Through-hole three-dimensional detection system and method

Abstract

The application discloses a kind of through-hole three-dimensional detection system and method, the system includes: double optical comb light source module, signal optical comb and local oscillator optical comb with preset repetition frequency difference are generated;Beam control module, signal optical comb is focused and scanned to through-hole with oblique incidence angle, so that it occurs multi-path reflection including light-approaching surface reflection and back surface reflection, and through changing azimuth angle and radial distance, polar coordinate multi-point scanning is realized;Main detection channel module, receive reflected signal and carry out coherent detection with local oscillator optical comb, to output corresponding detection signal;Data processing module, according to detection signal generation three-dimensional point cloud and carry out back surface geometric reconstruction, obtain the three-dimensional topography and geometric parameter of through-hole;Synchronous control module, according to the control parameter of data processing module control the time sequence of each adjustment action of beam control module.The application solves the problem of blind area of inclined through-hole back surface detection, has the advantages such as single hole detection speed fast, deep width ratio adaptability, high measurement accuracy, complete parameter etc..

Description

Technical Field

[0001] This invention relates to the field of non-destructive testing technology for advanced semiconductor packaging, and in particular to a three-dimensional inspection system and method for through-holes. Background Technology

[0002] In the manufacturing of 3D integrated circuits and high-bandwidth memory (HBM), through-holes (e.g., through-silicon vias, TSVs) are the core structure for realizing vertical interconnects of chips. Their quality parameters (hole diameter, hole depth, sidewall morphology, and perpendicularity) directly affect interconnect reliability and chip yield. As advanced packaging technologies develop towards higher aspect ratios (15:1 and even 20:1) and smaller hole diameters (5μm~10μm), higher precision and higher efficiency are required for the non-destructive testing of the 3D morphology of through-holes.

[0003] Existing through-hole inspection methods have the following main shortcomings:

[0004] (1) Insufficient detection capability for high aspect ratio. Vertical scanning white light interferometry is the current mainstream method for detecting via morphology. However, when the aspect ratio exceeds 10:1, the probe beam is difficult to reach the bottom of the hole, and the signal-to-noise ratio of the reflected signal drops sharply, which cannot meet the detection requirements of high aspect ratio vias in advanced packaging.

[0005] (2) The problem of blind zone on the back side of the tilted through hole. Most existing optical detection methods adopt the vertical incidence method. When the through hole itself is axially tilted, the vertically incident beam can only illuminate the light-facing side of the side wall, and the back side is in the illumination blind zone, resulting in missing side wall contour data, and the depth measurement is affected by the tilt, resulting in systematic deviation.

[0006] (3) Lack of rapid detection methods for the three-dimensional morphology of the entire hole wall. Existing technologies such as CN105043297A rely on the superposition of multi-focal plane images, and the measurement accuracy is limited by the imaging resolution; Weihua Technology SP8000S adopts dual-optical-path vertical incident point scanning for each hole, which has low detection efficiency and also cannot solve the problem of blind area on the backlight surface.

[0007] (4) Insufficient beam control capability. Traditional systems use mechanically moving lens groups to achieve zooming, which has problems such as large size and slow response speed. It is difficult to achieve dynamic compensation for the lateral deviation of the obliquely incident beam, which limits the scanning accuracy and detection efficiency. Summary of the Invention

[0008] The purpose of this invention is to provide a three-dimensional inspection system and method for through holes. By utilizing dual-comb asynchronous optical sampling, oblique incidence multipath reflection scanning, beam control and geometric reconstruction algorithms, non-destructive three-dimensional measurement of the aperture, depth, sidewall morphology and perpendicularity of through holes is achieved.

[0009] To achieve one of the aforementioned objectives, one embodiment of the present invention provides a three-dimensional detection system for through-holes, comprising the following modules: a dual-comb light source module for generating a signal comb and a local oscillator comb with a preset repetition frequency difference; a beam control module for focusing and scanning the signal comb into the through-hole of the sample at an oblique incidence angle relative to the through-hole axis, so that the signal comb undergoes multipath reflection within the through-hole, including reflection from the light-facing surface and reflection from the back surface; polar coordinate multi-point scanning of the through-hole is achieved by changing the azimuth angle and radial distance of the signal comb; and the reflected signal returned from the through-hole is transmitted to the main detection channel module; wherein, the azimuth angle is an angle determined on the plane of the through-hole opening with a preset reference direction as the starting direction. The radial distance is the distance by which the incident light spot deviates from the center of the through-hole opening on the plane of the through-hole opening; the main detection channel module is used to receive the reflected signal and perform coherent detection of the reflected signal with the local oscillator comb to output a corresponding detection signal; the data processing module is signal-connected to the main detection channel module and is used to generate a three-dimensional point cloud based on the detection signal and perform backlight surface geometry reconstruction on the three-dimensional point cloud to obtain the three-dimensional morphology and geometric parameters of the through-hole; the synchronization control module is signal-connected to the data processing module and the beam adjustment module and is used to control the timing of each adjustment action of the beam adjustment module according to the control parameters of the data processing module, so that the incident light spots of each scanning point are aligned sequentially in a predetermined order.

[0010] Furthermore, the beam control module includes an angle adjustment unit and a displacement compensation unit. The angle adjustment unit is used to adjust the oblique incident angle and the azimuth angle of the signal optical comb, and the displacement compensation unit is used to adjust the radial distance by compensating for the lateral offset of the signal optical comb. The synchronization control module controls the timing synchronization between the angle adjustment unit and the displacement compensation unit to ensure that the incident light spots at each scanning point are aligned sequentially in a predetermined order.

[0011] Furthermore, the angle adjustment unit adopts a dichroic scanning mirror, and the displacement compensation unit adopts an Alvarez lens group. The Alvarez lens group includes a first Alvarez lens group and a second Alvarez lens group. The first Alvarez lens group and the second Alvarez lens group achieve lateral offset compensation of the signal optical comb through their respective overall lateral translation, and adjust the focused spot diameter of the signal optical comb through their respective internal relative displacement.

[0012] Further, the polar coordinate multi-point scanning includes scanning multiple first-type scanning points and one second-type scanning point; the first-type scanning points are the scanning points whose radial distance is not equal to 0, the main detection channel module receives the multipath reflection signal generated in the through hole from the first-type scanning points, and outputs a first-type detection signal after coherently detecting the multipath reflection signal with the local oscillator optical comb; the data processing module extracts the depth data of the light-facing side and the depth data of the backlight side according to the first-type detection signal to generate a three-dimensional point cloud; the second-type scanning points are the scanning points whose radial distance is equal to 0, the main detection channel module receives the hole bottom reflection signal from the second-type scanning points, and outputs a second-type detection signal after coherently detecting the hole bottom reflection signal with the local oscillator optical comb; the data processing module extracts the hole bottom depth data according to the second-type detection signal to obtain the depth reference of the through hole.

[0013] Furthermore, the first type of scanning points includes multiple scanning points with different azimuth angles and multiple scanning points with different radial distances; wherein, the multiple scanning points with different azimuth angles are implemented by changing the azimuth angles with a preset step size, and the multiple scanning points with different radial distances are implemented by setting the radial distances to multiple different preset ratios of the through hole diameter.

[0014] Furthermore, the data processing module calculates the target coordinates of each of the first type of scanning points on the plane of the through hole based on the oblique incident angle, the azimuth angle, and the radial distance; the synchronization control module controls the displacement compensation unit to perform lateral offset compensation on the signal optical comb based on the target coordinates, so that the incident light spot is sequentially aligned with each of the first type of scanning points.

[0015] Furthermore, the scanning sequence of the polar coordinate multi-point scanning is as follows: first, each of the first type of scanning points is scanned in descending order of radial distance; after traversing all azimuth angles at the same radial distance, the scanning is switched to the next radial distance until all the first type of scanning points are scanned; finally, the second type of scanning points are scanned.

[0016] Furthermore, the oblique angle of incidence ranges from 5° to 12°.

[0017] Furthermore, the system also includes a preliminary hole positioning module, which is used to establish the global coordinates of all through-hole openings by scanning the entire sample. The preliminary hole positioning module includes: an infrared light source for emitting infrared light; a beam splitter disposed in the output light path of the infrared light source; and an infrared camera disposed in the reflected light path of the beam splitter. The infrared light emitted by the infrared light source is transmitted through the beam splitter and then through the dichroic scanning mirror to illuminate the sample surface. The infrared light reflected from the sample surface is transmitted through the dichroic scanning mirror and then through the beam splitter to be received by the infrared camera.

[0018] Furthermore, the beam control module also includes a switchable objective lens, positioned along the light incident direction in the optical path after the Alvarez lens group, for switching between infrared positioning mode and dual-comb detection mode; the switchable objective lens includes a large field-of-view infrared objective lens and a high numerical aperture focusing objective lens; in the infrared positioning mode, the switchable objective lens switches to the large field-of-view infrared objective lens, and infrared light is irradiated onto the sample surface through the large field-of-view infrared objective lens to establish the global coordinates of all through-hole openings; in the dual-comb detection mode, the switchable objective lens switches to the high numerical aperture focusing objective lens, and the signal light comb is focused and scanned into the through-hole through the high numerical aperture focusing objective lens to perform three-dimensional detection.

[0019] Further, the data processing module acquires the three-dimensional morphology and geometric parameters of the through hole by: acquiring the depth-reflection intensity curve of each of the first type of detection signals; extracting the reflection peak corresponding to the upper surface of the sample from the depth-reflection intensity curve as the zero depth point, and extracting the reflection peak of the light-facing side and the reflection peak of the backlight side as the light-facing side depth data and the backlight side depth data, respectively; generating a three-dimensional point cloud based on the coordinates of each scanning point, the zero depth point, and the corresponding light-facing side depth data and the backlight side depth data; performing backlight side geometry reconstruction on the three-dimensional point cloud, and outputting the three-dimensional morphology and geometric parameters of the through hole.

[0020] Further, the backlighting geometry reconstruction of the three-dimensional point cloud specifically includes: dividing the three-dimensional point cloud into multiple virtual slices along the depth direction; performing circle fitting on the light-facing point cloud in each virtual slice to obtain the center and radius of the corresponding circle for each virtual slice; interpolating and completing the missing point cloud on the backlighting surface based on the circular section assumption and axial continuity; performing three-dimensional line fitting on the center of the corresponding circle for each virtual slice to obtain the central axis of the through hole; and calculating the geometric parameters of the through hole based on the radius of each virtual slice and the central axis.

[0021] Furthermore, the interpolation completion of the missing point cloud on the backlit side specifically includes: for any depth layer, based on the center and radius of the circle fitted on the light-facing side of that layer, interpolating the radial coordinates in the missing area on the backlit side according to the circumferential angle; the interpolation process introduces curvature smoothing constraints to ensure a continuous transition between the completed data and the existing data of adjacent depth layers.

[0022] Furthermore, the beam control module also includes an optical circulator, an optical fiber collimator, and a quarter-wave plate. The optical circulator is disposed between the dual-comb light source module and the dichroic scanning mirror, and is used to transmit the signal optical comb as an incident beam to the dichroic scanning mirror, and to transmit the reflected beam reflected by the sample and returned by the dichroic scanning mirror to the main detection channel module. The optical fiber collimator is disposed between the optical circulator and the dichroic scanning mirror, and is used to expand and collimate the incident beam. The quarter-wave plate is disposed between the dichroic scanning mirror and the Alvarez lens group, and is used to make the polarization states of the incident beam and the reflected beam mutually orthogonal, so that the incident beam and the reflected beam are separated at the optical circulator.

[0023] One embodiment of the present invention also provides a method for three-dimensional detection of through-holes, which uses the three-dimensional detection system for through-holes as described above. The method includes the following steps: S1: performing system initialization and parameter setting, setting the oblique incidence angle and scanning trajectory parameters including multiple scanning points; S2: based on the scanning trajectory parameters, performing single-point dual-comb multipath reflection measurement, focusing the signal comb generated by the dual-comb light source module at the oblique incidence angle to scan the current scanning point in the through-hole of the sample, so that the signal comb generates reflections including light-facing and back-facing reflections within the through-hole. Multipath reflection: Receive the reflected signal returned from the through-hole, and perform coherent detection between the reflected signal and the local oscillator optical comb generated by the dual-comb light source module, outputting a corresponding detection signal; S3: Sequentially switch each scanning point according to the scanning trajectory parameters, repeat step S2 to obtain the detection signal of each scanning point, obtain the corresponding depth information based on the detection signal of each scanning point, and generate a three-dimensional point cloud based on the coordinates of each scanning point and the corresponding depth information; S4: Perform backlight surface geometric reconstruction on the three-dimensional point cloud to obtain the three-dimensional morphology and geometric parameters of the through-hole.

[0024] The beneficial effects of this invention are as follows: by combining oblique incidence multipath reflection scanning with a backlight surface geometric reconstruction algorithm, full-hole wall three-dimensional morphology detection is achieved, effectively solving the industry problem of blind spots in the backlight surface of tilted through holes; dual-comb asynchronous optical sampling technology enables high-precision absolute distance measurement without the need for axial mechanical scanning; combined with polar coordinate multi-point scanning guided by a dichroic scanning mirror and rapid beam control of the Alvarez lens group, detection efficiency is significantly improved; the system can flexibly adapt to the detection needs of through holes with different apertures and aspect ratios, and output multiple parameter detection results such as aperture distribution, hole depth, perpendicularity, taper, and sidewall smoothness in one go; the whole system adopts fiber optic optical path and piezoelectric ceramic drive, with a compact structure, easy integration, and good industrial application prospects. Attached Figure Description

[0025] Figure 1 This is an overall framework diagram of the through-hole three-dimensional detection system of the present invention; Figure 2 This is an overall structural diagram of the through-hole three-dimensional detection system of the present invention; Figure 3 (a) and (b) are schematic diagrams illustrating the working principle of the Alvarez lens group, taking the first Alvarez lens group and the second Alvarez lens group as examples, respectively. Figure 4 (a)-(c) are schematic diagrams of the scanning trajectories of the present invention at different radial distances; Figure 5 This is a flowchart of the three-dimensional detection method for through holes according to the present invention.

[0026] Reference numerals: 10—Dual-comb light source module, 101—Continuous laser, 102—1×2 fiber coupler, 103—Polarization and modulation components, 20—Beam control module, 201—Dichroic scanning mirror, 202—First Alvarez lens group, 203—Second Alvarez lens group, 204—Switchable objective lens, 205—Optical circulator, 206—Fiber collimator, 207—1 / 4 wave plate, 30—Main detection channel module, 301—2×2 fiber coupler, 302—Balanced detector, 40—Data processing module, 50—Synchronization control module, 60—Preliminary aperture positioning module, 601—Infrared light source, 602—Infrared camera, 603—Beam splitter. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0028] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0029] The main improvements of this invention are as follows: (1) Multipath reflection oblique incidence scanning strategy: Utilize the multipath reflection of the beam in the through hole (the first reflection illuminates the light-facing surface, and the second reflection illuminates the back-facing surface), combine multiple scanning points, obtain the discrete reflection point cloud of the entire hole wall at once, and then complete the continuous contour through the reconstruction algorithm, effectively solving the problem of the blind area of ​​the back-facing surface.

[0030] (2) Polar coordinate point-by-point scanning guided by dichroic scanning mirror: The dichroic scanning mirror is used to quickly change the oblique incidence azimuth angle φ and tilt angle α. Combined with the real-time position compensation of the first Alvarez lens group and the second Alvarez lens group, the millisecond-level traversal of dozens of scanning points within the micron-level aperture is realized, which significantly improves the detection efficiency.

[0031] (3) Backlight surface geometric reconstruction algorithm: Based on the circular cross section assumption and axial continuity, the complete outline of the backlight surface is recovered by interpolation from the sparse point cloud, and the hole wall coverage rate of more than 95% is achieved under the condition that the through hole tilt angle is ≤5°.

[0032] (4) Dual-comb asynchronous optical sampling: continuous depth reflection information (A-scan) from the aperture to the bottom of the aperture can be obtained with a single irradiation, without the need for axial mechanical scanning, and the measurement accuracy reaches the nm level.

[0033] like Figure 1 As shown, the present invention provides a three-dimensional detection system for through holes based on oblique incidence multipath reflection of dual light combs, comprising: dual light comb light source module 10, beam control module 20, main detection channel module 30, data processing module 40 and synchronization control module 50.

[0034] The dual-comb light source module 10 generates a signal comb and a local oscillator comb with a preset repetition frequency difference. The beam control module 20 focuses and scans the signal comb into the through-hole of the sample at an oblique incidence angle relative to the through-hole axis, causing multipath reflections including reflections from the light-facing and back-facing surfaces within the through-hole; it achieves polar coordinate multi-point scanning of the through-hole by changing the azimuth angle and radial distance of the signal comb; and it transmits the reflected signals returned from the through-hole to the main detection channel module. The main detection channel module 30 receives the reflected signals, performs coherent detection between the reflected signals and the local oscillator comb, and outputs a corresponding detection signal. The data processing module 40 is signal-connected to the main detection channel module 30, and generates a three-dimensional point cloud based on the detection signal, and performs back-facing surface geometric reconstruction on the three-dimensional point cloud to obtain the three-dimensional morphology and geometric parameters of the through-hole. The synchronization control module 50 is connected to the data processing module 40 and the beam control module 20 by signal. It is used to control the timing of each adjustment action of the beam control module 20 according to the control parameters of the data processing module 40, so that the incident light spots of each scanning point are aligned in a predetermined order.

[0035] Wherein, the azimuth angle is the angle determined on the plane of the via opening with a preset reference direction as the starting direction, and the radial distance is the distance of the incident light spot off the center of the via opening on the plane of the via opening. The preset reference direction can be selected as the direction of the flat edge of the sample surface or the direction of the wafer notch, or it can be selected as the row direction or column direction of the via array, or it can be determined according to the coordinate system calibration of the detection system. For example, in a specific embodiment, the direction passing through the center of the via opening and parallel to the flat edge of the sample is taken as the 0° starting direction, and the range from 0° to 360° is measured in a counterclockwise direction to determine the azimuth angle of each scanning point. It is understood that the specific selection method of the preset reference direction can be flexibly determined by those skilled in the art according to the coordinate system calibration and scanning strategy of the detection system, and the present invention does not specifically limit it in this regard.

[0036] The through-hole 3D detection system of this invention achieves a complete detection closed loop from beam excitation and multipath reflection signal acquisition to 3D topography reconstruction through the coordinated operation of the above five core modules. The beam detection signal provided by the dual-comb light source module, combined with the oblique incidence polar coordinate multi-point scanning of the beam control module, enables the detection beam to actively excite multipath reflections including the light-facing and back-facing surfaces within the through-hole, fundamentally solving the inherent defect of the back-facing surface being in the illumination blind zone in the vertical incidence scheme.

[0037] See Figure 2 The dual-comb light source module 10 may include a continuous laser 101, a 1×2 fiber coupler 102, and a polarization and modulation component 103.

[0038] The continuous laser emitted by the continuous laser 101 is split into two paths by a 1×2 fiber coupler 102. One path, after passing through a corresponding polarization component (polarization controller 1) and a corresponding modulation component (electro-optic modulator 1), generates a signal optical comb with a repetition frequency of [missing information]. f r The other path, after passing through the corresponding polarization component (polarization controller 2) and the corresponding modulation component (electro-optic modulator 2), generates a local oscillator optical comb with a repetition frequency of [missing information]. f r + Δf ,in Δf The signal frequency ranges from hundreds of Hz to kHz. The signal optical comb is transmitted to the beam control module 20 via optical fiber, and the local oscillator optical comb is transmitted to the main detection channel module 30 via optical fiber.

[0039] like Figure 2 As shown, the polarization states of the signal optical comb and the local oscillator optical comb are adjusted to the P state (i.e., the first polarization state) and the S state (i.e., the second polarization state), respectively. Of course, it is understandable that... Figure 2 The P state can be replaced by the S state, and the S state can be replaced by the P state, as long as the orthogonality requirement of the corresponding polarization state is met.

[0040] Preferably, the center wavelength of the dual-comb light source module 10 is selected in the near-infrared band (1500nm-1600nm). The dual-comb light source and fiber optic devices in this band are mature and have stable performance. In addition, silicon has extremely low absorption in this band, which can avoid the additional noise introduced by the photogenerated carrier effect and help improve the detection accuracy.

[0041] The beam control module 20 of the present invention may include an angle adjustment unit and a displacement compensation unit. The angle adjustment unit is used to adjust the oblique incident angle and the azimuth angle of the signal optical comb. The displacement compensation unit is used to adjust the radial distance by compensating for the lateral offset of the signal optical comb. The synchronization control module is used to control the timing synchronization between the angle adjustment unit and the displacement compensation unit so that the incident light spots of each scanning point are aligned sequentially in a predetermined order.

[0042] like Figure 2 The angle adjustment unit may be a dichroic scanning mirror 201, and the displacement compensation unit may be an Alvarez lens group (referred to as the A lens group), wherein the Alvarez lens group includes a first Alvarez lens group 202 and a second Alvarez lens group 203.

[0043] The dichroic scanning mirror 201 is a two-dimensional galvanometer galvanometer system, integrating two galvanometer motors to control the mirror's oscillation around the X and Y axes, respectively. The dichroic scanning mirror 201 reflects a signal light comb and adjusts its oblique incidence angle α and azimuth angle φ, achieving precise control over the beam direction. Here, the oblique incidence angle α is the tilt angle of the signal light comb relative to the through-hole axis, and the azimuth angle φ is the angle determined on the through-hole opening plane with a preset reference direction as the starting direction. In this embodiment, the oblique incidence angle α is preferably in the range of 5°-12°.

[0044] The first Alvarez lens group 202 (i.e., lens group 1) and the second Alvarez lens group 203 (i.e., lens group 2) compensate for the lateral offset of the signal optical comb through their respective overall lateral translations and their respective internal relative displacements. δ x δ y Adjust the focused spot diameter of the signal optical comb, such as... Figure 3 As shown in (a) and (b), the first Alvarez lens group 202 and the second Alvarez lens group 203 are each composed of two phase plates with completely complementary face shapes. The first Alvarez lens group 202 is responsible for lateral translation compensation in the X direction, and the second Alvarez lens group 203 is responsible for lateral translation compensation in the Y direction.

[0045] The Alvarez lens system works by using two phase plates with specific freeform surfaces, whose lateral relative displacement alters the combined optical power. When the two phase plates are perfectly aligned, the entire system functions as a parallel plate; when they are laterally displaced, it functions as either a convex or concave lens. Each Alvarez lens system has two independent operating modes: Overall lateral translation mode: The two phase plates move in the same direction and at the same speed, changing the propagation direction of the light beam and shifting the position of the light spot on the aperture plane to adjust the radial distance r. For example, when it is necessary to move the light spot in the X direction, the two phase plates of the first Alvarez lens group 202 move in the same direction; when it is necessary to move the light spot in the Y direction, the two phase plates of the second Alvarez lens group 203 move in the same direction.

[0046] Internal relative displacement mode: Two phase plates move in opposite directions, changing the equivalent optical power and altering the focusing diameter of the light spot to adapt to the detection requirements of through holes with different diameters. The Alvarez lens group can achieve a large magnification focal length change with a small vertical axis movement, which is beneficial for shortening the zoom stroke and achieving fast zooming.

[0047] The lateral translation and internal relative displacement of the Alvarez lens group are both achieved using piezoelectric ceramic actuators. These actuators feature fast response and high positioning accuracy, enabling micron-level or even nanometer-level displacement control.

[0048] In an embodiment of the present invention, polar coordinate multi-point scanning of the via is achieved by changing the azimuth angle φ and radial distance r of the signal optical comb. The polar coordinate multi-point scanning may include scanning multiple first-type scanning points and one second-type scanning point. The first-type scanning points are those with a radial distance r not equal to 0, and the second-type scanning points are those with a radial distance r equal to 0.

[0049] The main detection channel module 30 receives a multipath reflection signal containing the reflection information of the light-facing surface and the reflection information of the backlight surface from the first type of scanning point, and outputs a first type of detection signal after coherently detecting the multipath reflection signal with the local oscillator comb; the data processing module 40 extracts the depth data of the light-facing surface and the depth data of the backlight surface according to the first type of detection signal to generate a three-dimensional point cloud.

[0050] The main detection channel module 30 receives the reflected signal from the bottom of the hole (or the stage) at the second type of scanning point, and outputs the second type of detection signal after coherently detecting the reflected signal from the bottom of the hole with the local oscillator comb; the data processing module 40 extracts the depth data of the bottom of the hole according to the second type of detection signal to obtain the depth reference of the through hole.

[0051] Preferably, the scanning order of the first type of scanning points is as follows: first, the first type of scanning points are scanned in descending order of radial distance; after traversing all azimuth angles at the same radial distance, the scanning is switched to the next radial distance until all first type of scanning points are scanned; finally, the second type of scanning points are scanned.

[0052] Specifically, the first type of scanning points includes multiple scanning points with different azimuth angles and multiple scanning points with different radial distances. The multiple scanning points with different azimuth angles are achieved by changing the azimuth angles with a preset step size, and the multiple scanning points with different radial distances are achieved by setting the radial distances to multiple preset ratios of the through-hole diameter.

[0053] For example, see Figure 4 For (a)-(c), the focused spot radius is 0.5R, the azimuth angle φ step size is 30° (12 directions in total), and the radial distance r under each φ takes 3 values ​​(0.4R, 0.5R, 0.6R), for a total of 36 scanning points. R represents the orifice radius of the through hole. The scanning sequence is from the outer circle to the inner circle, with φ increasing in each circle.

[0054] It is understood that the values ​​of the azimuth angle φ step size and radial distance r given in the above embodiments are only exemplary parameters and can be flexibly adjusted according to actual detection needs.

[0055] In actual operation, after the upper scanning trajectory parameters are set, the synchronous control module 50 controls the dichroic scanning mirror 201 to deflect the beam to the preset tilt angle α and azimuth angle φ. After the dichroic scanning mirror 201 deflects, the actual landing point of the beam on the aperture plane will be laterally offset relative to the target coordinates. The offset amount is related to α and φ.

[0056] The data processing module 40 calculates the required lateral translation compensation amounts Δx(α, φ) and Δy(α, φ) based on α, φ, and the target coordinates of the current scanning point (x_target = r·cosφ, y_target = r·sinφ). The synchronization control module 50 controls the first Alvarez lens group 202 and the second Alvarez lens group 203 to perform overall lateral translation in the X and Y directions respectively based on these compensation amounts, to compensate for the lateral shift of the light spot caused by the angle change of the dichroic scanning mirror 201, ensuring that the focused light spot accurately falls at the target coordinates.

[0057] As a preferred implementation method, such as Figure 2 As shown, the system of the present invention may further include a preliminary hole positioning module 60 for establishing global coordinates of all through-hole openings by scanning the entire sample.

[0058] like Figure 2 As shown, the preliminary aperture positioning module 60 includes an infrared light source 601, an infrared camera 602, and a beam splitter 603. The infrared light source 601 emits infrared light (preferably 1310nm); the beam splitter 603 (preferably a 50:50 beam splitter) is disposed in the output light path of the infrared light source 601; the infrared camera 602 is disposed in the reflected light path of the beam splitter 603. The infrared light emitted by the infrared light source 601 is transmitted through the beam splitter 603, and then through the dichroic scanning mirror 201 in the beam control module 20, illuminating the sample surface; the infrared light reflected from the sample surface is transmitted through the dichroic scanning mirror 201, then reflected by the beam splitter 603, and received by the infrared camera 602.

[0059] Because the requirements for the objective lens differ between the infrared positioning mode and the dual-comb detection mode—infrared positioning requires a large field of view to quickly scan the entire wafer and establish global coordinates, while dual-comb detection requires a high numerical aperture to focus the beam into a micrometer-sized spot that enters a single through-hole—the beam control module 20 of this invention also includes a switchable objective lens 204, which is positioned along the light incident direction in the optical path after the Alvarez lens group, for switching between the infrared positioning mode and the dual-comb detection mode.

[0060] The switchable objective lens includes a large field-of-view infrared objective lens and a high numerical aperture focusing objective lens. In the infrared positioning mode, the switchable objective lens is switched to the large field-of-view infrared objective lens, and infrared light is irradiated onto the sample surface through the large field-of-view infrared objective lens to establish the global coordinates of all through-hole openings; in the dual-comb detection mode, the switchable objective lens is switched to the high numerical aperture focusing objective lens, and the signal comb is focused and scanned into the through-hole through the high numerical aperture focusing objective lens to perform three-dimensional detection.

[0061] To ensure that the infrared positioning mode and the dual-comb detection mode operate in a time-sharing manner without interference, the synchronization control module 50 is also signal-connected to the preliminary hole positioning module 60 and the dual-comb light source module 10 to control the timing switching between the infrared positioning mode and the dual-comb detection mode. Specifically, the synchronization control module 50 is signal-connected to the continuous laser 101 of the dual-comb light source module 10 and the infrared light source 601 of the preliminary hole positioning module 60, respectively.

[0062] Specifically, the synchronization control module 50 controls the infrared light source 601 to turn on, entering the infrared positioning mode. The infrared light emitted by the infrared light source 601 is transmitted through the beam splitter 603, then through the dichroic scanning mirror 201, and finally through the large field-of-view infrared objective lens in the switchable objective lens 204 to illuminate the sample surface. The infrared signal diffusely reflected from the sample surface returns along the original path, is transmitted through the dichroic scanning mirror 201, and then reflected by the beam splitter 603, and is received and imaged by the infrared camera 602. By controlling the rapid movement of the three-dimensional displacement stage to scan the entire sample, the global coordinates of all through-hole openings are established.

[0063] After infrared positioning is completed, the synchronization control module 50 controls the infrared light source 601 to turn off and the continuous laser 101 to turn on, entering the dual-comb detection mode. At the same time, the switchable objective lens 204 switches to the high numerical aperture focusing objective lens.

[0064] It should be understood that the core of this invention lies in the dual-comb oblique incidence multipath reflection detection technology. The preliminary hole positioning module 60 serves as an auxiliary positioning means, and its specific implementation is not limited to the infrared positioning scheme described above; other optical or mechanical positioning methods can also be used. Even without the preliminary hole positioning module 60, the dual-comb oblique incidence multipath reflection detection system of this invention can still perform three-dimensional detection of through holes at known locations.

[0065] In dual-comb detection mode, in order to separate the incident beam and the reflected beam while sharing the same optical path and to avoid mutual interference between the transmitted and received signals, such as Figure 2 As shown, the beam control module 20 of the present invention further includes an optical circulator 205, an optical fiber collimator 206, and a quarter-wave plate 207.

[0066] The optical circulator 205 is a three-port optical circulator. Its first port receives the signal optical comb transmitted from the dual optical comb light source module 10, its second port is connected to the fiber collimator 206, and its third port is connected to the main detection channel module 30. The optical circulator 205 is used to transmit the signal optical comb as the incident beam to the fiber collimator 206 and to transmit the reflected beam reflected by the sample to the main detection channel module 30.

[0067] The fiber collimator 206 is positioned between the optical circulator 205 and the dichroic scanning mirror 201 to expand and collimate the incident beam, converting the diverging light emitted from the fiber into parallel light, which facilitates the subsequent deflection of the dichroic scanning mirror 201 and the adjustment of the Alvarez lens group.

[0068] A quarter-wave plate 207 is positioned between the dichroic scanning mirror 201 and the Alvarez lens group to ensure that the polarization states of the incident beam incident on the sample surface and the reflected beam reflected back from the sample are orthogonal, thereby enabling separation at the optical circulator 205. The P-state signal comb undergoes a 45° polarization deflection after passing through the quarter-wave plate 207; after reflection from the sample, its polarization state is again deflected by 45° when passing through the quarter-wave plate 207 again, thus ensuring that the polarization states of the reflected beam and the incident beam are orthogonal.

[0069] After the beam control module 20 focuses and scans the signal optical comb into the through hole at an oblique incidence angle, the main detection channel module 30 receives the reflected signal returned from the through hole, performs coherent detection between the reflected signal and the local oscillator optical comb, and outputs the corresponding detection signal.

[0070] like Figure 2 As shown, the main detection channel module 30 includes a 2×2 fiber optic coupler 301 and a balanced detector 302. The first input port of the 2×2 fiber optic coupler 301 receives a reflected beam containing information on the front and back surfaces of the light beam transmitted from the third port of the optical circulator 205; the second input port receives the local oscillator optical comb output from the dual-comb light source module 10. The 2×2 fiber optic coupler 301 outputs an interference signal by mixing the reflected beam with the local oscillator optical comb.

[0071] The balanced detector 302 is connected to the two output ports of the 2×2 fiber optic coupler 301 to receive interference signals and convert them into electrical signals for output. The balanced detector 302 can effectively suppress common-mode noise and improve the detection signal-to-noise ratio.

[0072] As described above, the main detection channel module 30 can receive multipath reflection signals containing the reflection information of the light-facing surface and the reflection information of the backlight surface from the first type of scanning point, and output a first type of detection signal after coherently detecting the multipath reflection signals with the local oscillator optical comb. The main detection channel module 30 receives aperture bottom reflection signals from the second type of scanning point, and outputs a second type of detection signal after coherently detecting the aperture bottom reflection signals with the local oscillator optical comb.

[0073] The detection signal output by the main detection channel module 30 is transmitted to the data processing module 40 for further processing. The data processing module 40 extracts the depth data of the light-facing side and the light-reflecting side based on the first type of detection signal to generate a three-dimensional point cloud. The data processing module 40 also extracts the bottom depth data of the hole based on the second type of detection signal to obtain the depth reference of the through hole.

[0074] Specifically, the data processing module 40 acquires the depth-reflection intensity curve (A-scan) of each first-type scanning point based on the first-type detection signal.

[0075] The data processing module 40 performs a Fourier transform on the beat frequency signal to obtain an A-scan. In the A-scan, different reflection peaks correspond to different depths of the reflection interface—peak 0 corresponds to the upper surface of the sample, peak 1 corresponds to the side wall of the light-facing side (the first reflection point), peak 2 corresponds to the side wall of the back-facing side (the second reflection point), and peak 3 corresponds to the bottom of the hole (the purest when r=0).

[0076] The data processing module 40 extracts the reflection peak (peak 0) corresponding to the upper surface of the sample from the A-scan as the zero depth point, and extracts the reflection peak (peak 1) on the light-facing side and the reflection peak (peak 2) on the shadow side, which are used as the depth data of the light-facing side and the shadow side, respectively. For the second type of scanning point (r=0), the data processing module 40 extracts the reflection peak (peak 3) at the bottom of the hole as the depth reference of the through hole based on the second type of detection signal.

[0077] In the scanning of 36 polar coordinate points, peak 0 information can be obtained at each point, and the accuracy of the depth zero-point calibration can be further improved by averaging. The depth values ​​of all reflection peaks (peak 1, peak 2, peak 3) are corrected relative to peak 0, that is, actual depth = peak depth value - peak 0 depth value, thereby eliminating systematic errors introduced by sample surface unevenness and system optical path drift.

[0078] The data processing module 40 generates a three-dimensional point cloud based on the coordinates, zero depth point, and corresponding depth data of the light-facing and shadow-facing surfaces of each scanning point. The three-dimensional point cloud includes light-facing and shadow-facing point clouds, corresponding to the three-dimensional coordinate data of the light-facing and shadow-facing sidewalls of the through-hole, respectively. For the first type of scanning point (r≠0), the light beam undergoes multipath reflection within the through-hole, generating multipath reflection signals containing both light-facing and shadow-facing reflection information. The data processing module 40 extracts the light-facing and shadow-facing depth data based on these multipath reflection signals to generate the three-dimensional point cloud. For the second type of scanning point (r=0), the data processing module 40 extracts the bottom depth data based on the bottom reflection signal to obtain the depth reference of the through-hole.

[0079] During point cloud generation, reflection points with signal strength below a preset threshold are marked as missing and excluded from subsequent point cloud reconstruction. Point cloud filtering effectively removes noise points, ensuring point cloud quality.

[0080] Data processing module 40 performs backlight surface geometry reconstruction on the 3D point cloud according to the following steps: (1) Axial virtual slicing: The three-dimensional point cloud is divided into multiple virtual slices S0, S1… S along the depth direction at a preset interval Δh (preferably 0.1μm~10μm). N-1 ; (2) Sunlit surface contour fitting: For each virtual slice layer, extract the radial coordinates of all sunlit surface point clouds within that layer, and perform circle fitting (preferably least squares circle fitting) to obtain the circle center O(x) corresponding to each virtual slice layer. k ,y k ) and radius R k ; (3) Point cloud interpolation for missing backlight areas: For each slice, if the point cloud of the backlight area is missing, based on the circular section assumption (the layer is circular) and the continuity of adjacent slices, spline interpolation or radial basis function interpolation is used to fill in the radial coordinates of the backlight points. Curvature smoothing constraints are introduced in the interpolation process to ensure that the completed area transitions naturally with the existing frontlight data. (4) Global axis fitting: The center O of each virtual slice is fitted. k Three-dimensional straight line fitting is performed to obtain the direction vector v of the central axis of the through hole and the spatial position P0. This axis serves as a global consistency constraint, triggering re-interpolation for layers that deviate too much. (5) Calculation of morphological parameters: Based on the radius and central axis corresponding to each virtual slice, calculate the geometric parameters of the through hole: Aperture distribution: D k =2R k Calculate the minimum, maximum, mean, and standard deviation; Hole depth H: from top slice S0 to bottom slice S N 1 axial distance; Verticality θ=arctan (∣ v xy | / v z ); taper γ=(D top D bottom ) / H ; Sidewall smoothness: Standard deviation of roundness error of each layer; Defect Report: Automatically identifies defects such as shrinkage, skewing, burrs, and blockage based on abrupt changes in aperture and out-of-tolerance roundness.

[0081] Preferably, the data processing module 40 can also automatically output a test report, including a pore size distribution chart, pore depth, verticality, taper, sidewall smoothness score, and three-dimensional contour rendering.

[0082] In another embodiment of the present invention, a three-dimensional detection method for through holes is also provided, such as... Figure 5 As shown, the method includes: S1: performing system initialization and parameter setting, setting the oblique incidence angle and scanning trajectory parameters including multiple scanning points; S2: based on the scanning trajectory parameters, performing single-point dual-comb multipath reflection measurement, focusing the signal comb generated by the dual-comb light source module at the oblique incidence angle to the current scanning point in the through hole of the sample, causing the signal comb to undergo multipath reflection including light-facing surface reflection and back-facing surface reflection in the through hole, receiving the reflected signal returned from the through hole, and performing coherent detection with the local oscillator light comb generated by the dual-comb light source module, outputting the corresponding detection signal; S3: sequentially switching each scanning point according to the scanning trajectory parameters, repeating step S2 to obtain the detection signal of each scanning point, obtaining the corresponding depth information based on the detection signal of each scanning point, and generating a three-dimensional point cloud based on the coordinates of each scanning point and the corresponding depth information; S4: performing back-facing surface geometric reconstruction on the three-dimensional point cloud to obtain the three-dimensional morphology and geometric parameters of the through hole.

[0083] In step S1, infrared positioning can be performed first through the preliminary hole positioning module 60 to establish the global coordinates of all via openings. The synchronization control module 50 controls the infrared light source 601 to turn on, entering infrared positioning mode. The infrared light emitted by the infrared light source 601 passes through the beam splitter 603, then through the dichroic scanning mirror 201 (transmitted infrared 1310nm, reflected infrared 1500-1600nm), the quarter-wave plate 207, the Alvarez lens group, and the switchable objective lens (large field-of-view infrared objective lens) 204, before vertically illuminating the wafer surface. The diffusely reflected infrared signal returns along the original path and is transmitted through the dichroic scanning mirror 201, then illuminates the sample surface through the large field-of-view infrared objective lens in the switchable objective lens 204. The diffusely reflected infrared signal from the sample surface returns along the original path, is reflected after passing through the beam splitter 603, and is received and imaged by the infrared camera 602. By controlling the rapid movement of the three-dimensional displacement stage, the entire wafer is scanned to establish the global coordinates of all via openings.

[0084] After infrared positioning is completed, the synchronization control module 50 controls the infrared light source 601 to turn off and the continuous laser 101 to turn on, entering the dual-comb detection mode. At the same time, the switchable objective lens 204 switches to the high numerical aperture focusing objective lens.

[0085] When entering the dual-comb detection mode, first set the oblique incidence angle α and the scanning trajectory parameters (r, φ).

[0086] Specifically, the oblique incident angle α is set (preferably 5°-12°).

[0087] Set the scanning trajectory: Use polar coordinate scanning with an azimuth angle φ step size of 30° (12 directions in total). The radial distance r under each φ has 3 values ​​(0.4R, 0.5R, 0.6R), for a total of 36 scanning points. In addition, the center point of the hole bottom (r=0) is used as the depth reference.

[0088] The scanning sequence is as follows: First, scan each type of first-class scanning point in descending order of radial distance. Then, traverse all azimuth angles at the same radial distance and switch to the next radial distance until all type-1 scanning points are scanned. Finally, scan the type-2 scanning points.

[0089] In step S2, single-point dual-comb multipath reflection measurement is performed based on the set oblique incidence angle α and scanning trajectory parameters (r, φ).

[0090] Specifically, the synchronization control module 50 controls the dichroic scanning mirror 201 to deflect the light beam to a preset tilt angle α and azimuth angle φ. After the dichroic scanning mirror 201 deflects, the actual landing point of the light beam on the aperture plane is laterally offset relative to the target coordinates, and the offset amount is related to α and φ. The data processing module 40 calculates the required lateral translation compensation amounts Δx(α, φ) and Δy(α, φ) based on α, φ and the target coordinates of the current scanning point (x_target = r·cosφ, y_target = r·sinφ). The synchronization control module 50 controls the first Alvarez lens group 202 and the second Alvarez lens group 203 to perform overall lateral translation in the X and Y directions, respectively, to compensate for the lateral offset of the light spot caused by the angle change of the dichroic scanning mirror 201, so that the focused light spot falls accurately at the target coordinates.

[0091] The synchronization control module 50 controls the continuous laser to emit continuous laser light, thereby outputting a P-state signal optical comb and an S-state local oscillator optical comb from the dual optical comb light source module 10. The P-state signal optical comb passes through an optical circulator 205, an optical fiber collimator 206, a dichroic scanning mirror 201, a quarter-wave plate 207, a first Alvarez lens group 202, a second Alvarez lens group 203, and a switchable objective lens 204, before entering the through-hole at an oblique incidence angle α. The beam undergoes multipath reflection within the hole: a first reflection occurs on the sidewall facing the light source, with some diffused light returning upwards; the reflected light is reflected a second time to the sidewall facing away from the light source, also resulting in upward diffuse scattering.

[0092] The returned diffuse light passes through switchable objective lens 204, second Alvarez lens group 203, first Alvarez lens group 202, quarter-wave plate 207, and dichroic scanning mirror 201, becoming an S-state reflection comb. This comb is output through circulator port 3 and enters a 2×2 fiber coupler 301, where it is mixed with the local oscillator comb (S-state local oscillator comb). The balanced detector 302 outputs a beat frequency signal, which is acquired by an ADC and sent to the data processing module 40.

[0093] The data processing module 40 performs a Fourier transform on the beat frequency signal to obtain an A-scan (depth-reflection intensity curve). The multiple reflection peaks in the A-scan correspond to: peak 0 (upper surface of the sample), peak 1 (depth of the first reflection point, i.e., the sidewall of the light-facing side), peak 2 (depth of the second reflection point, i.e., the sidewall of the backlight side), and peak 3 (reflection from the bottom of the hole or the stage).

[0094] Record the coordinates (x, y) of the scan point and the depth values ​​of each reflection peak. For the first and second reflection points, their reflection intensities also need to be recorded (for subsequent point cloud filtering).

[0095] In step S3, multi-point scanning and point cloud generation are performed. According to preset scanning trajectory parameters, for example, the scanning points can be switched sequentially in descending order of radial distance r, and step S2 is repeated to obtain the detection signal of each scanning point.

[0096] The data processing module 40 acquires the corresponding depth information based on the detection signals of each scanning point: Based on the first type of detection signal, it acquires the A-scan of each first type of scanning point; it extracts the reflection peak (peak 0) corresponding to the upper surface of the sample from the A-scan as the zero depth point, and extracts the reflection peak (peak 1) on the light-facing side and the reflection peak (peak 2) on the back-facing side, respectively, as the depth data of the light-facing side and the depth data of the back-facing side. For the second type of scanning point (r=0), the data processing module 40 extracts the reflection peak (peak 3) at the bottom of the hole (or the stage) as the depth data of the bottom of the hole based on the second type of detection signal, so as to obtain the depth reference of the through hole.

[0097] The data processing module 40 generates a 3D point cloud based on the coordinates, zero depth point, and corresponding depth data of the illuminated and shaded sides of each scanning point. The 3D point cloud includes the point cloud of the illuminated side and the point cloud of the shaded side.

[0098] During the point cloud generation process, reflection points with signal strength below a preset threshold are marked as missing and do not participate in subsequent point cloud reconstruction.

[0099] In step S4, the data processing module 40 performs backlight surface geometry reconstruction on the 3D point cloud, specifically including the following steps: (1) Axial virtual slicing: The three-dimensional point cloud is divided into multiple virtual slices S0, S1… S along the depth direction at a preset interval Δh (preferably 0.1μm~10μm). N-1 ; (2) Sunlit surface contour fitting: For each virtual slice layer, extract the radial coordinates of all sunlit surface point clouds within that layer, and perform circle fitting (preferably least squares circle fitting) to obtain the circle center O(x) corresponding to each virtual slice layer. k ,y k ) and radius R k ; (3) Point cloud interpolation for missing backlight areas: For each slice, if the point cloud of the backlight area is missing, based on the circular section assumption (the layer is circular) and the continuity of adjacent slices, spline interpolation or radial basis function interpolation is used to fill in the radial coordinates of the backlight points. Curvature smoothing constraints are introduced in the interpolation process to ensure that the completed area transitions naturally with the existing frontlight data. (4) Global axis fitting: The center O of each virtual slice is fitted. k Three-dimensional straight line fitting is performed to obtain the direction vector v of the central axis of the through hole and the spatial position P0. This axis serves as a global consistency constraint, triggering re-interpolation for layers that deviate too much. (5) Calculation of morphological parameters: Based on the radius and central axis of each virtual slice, calculate the geometric parameters of the through hole, including hole diameter distribution, hole depth, verticality, taper, sidewall smoothness, etc.

[0100] The data processing module 40 automatically outputs the inspection report, which includes a pore size distribution chart, pore depth, verticality, taper, sidewall smoothness score, and a three-dimensional contour rendering.

[0101] Beneficial technical effects of the present invention: (1) Full wall coverage: Through multipath reflection scanning + backlight reconstruction, under the condition that the axial tilt angle of the through hole is ≤5°, the measurement coverage of the backlight area is increased from less than 50% under vertical incidence to more than 95%, thus realizing full wall three-dimensional reconstruction.

[0102] (2) High precision: The absolute ranging accuracy of the dual optical comb reaches the nm-μm level, and the refractive index adaptive compensation further eliminates the cross-medium system deviation, with a depth measurement uncertainty of ≤20nm.

[0103] (3) High efficiency: 36 scanning points per hole, each point takes 10ms to measure. With the addition of the response time of the dichroic scanning mirror and Alvarez, the total measurement time per hole is <500ms, which is dozens of times more efficient than the existing hole-by-hole scanning scheme.

[0104] (4) Strong adaptability: By zooming through the first Alvarez lens group and the second Alvarez lens group (the spot diameter is adjustable from 2.5μm to 5μm), it can detect through holes with an aperture of 5μm to 10μm and an aspect ratio of 20:1.

[0105] (5) Complete output of multiple parameters: output the detection results of aperture distribution, depth, verticality, taper, sidewall smoothness, etc. at one time.

[0106] (6) High system integration: The system adopts fiber optic optical path, dichroic scanning mirror and piezoelectric ceramic to drive the first Alvarez lens group and the second Alvarez lens group. The overall structure is compact and easy to integrate into wafer inspection equipment.

[0107] It should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This way of describing the specification is only for clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

[0108] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A three-dimensional inspection system for through holes, characterized in that, Includes the following modules: A dual optical comb light source module is used to generate a signal optical comb and a local oscillator optical comb with a preset repetition frequency difference; The beam control module is used to focus and scan the signal light comb into the through-hole of the sample at an oblique incidence angle relative to the through-hole axis, so that the signal light comb undergoes multipath reflection within the through-hole, including reflection from the light-facing surface and reflection from the back surface; to achieve polar coordinate multi-point scanning of the through-hole by changing the azimuth angle and radial distance of the signal light comb; and to transmit the reflected signal returned from the through-hole to the main detection channel module; wherein, the azimuth angle is the angle determined on the plane of the through-hole opening with a preset reference direction as the starting direction, and the radial distance is the distance of the incident light spot from the center of the through-hole opening on the plane of the through-hole opening; The main detection channel module is used to receive the reflected signal and perform coherent detection between the reflected signal and the local oscillator optical comb to output a corresponding detection signal; The data processing module is signal-connected to the main detection channel module and is used to generate a three-dimensional point cloud based on the detection signal, and to perform backlight surface geometry reconstruction on the three-dimensional point cloud to obtain the three-dimensional morphology and geometric parameters of the through hole. A synchronization control module is signal-connected to the data processing module and the beam control module to control the timing of each adjustment action of the beam control module according to the control parameters of the data processing module, so that the incident light spots at each scanning point are aligned sequentially in a predetermined order.

2. The system according to claim 1, characterized in that, The beam control module includes an angle adjustment unit and a displacement compensation unit. The angle adjustment unit is used to adjust the oblique incident angle and the azimuth angle of the signal optical comb. The displacement compensation unit is used to adjust the radial distance by compensating for the lateral offset of the signal optical comb. The synchronization control module controls the timing synchronization between the angle adjustment unit and the displacement compensation unit, so that the incident light spots at each scanning point are aligned sequentially in a predetermined order.

3. The system according to claim 2, characterized in that, The angle adjustment unit uses a dichroic scanning mirror, and the displacement compensation unit uses an Alvarez lens group. The Alvarez lens group includes a first Alvarez lens group and a second Alvarez lens group. The first Alvarez lens group and the second Alvarez lens group compensate for the lateral offset of the signal optical comb by their respective overall lateral translation, and adjust the focused spot diameter of the signal optical comb by their respective internal relative displacement.

4. The system according to claim 3, characterized in that, The polar coordinate multi-point scanning includes scanning multiple first-type scanning points and one second-type scanning point; The first type of scanning point is the scanning point where the radial distance is not equal to 0. The main detection channel module receives the multipath reflection signal generated in the through hole from the first type of scanning point, and outputs the first type of detection signal after coherently detecting the multipath reflection signal with the local oscillator comb. The data processing module extracts the depth data of the front surface and the depth data of the back surface according to the first type of detection signal to generate a three-dimensional point cloud. The second type of scanning point is the scanning point where the radial distance is equal to 0. The main detection channel module receives the bottom reflection signal from the second type of scanning point, and outputs the second type of detection signal after coherently detecting the bottom reflection signal with the local oscillator comb. The data processing module extracts the bottom depth data of the hole based on the second type of detection signal to obtain the depth reference of the through hole.

5. The system according to claim 4, characterized in that, The first type of scanning point includes multiple scanning points with different azimuth angles and multiple scanning points with different radial distances; wherein, the multiple scanning points with different azimuth angles are implemented by changing the azimuth angles with a preset step size, and the multiple scanning points with different radial distances are implemented by setting the radial distances to multiple preset ratios of the through hole diameter.

6. The system according to claim 5, characterized in that, The data processing module calculates the target coordinates of each of the first type of scanning points on the plane of the through hole based on the oblique incidence angle, the azimuth angle, and the radial distance; the synchronization control module controls the displacement compensation unit to perform lateral offset compensation on the signal optical comb based on the target coordinates, so that the incident light spot is sequentially aligned with each of the first type of scanning points.

7. The system according to claim 5, characterized in that, The scanning sequence of the polar coordinate multi-point scan is as follows: first, the first type of scanning points are scanned in descending order of radial distance; after traversing all azimuth angles at the same radial distance, the scan is switched to the next radial distance until all first type of scanning points are scanned; finally, the second type of scanning points are scanned.

8. The system according to claim 3, characterized in that, The range of the oblique incidence angle is 5°-12°.

9. The system according to claim 3, characterized in that, The system also includes a preliminary hole positioning module, which is used to establish the global coordinates of all through-hole openings by scanning the entire sample. The preliminary hole positioning module includes: Infrared light source, used to emit infrared light; A beam splitter is disposed on the output light path of the infrared light source; An infrared camera is positioned on the reflected light path of the beam splitter; The infrared light emitted by the infrared light source is transmitted through the beam splitter and then through the dichroic scanning mirror to illuminate the sample surface; the infrared light reflected from the sample surface is transmitted through the dichroic scanning mirror and then reflected by the beam splitter to be received by the infrared camera.

10. The system according to claim 9, characterized in that, The beam control module also includes a switchable objective lens, which is set in the optical path after the Alvarez lens group along the light incident direction, for switching between infrared positioning mode and dual-comb detection mode; The switchable objective lens includes a large field-of-view infrared objective lens and a high numerical aperture focusing objective lens; In the infrared positioning mode, the switchable objective lens is switched to the large field-of-view infrared objective lens, and infrared light is irradiated onto the sample surface through the large field-of-view infrared objective lens to establish the global coordinates of all through holes. In the dual-comb detection mode, the switchable objective lens is switched to the high numerical aperture focusing objective lens, and the signal optical comb is focused and scanned into the through hole by the high numerical aperture focusing objective lens for three-dimensional detection.

11. The system according to claim 4, characterized in that, The data processing module acquires the three-dimensional morphology and geometric parameters of the through hole, specifically including: Based on the first type of detection signal, the depth-reflection intensity curve of each first type of scanning point is obtained; The reflection peak corresponding to the upper surface of the sample is extracted from the depth-reflection intensity curve as the zero depth point, and the reflection peaks of the light-facing side and the back-facing side are extracted as the light-facing side depth data and the back-facing side depth data, respectively. A three-dimensional point cloud is generated based on the coordinates of each scanning point, the zero depth point, and the corresponding depth data of the light-facing side and the light-reflecting side. The backlight surface geometry is reconstructed from the three-dimensional point cloud to output the three-dimensional morphology and geometric parameters of the through hole.

12. The system according to claim 11, characterized in that, The specific steps for reconstructing the backlight surface geometry of the three-dimensional point cloud include: The three-dimensional point cloud is divided into multiple virtual slices according to the depth direction; By performing circle fitting on the point cloud of the sunlit side in each virtual slice, the center and radius of the circle corresponding to each virtual slice are obtained; Based on the assumption of a circular cross section and axial continuity, interpolation is used to complete the missing point cloud on the backlit surface; By fitting a three-dimensional straight line to the center of the circle corresponding to each virtual slice, the central axis of the through hole is obtained; The geometric parameters of the through hole are calculated based on the radius corresponding to each virtual slice and the central axis.

13. The system according to claim 12, characterized in that, The specific steps for interpolating and completing the missing point cloud on the backlit side include: For any depth layer, the radial coordinates are interpolated and supplemented in the missing area on the back side according to the circumferential angle based on the center and radius of the circle fitted on the light-facing side of the layer. The interpolation process introduces curvature smoothing constraints to make the supplemented data transition continuously with the existing data of adjacent depth layers.

14. The system according to claim 3, characterized in that, The beam control module also includes an optical circulator, an optical fiber collimator, and a quarter-wave plate, wherein... The optical circulator is disposed between the dual optical comb light source module and the dichroic scanning mirror, and is used to transmit the signal optical comb as an incident beam to the dichroic scanning mirror, and transmit the reflected beam reflected by the sample and returned by the dichroic scanning mirror to the main detection channel module. The fiber collimator is disposed between the optical circulator and the dichroic scanning mirror, and is used to expand and collimate the incident beam. The quarter-wave plate is positioned between the dichroic scanning mirror and the Alvarez lens group to make the polarization states of the incident beam and the reflected beam orthogonal to each other, so that the incident beam and the reflected beam are separated at the optical circulator.

15. A method for three-dimensional inspection of through holes, characterized in that, Using the through-hole three-dimensional inspection system as described in any one of claims 1-14, the method comprises the following steps: S1: Perform system initialization and parameter settings, setting the oblique incidence angle and scanning trajectory parameters including multiple scanning points; S2: Based on the scanning trajectory parameters, perform single-point dual-comb multipath reflection measurement. Focus the signal comb generated by the dual-comb light source module at the oblique incident angle to the current scanning point in the through hole of the sample, so that the signal comb will generate multipath reflection including light-facing surface reflection and back-facing surface reflection in the through hole. Receive the reflected signal returned from the through hole, and perform coherent detection between the reflected signal and the local oscillator comb generated by the dual-comb light source module, and output the corresponding detection signal. S3: Switch each scanning point sequentially according to the scanning trajectory parameters, repeat step S2 to obtain the detection signal of each scanning point, obtain the corresponding depth information according to the detection signal of each scanning point, and generate a three-dimensional point cloud according to the coordinates of each scanning point and the corresponding depth information. S4: Perform backlight surface geometry reconstruction on the three-dimensional point cloud to obtain the three-dimensional morphology and geometric parameters of the through hole.

Images