vision measurement module is used to acquire the visual reference marks and estimate translation, rotation, scaling and nonlinear deformation errors; The bonding and curing module is used to bond and cure the flexible electronic backplane to the optical substrate through an optically transparent adhesive layer in the alignment state. An online multi-physical quantity measurement module, including at least one of a visual measurement module, an electrical measurement module and an optical measurement module, is used to measure alignment error and device performance in at least two stages: before bonding, during bonding and after bonding. The closed-loop controller communicates with the machine vision measurement module and the online multi-physical quantity measurement module. The closed-loop controller is used to perform alignment error decomposition and adaptively adjust the bonding parameters and update the alignment compensation amount according to the measurement results. The system is configured to make a rework decision based on online measurement results within a preset rework window and output a traceability data packet bound to the device.

[0008] Furthermore, the alignment reference structure includes multi-scale reference markers, which include at least coarse alignment markers and fine alignment markers.

[0009] Furthermore, the closed-loop controller decomposes the alignment error into translation errors Δx and Δy, rotation error Δθ, scaling error Δs, and warping / nonlinear deformation term, and outputs a compensation transformation matrix T*; where Δs is the deviation of the scaling factor s from 1.

[0010] Furthermore, the stage and clamping mechanism include a tension sensor and a tension closed-loop control unit. The tension closed-loop control unit adjusts the traction speed or clamping displacement of the clamping mechanism according to the feedback from the tension sensor, so that the in-plane strain of the flexible electronic backplane is within a preset range, thereby reducing the secondary displacement caused by stress and curing shrinkage after bonding.

[0011] Furthermore, the online multi-physical quantity measurement module includes a wavefront measurement module, which is used to measure the wavefront phase Φ(x,y) after bonding and calculate the equivalent optical path difference OPD(x,y)=λ·Φ(x,y) / (2π) to evaluate the wavefront error caused by alignment and curing.

[0012] Furthermore, the online multi-physical quantity measurement module includes an electrical continuity test module, which is used to detect electrode open circuits, short circuits, and sheet resistance drift according to a preset test sequence.

[0013] Furthermore, the bonding and curing module includes at least one of a UV curing unit and a thermal curing unit, and performs closed-loop control on the corresponding curing dosage curve or temperature curve to suppress secondary offset caused by curing shrinkage.

[0014] Furthermore, the system also includes a rework station for performing disassembly, cleaning, re-adhesion, and re-bonding within the rework window.

[0015] Furthermore, the closed-loop controller updates the process parameter vector p=[p1,…,pm] based on the online metering results and forms a batch normalization strategy.

[0016] Furthermore, the system is adapted to at least one production line configuration, either roll-to-roll or sheet-to-sheet, and maintains alignment accuracy when the production line speed changes.

[0017] The high-precision alignment and bonding and online measurement rework process for flexible optical components, applicable to the aforementioned high-precision alignment and bonding and online measurement rework system, includes the following steps: Step A: Establish the coordinate system Σ_E of the flexible electronic backplane and the coordinate system Σ_O of the optical substrate, and collect visual reference marks; Step B: Estimate the geometric transformation from Σ_E to Σ_O and calculate the alignment error; Step C: Decompose the alignment error and output the compensation transformation T* to adjust the stage attitude, tension or bonding path; Step D: After adjustment, perform bonding and curing, and perform online multi-physical quantity measurement in at least two stages: before bonding, during bonding, and after bonding; Step E: Determine whether rework is required based on the measurement results in the rework window; Step F: Output the traceability data packet and bind it to the device serial number; Step G: Map the alignment error statistics, cause codes, and key measurement indicators to process parameter update quantities Δp, and execute adaptive compensation and batch normalization strategies in subsequent batches.

[0018] Furthermore, in step B, the geometric transformation is estimated using a homography matrix or an affine model, and a piecewise deformation model or a spline deformation model is introduced when warping exists.

[0019] Furthermore, in step C, the compensation transformation is estimated by minimizing the objective function J, where: J=Σ_{i=1}^{N}||q_i-Tp_i||_2^2+η_reg||w||_2^2; Where p_i is the reference point in coordinate system Σ_E, q_i is the corresponding point in coordinate system Σ_O, T is the similarity transformation matrix, w is the deformation field parameter vector, and η_reg is the regularization coefficient.

[0020] Furthermore, in step D, the dosage curve or temperature curve of the curing process is controlled by feedback to ensure that the secondary offset caused by curing shrinkage is less than the threshold.

[0021] Furthermore, in step D, the online multi-physical quantity measurement includes at least electrical continuity testing and wavefront measurement, and jointly determines whether the assembly threshold is met.

[0022] Furthermore, the determination of rework in step E is based on at least one of the following: alignment error threshold, wavefront RMS threshold, and electrical yield threshold.

[0023] Furthermore, within the rework window, the process involves disassembly, cleaning, re-adhesive application, and re-bonding, with an upper limit on the number of rework attempts.

[0024] Furthermore, the alignment error and wavefront error are mapped to process parameter update amounts Δp, and adaptive compensation is performed in subsequent batches.

[0025] Furthermore, the traceability data package includes at least: alignment reference version, algorithm / model version, bonding and solidification parameters, online measurement results, rework records, original data hash digest, reason code, and digital signature field.

[0026] Furthermore, it also includes closed-loop control of the tension of the flexible backsheet to keep the in-plane strain within a preset range.

[0027] Furthermore, it also includes online estimation of the adhesive layer thickness and incorporating the thickness deviation into the error decomposition model.

[0028] Furthermore, after bonding, an optical performance acceptance operation is performed and a parameter rollback point is generated that can be used for after-sales review.

[0029] Furthermore, the traceability data package is used to achieve closed-loop management of manufacturing → calibration → after-sales service.

[0030] Furthermore, the joint determination includes hard threshold gating and fusion score gating; when any key measurement indicator exceeds the corresponding hard threshold, it is determined as a failure and a reason code is generated; when all key measurement indicators meet the hard threshold, the fusion score S is calculated and compared with the score threshold to determine whether it passes or enters the rework process.

[0031] Furthermore, the key metrology index includes the image quality index MTF calculated based on the wavefront metrology results. The MTF is obtained by the Fourier transform of the point spread function PSF, and the minimum MTF value at a preset spatial frequency is used in the calculation of the fusion score S.

[0032] Furthermore, a calibration transformation is performed on the wavefront measurement coordinate system and the pupil coordinate system Σ_P, and the wavefront root mean square error W_RMS, image quality MTF and their spatial weights are calculated in the pupil coordinate system Σ_P.

[0033] Furthermore, a hash digest is calculated for the traceability data packet and written into an immutable log chain, and the hash digest is digitally signed using a private key to support cross-workstation consistency verification and after-sales traceability.

[0034] Compared with the prior art, the advantages of this invention are: (1) This solution establishes a calculable error model during the bonding process between the flexible substrate and the optical microstructure, and completes the estimation within the production cycle. It integrates visual, electrical, and optical measurement results and outputs a unified "pass / rework / downgrade / scrap" decision. The alignment error statistics and defect cause codes are fed back into the manufacturing parameters and design parameters, realizing the transformation of alignment error from "experience-based machine adjustment" into an estimable, compensable, and traceable quantitative closed loop. Through online measurement and rework window control, the scrap rate is significantly reduced, the yield and consistency are improved, and a systematic barrier of "process control loop + data packet traceability interface" is formed. Attached Figure Description

[0035] Figure 1 This is a schematic diagram of the overall architecture of the alignment and bonding and online measurement rework system of the present invention; Figure 2 This is a schematic diagram of the alignment error decomposition model (translation / rotation / scaling / warping) of the present invention; Figure 3 This is a schematic diagram of the visual measurement and feature extraction process of the present invention; Figure 4 This is a schematic diagram of the electrical measurement (open circuit / short circuit / resistance network consistency) of the present invention; Figure 5 This is a schematic diagram of the optical metrology (wavefront / haze / transmittance) station of the present invention; Figure 6 This is a closed-loop control block diagram of the metering-compensation-recharge system of the present invention; Figure 7 This is a schematic diagram of the rework window and decision tree (pass / rework / downgrade / scrap) of the present invention; Figure 8 This is a schematic diagram of the warp / scaling field estimation and correction of the present invention; Figure 9 This is a cross-sectional view of the laminated structure and a schematic diagram of the adhesive layer stress relief structure of the present invention; Figure 10 This is a schematic diagram comparing the wavefront error before and after alignment compensation according to the present invention; Figure 11This is a schematic diagram of the interface between the alignment mark edge and the adhesive layer in the simulated SEM style of this invention. Figure 12 This is a schematic diagram of the traceability data packet fields and log chain summary of the present invention; Figure 13 This is a schematic diagram illustrating the definitions of similarity transformation, thickness coupling term, and nonlinear displacement field parameters in the composite alignment model of this invention. Detailed Implementation

[0036] The technical solutions will now be clearly and completely described with reference to the accompanying drawings in the embodiments of the present invention.

[0037] First implementation method: Please see Figure 1 A high-precision alignment and bonding system for flexible optical components, along with an online measurement and rework system, includes: The stage and clamping mechanism (i.e., alignment platform) are used to clamp the flexible electronic backplane and the optical substrate with optical microstructure respectively, and to apply tension control to the flexible electronic backplane. The alignment reference structure includes visual reference marks and mechanical references set on the flexible electronic backplane and the optical substrate, which are used to establish at least one pair of alignment coordinate systems; The machine vision measurement module is used to acquire the visual reference marks and estimate translation, rotation, scaling and nonlinear deformation errors; The bonding and curing module is used to bond and cure the flexible electronic backplane to the optical substrate through an optically transparent adhesive layer in the alignment state. An online multi-physical quantity measurement module, including at least one of a visual measurement module, an electrical measurement module and an optical measurement module, is used to measure alignment error and device performance in at least two stages: before bonding, during bonding and after bonding. The closed-loop controller communicates with the machine vision measurement module and the online multi-physical quantity measurement module. The closed-loop controller is used to perform alignment error decomposition and adaptively adjust the bonding parameters and update the alignment compensation amount according to the measurement results. The system is configured to make a rework decision based on online measurement results within a preset rework window and output a traceability data packet bound to the device.

[0038] The alignment reference structure includes multi-scale reference markers, which include at least coarse alignment markers and fine alignment markers to support progressively converging alignment from millimeter to micrometer scale.

[0039] like Figure 2As shown, the closed-loop controller decomposes the alignment error into translation errors Δx and Δy, rotation error Δθ, scaling error Δs, and warping / nonlinear deformation term, and outputs a compensation transformation matrix T*; where Δs is the deviation of the scaling factor s from 1.

[0040] The stage and clamping mechanism include a tension sensor and a tension closed-loop control unit. The tension closed-loop control unit adjusts the traction speed or clamping displacement of the clamping mechanism based on feedback from the tension sensor, so that the in-plane strain of the flexible electronic backplane is within a preset range (the stress release direction on the flexible electronic backplane is as follows). Figure 9 As shown, Figure 9 The cross-sectional structure of the flexible electronic backplane, shown from top to bottom (from the lower support structure to the upper transparent structure), is designed to reduce secondary displacement caused by stress and curing shrinkage after bonding.

[0041] The online multi-physical quantity measurement module includes a wavefront measurement module, which is used to measure the wavefront phase Φ(x,y) after bonding and calculate the equivalent optical path difference OPD(x,y)=λ·Φ(x,y) / (2π) to evaluate the wavefront error caused by alignment and curing.

[0042] The online multi-physical quantity measurement module includes an electrical continuity test module, which is used to detect electrode open circuits, short circuits, and sheet resistance drift according to a preset test sequence.

[0043] The bonding and curing module includes at least one of a UV curing unit and a thermal curing unit, and performs closed-loop control on the corresponding curing dose curve or temperature curve to suppress secondary offset caused by curing shrinkage.

[0044] The system also includes a rework station, which is used to perform disassembly, cleaning, re-adhesion and re-bonding within the rework window.

[0045] The closed-loop controller updates the process parameter vector p=[p1,…,pm] based on the online metering results and forms a batch normalization strategy.

[0046] The system is compatible with at least one production line configuration, either roll-to-roll (R2R) or sheet-to-sheet (S2S), and maintains alignment accuracy when the production line speed changes.

[0047] Combination Figure 6 As shown, the high-precision alignment and bonding process for flexible optical components, along with online measurement and rework, includes the following steps: Step A: Establish the coordinate system Σ_E of the flexible electronic backplane and the coordinate system Σ_O of the optical substrate, and collect visual reference marks.

[0048] Step B: Estimate the geometric transformation from Σ_E to Σ_O and calculate the alignment error; Specifically: the geometric transformation is estimated using a homography matrix or affine model, and a piecewise deformation model or spline deformation model is introduced when warping exists.

[0049] Step C: Decompose the alignment error and output the compensation transformation T* to adjust the stage attitude, tension or bonding path; Specifically: the compensation transformation is estimated by minimizing the objective function J, where: J=Σ_{i=1}^{N}||q_i-Tp_i||_2^2+η_reg||w||_2^2; Where p_i is the reference point in coordinate system Σ_E, q_i is the corresponding point in coordinate system Σ_O, T is the similarity transformation matrix, w is the deformation field parameter vector, and η_reg is the regularization coefficient.

[0050] Step D: After adjustment, perform bonding and curing, and perform online multi-physical quantity measurement in at least two stages: before bonding, during bonding, and after bonding; Specifically: During the curing process, feedback control of the dosage curve or temperature curve is required to ensure that the secondary offset caused by curing shrinkage is less than a threshold; online multi-physical quantity measurement includes at least electrical continuity testing and wavefront measurement. In one embodiment, step D is further subdivided into: step D1, performing visual reference measurement before bonding; step D2, performing online measurement of tension, thickness, and electrical parameters during bonding; step D3, performing optical performance acceptance after bonding and initial curing; and step D4, generating parameter rollback points based on the optical indicators output from step D3. The rework determination in step E uses the outputs of the aforementioned steps D3 and D4 as inputs.

[0051] Step E: Determine whether rework is required based on the measurement results in the rework window; Specifically: the determination of rework is based on at least one of the alignment error threshold, wavefront RMS threshold and electrical yield threshold. Within the rework window, the process is to disassemble → clean → re-apply adhesive → re-bond, and an upper limit is imposed on the number of reworks. Step F: Output the traceability data packet and bind it to the device serial number; Specifically, the traceability data package includes at least: alignment reference version, algorithm / model version, bonding and solidification parameters, online measurement results, rework records, original data hash digest, reason code, and digital signature field. A hash digest is calculated for the traceability data package and written to an immutable log chain. The hash digest is then digitally signed using a private key to support cross-station consistency verification and after-sales traceability. The traceability data package is used to implement closed-loop management from manufacturing to calibration to after-sales service, enabling the reproduction of assembly and measurement conclusions during the after-sales phase.

[0052] Step G: Map the alignment error statistics, cause codes, and key measurement indicators to process parameter update quantities Δp, and execute adaptive compensation and batch normalization strategies in subsequent batches.

[0053] The alignment error and wavefront error are mapped to process parameter update amounts Δp, and adaptive compensation is performed in subsequent batches.

[0054] The above process also includes: 1. Closed-loop control of the tension of the flexible backsheet to keep the in-plane strain within a preset range; 2. Online estimation of the adhesive layer thickness field h(x,y), decomposing the thickness field into average thickness deviation h_bar and thickness gradients g_x and g_y, and obtaining at least one of the lateral correction amounts Δx_h and Δy_h and the optical path difference correction amount ΔOPD caused by the thickness deviation through pre-calibrated thickness-lateral coupling relationship and thickness-optical coupling relationship, so as to incorporate the thickness deviation into the error decomposition model; 3. After completing the bonding in step D, online optical measurement is performed and before the rework judgment in step E is executed, optical performance acceptance operation is performed, and parameter rollback points that can be used for after-sales review are generated based on at least one of wavefront root mean square error, haze, transmittance and image quality MTF.

[0055] Thickness deviation is not directly equivalent to translation error Δx, Δy, rotation error Δθ, or scaling error Δs. Instead, it participates in subsequent calculations as an additional coupling quantity: First, when there is tilted observation, lamination compression, or local height difference, the thickness deviation is mapped to a lateral correction quantity [Δx_h,Δy_h]^T=K_h[h_bar,g_x,g_y]^T through a pre-calibrated sensitivity matrix K_h; Second, the thickness deviation introduces an optical path difference correction quantity ΔOPD(x,y)=(n_adh-n_0)·δh(x,y), which further affects the root mean square error of the wavefront and the image quality MTF; Third, the thickness deviation is also used to update the adhesive amount, pressing pressure, bonding speed, and gap settings, thereby forming a closed-loop update of design-process-metering.

[0056] The joint determination includes hard threshold gating and fusion scoring gating; combined with Figure 5As shown, when any of the key metrological indicators among the wavefront root mean square error W_RMS, leakage current I_leak, resistive network deviation R_dev, and haze H_haze exceeds the corresponding hard thresholds τ_W (corresponding to wavefront root mean square error W_RMS), τ_I (corresponding to leakage current), τ_R (corresponding to resistive network deviation R_dev), and τ_H (corresponding to haze H_haze), it is judged as a failure and a reason code is generated. When all key indicators meet the hard thresholds, the fusion score S is calculated and compared with the score threshold τ_S to determine whether it passes or enters the rework process. The key metrological indicators also include the image quality index MTF calculated based on the wavefront metrological results. The MTF is obtained by the Fourier transform of the point spread function PSF, and the minimum MTF value at the preset spatial frequency f_0 is used in the calculation of the fusion score S (for the specific method of joint judgment, please refer to the "Joint Judgment Process" below).

[0057] Based on the above system and method, the implementation process of this application is described in detail below: I. Key Models and Decision Logic: (1) Definition of coordinate system and variables: Σ_E is the backplate reference coordinate system, Σ_O is the optical substrate coordinate system, and Σ_P is the pupil / wavefront measurement coordinate system; p_i=[x_i,y_i]^T represents the design coordinate of the i-th reference point in coordinate system Σ_E, and q_i=[x_i_prime,y_i_prime]^T represents the measured coordinate of the corresponding reference point in coordinate system Σ_O; Δx and Δy are translation errors, Δθ is rotation error, and Δs is scaling error, where Δs=s-1; h(x,y) is the adhesive layer thickness field, h_bar is the average thickness deviation, g_x=∂h / ∂x and g_y=∂h / ∂y are the thickness gradients; δh(x,y)=h(x,y)-h_ref(x,y) is the relative reference thickness deviation.

[0058] (2) To simultaneously compensate for translation, rotation, scaling, thickness coupling, and nonlinear deformation, a composite model is adopted: q_i=Tp_i+d_h(x_i,y_i)+[u(x_i,y_i),v(x_i,y_i)]^T+ε_i, where ε_i is the residual term, T is the similarity transformation matrix, and d_h(x_i,y_i) is the coupling correction term caused by thickness deviation. The similarity transformation matrix can be written as T=[[scosθ,-ssinθ,t_x],[ssinθ,scosθ,t_y],[0,0,1]] (e.g. Figure 13 As shown in the figure, t_x and t_y are the translations in the x and y directions, respectively.

[0059] The in-plane nonlinear displacement fields u(x,y) and v(x,y) are expanded using basis functions: u(x,y)=Σ_{k=1}^{K}a_kψ_k(x,y), v(x,y)=Σ_{k=1}^{K}b_kψ_k(x,y). Here, K is the number of basis functions; ψ_k(x,y) is a pre-selected two-dimensional spatial basis function. In one embodiment, it can be a tensor product cubic B-spline basis function ψ_k(x,y)=B_3((x-ξ_m) / h_x)·B_3((y-η_n) / h_y), where k corresponds to the control node (m,n); in another embodiment, Zernike basis functions can also be used. a_k represents the coefficient of the k-th basis function in the x-direction displacement field, and b_k represents the coefficient of the k-th basis function in the y-direction displacement field; both have the dimension of length.

[0060] The thickness coupling term d_h(x_i,y_i) is used to characterize the effect of thickness deviation on lateral and optical errors. In one embodiment, d_h(x_i,y_i)=[Δx_h(x_i,y_i),Δy_h(x_i,y_i)]^T, where: Δx_h=c_x0h_bar+c_x1g_x+c_x2g_y; Δy_h = c_y0h_bar + c_y1g_x + c_y2g_y; the coefficients c_x0, c_x1, c_x2, c_y0, c_y1, and c_y2 are obtained from offline calibration. Therefore, the thickness deviation does not replace Δx, Δy, Δθ, and Δs, but rather serves as an additional coupling quantity to correct the geometric estimate and simultaneously provide the optical path difference correction.

[0061] Unknown parameters are solved using least squares or iterative optimization: Its objective function can be written as: J=Σ_{i=1}^{N}||q_i-Tp_i-d_h(x_i,y_i)-[u(x_i,y_i),v(x_i,y_i)]^T||_2^2+η_reg(Σ_{k=1}^{K}a_k^2+Σ_{k=1}^{K}b_k^2)+η_h(h_bar^2+g_x^2+g_y^2). Here, η_reg is the deformation field regularization coefficient, and η_h is the thickness coupling term regularization coefficient. The obtained parameter set is then mapped to compensation control variables, for example, Δp=[Δx,Δy,Δθ,Δs,Δx_h,Δy_h,ΔOPD]^T.

[0062] (3) Specific implementation of visual metrology and electrical metrology: Please refer to Figure 3, in visual metrology, first obtain the alignment mark image I(x, y), and perform flat-field correction and denoising processing; in one embodiment, Gaussian filtering, bilateral filtering or median filtering can be used to obtain I_f(x, y). Subsequently, adaptive threshold segmentation or Otsu binarization is performed on I_f(x, y) to obtain the marked area B(x, y). For circular or annular marks, edge extraction and ellipse fitting can be used to obtain the sub-pixel center; for cross-shaped, L-shaped or chessboard corner marks, the corner response function C = det(M) - k·tr(M)^2 can be used to obtain the initial corner points, and then sub-pixel refinement can be achieved through local quadratic surface fitting or gradient least squares, so as to output the reference point coordinates q_i and their confidence levels w_i.

[0063] Please refer to Figure 4 , in electrical metrology, connect each ring electrode, wire or resistor network node to the test matrix through terminals or FFC interfaces; measure the resistance R_meas,j of each branch using the two-wire method or four-wire method, and compare it with the nominal value R_nom,j. The resistance network deviation can be defined as R_dev = sqrt((1 / M)Σ_{j = 1}^{M}((R_meas,j - R_nom,j) / R_nom,j)^2), the leakage current index I_leak is the maximum leakage current between adjacent electrode pairs under the preset test voltage V_test, the short-circuit criterion can be written as R_mn < R_short, and the open-circuit criterion can be written as R_mn > R_open. The test results are encoded as cause codes, such as RC_OPEN, RC_SHORT, RC_RSPEC and RC_LEAK, and enter the joint determination together with the optical metrology conclusion.

[0064] (4) Joint determination process: A "hard threshold gating + fusion scoring" strategy is adopted. The wavefront root mean square error W_RMS is defined as the standard deviation of the wavefront after removing spike, tip, and tilt within the pupil region Ω_P; I_leak represents the maximum leakage current of adjacent electrode pairs or resistor networks under the preset test voltage V_test; R_dev represents the statistical deviation of the measured resistance spectrum from the nominal resistance spectrum; H_haze represents haze, i.e., the percentage of large-angle scattered transmission flux relative to the total transmission flux. Hard threshold gating is used to quickly isolate unacceptable samples: when W_RMS>τ_W, I_leak>τ_I, R_dev>τ_R, or H_haze>τ_H, failure is directly determined and a reason code is output. The fusion score S is used to make robust judgments on boundary samples. One embodiment is S=Σ_{j=1}^{M}α_j·(1-min(1,m_j / τ_j)), where α_j is the weight of the j-th index and satisfies Σα_j=1, m_j is the measured value of the j-th index, and τ_j is the corresponding normalization threshold. When S<τ_S, the rework process is initiated, and when S≥τ_S, the judgment is passed.

[0065] (5) Origin and application of image quality MTF: The wavefront measurement coordinate system and the pupil coordinate system Σ_P are calibrated and transformed, and the root mean square error of the wavefront W_RMS, the image quality MTF and its spatial weights are calculated in the pupil coordinate system Σ_P. To avoid symbol confusion, the detection wavelength is denoted as λ_0 and the wavefront distribution is denoted as W_ab(ρ), where ρ=[ρ_x,ρ_y]^T is the normalized pupil coordinate; The generalized pupil function P(ρ) is calculated based on the wavefront distribution W_ab(ρ) measured under Σ_P. P(ρ) takes the value 1 within the effective aperture and 0 outside the aperture. When considering the gradual change in transmittance, P(ρ) can also be written as A(ρ)·M(ρ), where A(ρ) is the amplitude weighting term and M(ρ) is the aperture mask. The point spread function is defined as PSF(x,y)=|F^{-1}{P(ρ)·exp[i·2π·W_ab(ρ) / λ_0]}|^2, the corresponding optical transfer function is OTF(f)=F{PSF(x,y)} / F{PSF(0,0)}, and the image quality modulation transfer function is MTF(f)=|OTF(f)|. In this application, the minimum MTF value at the preset spatial frequency f_0 is denoted as MTF_min(f_0), which participates in the calculation of the fusion score S and can also be used as a hard threshold gate term to characterize whether the image quality after bonding meets the acceptance requirements. Haze H_haze is used to characterize wide-angle scattering caused by adhesive bubbles, interface contamination, or scattering defects, and its calculation can be done by H_haze=Φ_scatter / Φ_total×100%. Therefore, W_RMS, MTF, and H_haze correspond to three different dimensions: wavefront distortion, image sharpness, and scattering contamination, respectively.

[0066] (6) Rework window: The rework window is jointly defined by cure degree, removability, and number of reworks. In one embodiment, the cumulative curing energy \(E_{cum}(t)=\int_{0}^{t}I_{uv}(\tau)d\tau\), and the cure degree can be approximately expressed as \(c(t) = 1-\exp[-k_c\cdot E_{cum}(t)]\). When \(c(t)\leq c_{max}\) (maximum cure degree), the peel force \(F_{peel}\leq F_{max}\) (maximum peel force), and the number of reworks \(n_{rw}<n_{max}\) (i.e., the maximum number of reworks), peeling rework is allowed; otherwise, rework is not allowed and it enters the downgrade or scrapping judgment (as Figure 7 shown). When reworking, record the number of reworks \(n_{rw}\), the error \(\Delta e\) before and after rework, and the update amount \(\Delta p\), and write them into the trace data packet.

[0067] II. Example: Take the lamination of a flexible optical component as an example: First, before lamination, identify multi-scale fiducial points through machine vision and estimate \(T\) and \((u, v)\), and stabilize the initial in-plane strain within a preset window in combination with tension closed-loop control; then perform pre-lamination, main lamination, and curing. After lamination, perform wavefront metrology and electrical continuity testing, and output a conclusion based on the above joint determination.

[0068] In a group of comparative samples, after introducing deformation field compensation, the alignment residual error is reduced from 35 μm to 8 μm (see Figure 8 illustration); at the same time, the wavefront RMS is reduced from 0.42λ to 0.15λ (see Figure 10 illustration), and the number of reworks and rework costs are significantly reduced.

[0069] In this embodiment, the optical performance acceptance results output by step D3 at least include \(W_{RMS}\), \(MTF_{min}(f_0)\), \(H_{haze}\), and transmittance \(T(\lambda)\); step D4 then writes the above results together with the corresponding process parameter vector \(p\), compensation matrix \(T^*\), and thickness field summary \(h_{bar} / g_x / g_y\) into the parameter rollback point for subsequent after-sales review or batch traceability.

[0070] The rework window can also be jointly defined by the cumulative curing energy \(E_{cum}\) and time \(t\): When \(E_{cum}\leq E_{rw}\) and \(t\leq t_{rw}\), it is considered to still be within the reworkable window.

[0071] A quantitative index \(R_d\) of "removability" (such as peel force, residual glue rate, or interface integrity score) can also be introduced to form a composite criterion: \(G_{rw} = 1\) if and only if \(c(t)\leq c_{max}\), \(R_d\leq R_{dmax}\), and \(n_{rw}<n_{max}\); otherwise \(G_{rw} = 0\).

[0072] The aforementioned composite criterion is used to unify material condition, process sequence, and rework history into the same rework decision logic, thereby avoiding continued rework when the interface is already irreversibly damaged.

[0073] Example of a compensation control law (linear): Δp = -K_ge_geo - K_we_wave - K_ee_ele - K_he_thk, where e_geo, e_wave, e_ele, and e_thk represent the geometric, electrical, optical, and thickness coupling error vectors, respectively, and K_g, K_w, K_e, and K_h are the corresponding gain matrices.

[0074] More generally, the compensation solution can be written as a constrained optimization: Δp*=argmin_{Δp}J_ctrl(Δp), where J_ctrl(Δp)=||AΔp-e||_2^2+μ||Δp||_2^2; the constraints are p_min≤p+Δp≤p_max, v_min≤v_line≤v_max, P_min≤P_press≤P_max.

[0075] Where A represents the sensitivity matrix from error to process parameter update, e is the aggregated error vector, μ is the control regularization coefficient, and J_ctrl(Δp) represents the control objective function used to solve for the compensation update. In one embodiment, when J_ctrl(Δp*)≤τ_ctrl and the key measurement indicators meet the hard threshold gate, the update Δp* is accepted; otherwise, the process proceeds to rework, downgrade, or scrap.

[0076] III. Data Packet Traceability and Tamper-Proof Audit Chain: To achieve supply chain-level traceability, key events and measurement results are written to a log chain: log_k={SN,station_id,ts_k,metric_hash_k,rc_k,sig_k}, where SN is the device serial number, station_id is the workstation identifier, ts_k is the timestamp, metric_hash_k is the summary of the measurement data for this time, rc_k is the reason code, and sig_k is the signature field.

[0077] A chained digest can be represented as: h_k = Hash(h_{k-1}||SN||station_id||ts_k||metric_hash_k||rc_k), where Hash(·) represents the hash function. Any modification to the data at any stage will cause h_k to mismatch with subsequent chain values, thus achieving tamper-proof verification.

[0078] Here, Hash(·) can be a hash function such as SHA-256; examples of fields that can be included in the traceable data packet are as follows. Figure 12 As shown.

[0079] This system and method can be implemented by a processor in an electronic device executing computer program instructions. The electronic device includes at least a processor, a memory, and interfaces for communicating with the machine vision metrology module, wavefront metrology module, and electrical metrology module, respectively. The processor is configured to generate audit logs and calculate hash digests and digital signatures for traceability data packets to support cross-workstation consistency verification and after-sales tamper-proof verification.

[0080] The above description is merely a preferred embodiment of the present invention; it encompasses all the protection scope of the present invention. Any equivalent substitutions or modifications made by those skilled in the art within the technical scope disclosed in the present invention, based on the technical solutions and improved concepts of the present invention, should be covered within the protection scope of the present invention.

Claims

1. A high-precision alignment and bonding system for flexible optical components, characterized in that: include: The stage and clamping mechanism are used to clamp the flexible electronic backplane and the optical substrate with optical microstructure respectively, and to apply tension control to the flexible electronic backplane. The alignment reference structure includes visual reference marks and mechanical references set on the flexible electronic backplane and the optical substrate, which are used to establish at least one pair of alignment coordinate systems; The machine vision measurement module is used to acquire the visual reference marks and estimate translation, rotation, scaling and nonlinear deformation errors; The bonding and curing module is used to bond and cure the flexible electronic backplane to the optical substrate through an optically transparent adhesive layer in the alignment state. An online multi-physical quantity measurement module, including at least one of a visual measurement module, an electrical measurement module and an optical measurement module, is used to measure alignment error and device performance in at least two stages: before bonding, during bonding and after bonding. The closed-loop controller communicates with the machine vision measurement module and the online multi-physical quantity measurement module. The closed-loop controller is used to perform alignment error decomposition and adaptively adjust the bonding parameters and update the alignment compensation amount according to the measurement results. The system is configured to make a rework decision based on online measurement results within a preset rework window and output a traceability data packet bound to the device.

2. The high-precision alignment and bonding system for flexible optical components and the online measurement and rework system according to claim 1, characterized in that: The alignment reference structure includes multi-scale reference markers, which include at least coarse alignment markers and fine alignment markers.

3. The high-precision alignment and bonding system for flexible optical components and the online measurement and rework system according to claim 2, characterized in that: The closed-loop controller decomposes the alignment error into translation errors Δx and Δy, rotation error Δθ, scaling error Δs, and warping / nonlinear deformation term, and outputs a compensation transformation matrix T*; where Δs is the deviation of the scaling factor s from 1.

4. The high-precision alignment and bonding system for flexible optical components and online measurement and rework system according to claim 3, characterized in that: The stage and clamping mechanism include a tension sensor and a tension closed-loop control unit. The tension closed-loop control unit adjusts the traction speed or clamping displacement of the clamping mechanism according to the feedback from the tension sensor, so that the in-plane strain of the flexible electronic backplane is within a preset range, thereby reducing the secondary displacement caused by stress and curing shrinkage after bonding.

5. The high-precision alignment and bonding system for flexible optical components and the online measurement and rework system according to claim 4, characterized in that: The online multi-physical quantity measurement module includes a wavefront measurement module, which is used to measure the wavefront phase Φ(x,y) after bonding and calculate the equivalent optical path difference OPD(x,y)=λ·Φ(x,y) / (2π) to evaluate the wavefront error caused by alignment and curing.

6. The high-precision alignment and bonding system for flexible optical components and online measurement and rework system according to claim 5, characterized in that: The online multi-physical quantity measurement module includes an electrical continuity test module, which is used to detect electrode open circuits, short circuits, and sheet resistance drift according to a preset test sequence.

7. The high-precision alignment and bonding system for flexible optical components and online measurement and rework system according to claim 6, characterized in that: The bonding and curing module includes at least one of a UV curing unit and a thermal curing unit, and performs closed-loop control on the corresponding curing dose curve or temperature curve to suppress secondary offset caused by curing shrinkage.

8. The high-precision alignment and bonding system for flexible optical components and online measurement and rework system according to claim 7, characterized in that: The system also includes a rework station, which is used to perform disassembly, cleaning, re-adhesion and re-bonding within the rework window.

9. The high-precision alignment and bonding system for flexible optical components and online measurement and rework system according to claim 8, characterized in that: The closed-loop controller updates the process parameter vector p=[p1,…,pm] based on the online metering results and forms a batch normalization strategy.

10. The high-precision alignment and bonding system for flexible optical components and online measurement and rework system according to claim 9, characterized in that: The system is adapted to at least one production line configuration, either roll-to-roll or sheet-to-sheet, and maintains alignment accuracy when the production line speed changes.

11. A high-precision alignment and bonding and online measurement rework process for flexible optical components, applicable to the high-precision alignment and bonding and online measurement rework system described in claim 10, characterized in that: Includes the following steps: Step A: Establish the coordinate system Σ_E of the flexible electronic backplane and the coordinate system Σ_O of the optical substrate, and collect visual reference marks; Step B: Estimate the geometric transformation from Σ_E to Σ_O and calculate the alignment error; Step C: Decompose the alignment error and output the compensation transformation T* to adjust the stage attitude, tension or bonding path; Step D: After adjustment, perform bonding and curing, and perform online multi-physical quantity measurement in at least two stages: before bonding, during bonding, and after bonding; Step E: Determine whether rework is required based on the measurement results in the rework window; Step F: Output the traceability data packet and bind it to the device serial number; Step G: Map the alignment error statistics, cause codes, and key measurement indicators to process parameter update quantities Δp, and execute adaptive compensation and batch normalization strategies in subsequent batches.

12. The high-precision alignment and bonding and online measurement rework process for flexible optical components according to claim 11, characterized in that: In step B, the geometric transformation is estimated using a homography matrix or affine model, and a piecewise deformation model or spline deformation model is introduced when warping exists.

13. The high-precision alignment and bonding and online measurement rework process for flexible optical components according to claim 12, characterized in that: In step C, the compensation transformation is estimated by minimizing the objective function J, where: J=Σ_{i=1}^{N}||q_i-Tp_i||_2^2+η_reg||w||_2^2; Where p_i is the reference point in coordinate system Σ_E, q_i is the corresponding point in coordinate system Σ_O, T is the similarity transformation matrix, w is the deformation field parameter vector, and η_reg is the regularization coefficient.

14. The high-precision alignment and bonding and online measurement rework process for flexible optical components according to claim 13, characterized in that: In step D, the dosage curve or temperature curve of the curing process is controlled by feedback to ensure that the secondary offset caused by curing shrinkage is less than the threshold.

15. The high-precision alignment and bonding and online measurement rework process for flexible optical components according to claim 14, characterized in that: In step D, the online multi-physical quantity measurement includes at least electrical continuity testing and wavefront measurement, and jointly determines whether the assembly threshold is met.

16. The high-precision alignment and bonding and online measurement rework process for flexible optical components according to claim 15, characterized in that: The determination of rework in step E is based on at least one of the following: alignment error threshold, wavefront RMS threshold, and electrical yield threshold.

17. The high-precision alignment and bonding and online measurement rework process for flexible optical components according to claim 16, characterized in that: Within the rework window, the process involves disassembly, cleaning, re-adhesive application, and re-bonding, with an upper limit on the number of rework attempts.

18. The high-precision alignment and bonding and online measurement rework process for flexible optical components according to claim 17, characterized in that: The alignment error and wavefront error are mapped to process parameter update amounts Δp, and adaptive compensation is performed in subsequent batches.

19. The high-precision alignment and bonding and online measurement rework process for flexible optical components according to claim 18, characterized in that: The traceability data package includes at least: alignment reference version, algorithm / model version, bonding and curing parameters, online measurement results, rework records, original data hash digest, reason code, and digital signature field.

20. The high-precision alignment and bonding and online measurement rework process for flexible optical components according to claim 19, characterized in that: It also includes closed-loop control of the tension of the flexible backsheet to keep the in-plane strain within a preset range.

21. The high-precision alignment and bonding and online measurement rework process for flexible optical components according to claim 20, characterized in that: It also includes online estimation of the adhesive layer thickness and incorporating thickness deviation into the error decomposition model.

22. The high-precision alignment and bonding and online measurement rework process for flexible optical components according to claim 21, characterized in that: After bonding, an optical performance acceptance operation is performed and a parameter rollback point is generated that can be used for after-sales review.

23. The high-precision alignment and bonding and online measurement rework process for flexible optical components according to claim 22, characterized in that: The traceability data package is used to achieve closed-loop management of manufacturing → calibration → after-sales service.

24. The high-precision alignment and bonding and online measurement rework process for flexible optical components according to claim 23, characterized in that: The joint determination includes hard threshold gating and fusion score gating; when any of the key measurement indicators, such as wavefront root mean square error W_RMS, leakage current I_leak, resistor network deviation R_dev, and haze H_haze, exceeds the corresponding hard threshold, it is determined as a failure and a reason code is generated; when all key measurement indicators meet the hard threshold, the fusion score S is calculated and compared with the score threshold to determine whether it passes or enters the rework process.

25. The high-precision alignment and bonding and online measurement rework process for flexible optical components according to claim 24, characterized in that: The key metrology indicators also include the image quality index MTF calculated based on the wavefront metrology results. The MTF is obtained by the Fourier transform of the point spread function PSF, and the minimum MTF value at the preset spatial frequency f_0 is used in the calculation of the fusion score S.

26. The high-precision alignment and bonding and online measurement rework process for flexible optical components according to claim 25, characterized in that: A calibration transformation is performed between the wavefront measurement coordinate system and the pupil coordinate system Σ_P, and the root mean square error of the wavefront W_RMS, the image quality MTF, and their spatial weights are calculated in the pupil coordinate system Σ_P.

27. The high-precision alignment and bonding and online measurement rework process for flexible optical components according to claim 26, characterized in that: The hash digest of the traceability data packet is calculated and written into an immutable log chain, and the hash digest is digitally signed using a private key to support cross-workstation consistency verification and after-sales traceability.

Images