Subsurface alignment metrology system for packaging applications

Abstract

An apparatus for detecting metrology data in a semiconductor packaging process, the apparatus determining alignment metrology data for hybrid bonding using a high-speed focus and collection technique, in some embodiments, the apparatus may include a source configured to illuminate a focal point with a wavelength selected from wavelengths longer than 1100 nm, an optical lens that forms the illumination beam when illuminated by the source, an acousto-optic scanner that moves the illumination beam back and forth according to a scan pattern, a splitter that enables the illumination beam to be directed to a metrology sampling location while allowing a reflected beam produced by the illumination beam to pass through the splitter to a detector, a set of optical components configured to focus the illumination beam to one or more focal points in a Z direction to acquire subsurface images, and a substrate platform configured to hold a substrate and move the substrate during scanning.

Description

[Technical Field]

[0001] FIELD OF THE INVENTION Embodiments of the present principles relate generally to semiconductor processing of semiconductor substrates. [Background technology]

[0002] Typically, during semiconductor manufacturing, mask alignment is performed by using a system that steps to a location, acquires data from that location, and then moves to the next location. However, packaging applications may require processing thousands of locations on a single substrate, and the inventors of the present invention have recognized that machine-based step-and-repeat systems are not fast enough to process a large number of locations without significantly impacting throughput.

[0003] Accordingly, the present inventors have provided a method, apparatus and system for alignment metrology that is conducive to packaging applications. Summary of the Invention

[0004] Described herein are methods, apparatus and systems for providing alignment metrology for packaging applications.

[0005] In some embodiments, an apparatus for detecting metrology data may include a source using a laser configured to illuminate a focal point through silicon, the wavelength of the source being selected from wavelengths longer than 1100 nm and configured to produce a diffraction-limited focal point for subsurface imaging; an optical lens configured to form an illumination beam when illuminated by the source; an acousto-optic scanner configured to move the illumination beam back and forth according to a scan pattern; a splitter configured to enable the illumination beam to be directed to a metrology sampling location while allowing a reflected beam produced by the illumination beam to pass through the splitter to a detector; a set of optical components configured to focus the illumination beam to a focal point in a Z direction to acquire a subsurface image; and a substrate platform configured to hold a substrate and move the substrate in an X and Y direction based on the metrology data collection pattern, wherein the apparatus is configured to acquire metrology data for a semiconductor packaging process.

[0006] In some embodiments, a system for correcting packaging alignment errors comprises an apparatus for detecting metrology data, the apparatus comprising: at least one source using a laser configured to illuminate a focal point through silicon, the wavelength of the source configured to produce a diffraction-limited focal point for subsurface imaging; at least one optical lens configured to form at least one illumination beam when illuminated by the at least one source; at least one scanner configured to move the at least one illumination beam back and forth according to a scan pattern; at least one splitter configured to enable directing the at least one illumination beam to a metrology sampling location while allowing at least one reflected beam generated by the at least one illumination beam to pass through the at least one splitter to reach at least one detector; and at least one scanner configured to focus the at least one illumination beam to one or more focal planes in a Z direction to acquire the subsurface image. and a substrate platform configured to hold the substrate and move the substrate in the X and Y directions based on a metrology data collection pattern, wherein the apparatus may include the apparatus configured to acquire metrology data for the semiconductor packaging process, a first controller in communication with the apparatus, the at least one scanner and the at least one set of optical components, the first controller configured to automatically adjust the scan pattern and focus based on a quantity of metrology data for a particular location on the substrate, at least one detector configured to receive the at least one reflected beam and generate a sub-surface image, an alignment correlator in communication with the at least one detector, the alignment correlator configured to determine an alignment error from the sub-surface image from the at least one detector, and a second controller in communication with the hybrid bonder, the second controller configured to adjust the alignment of the chip on the substrate based on the alignment error from the alignment correlator.

[0007] In some embodiments, the apparatus for detecting metrology data includes a source using a laser configured to illuminate a focal point through silicon, the wavelength of the source being selected from wavelengths longer than 1100 nm and configured to generate a diffraction-limited focal point for subsurface imaging; an optical lens configured to form an illumination beam when illuminated by the source; an acousto-optic scanner configured to move the illumination beam back and forth according to a scan pattern; a splitter configured to allow the illumination beam to be directed to a metrology sampling location while allowing a reflected beam generated by the illumination beam to pass through the splitter to a detector; and a splitter configured to steer the illumination beam in a Z direction to obtain the subsurface image. a set of optical components configured to focus the illumination beam to one or more focal points in a direction, the set of optical components including a correction for spherical aberration caused by the illumination beam passing through at least one surface plane; a substrate platform configured to hold the substrate and move the substrate in the X and Y directions based on a metrology data collection pattern; and a controller in communication with the acousto-optic scanner and the set of optical components, the controller configured to automatically adjust the scan pattern and focus based on a quantity of metrology data for a particular position on the substrate, wherein the apparatus is configured to acquire metrology data for a hybrid bonder in a semiconductor packaging process.

[0008] Other and further embodiments are disclosed below.

[0009] Embodiments of the present principles, briefly outlined above and discussed in more detail below, can be understood by reference to exemplary embodiments of the present principles as illustrated in the accompanying drawings. However, the accompanying drawings illustrate only typical embodiments of the present principles and, therefore, should not be considered limiting in scope, as the present principles may embrace other embodiments that are equally effective. [Brief explanation of the drawings]

[0010] [Figure 1] 1 is an isometric view of a chip bonded to a substrate, in accordance with some embodiments of the present principles; [Figure 2] 1A-1C are cross-sectional and top views of the alignment of a chip bonded to a substrate, according to some embodiments of the present principles; [Figure 3] 1A-1C are cross-sectional and top views of the alignment of a chip bonded to a substrate, according to some embodiments of the present principles; [Figure 4] 1 is a cross-sectional view of a measurement system, in accordance with some embodiments of the present principles; [Figure 5] 1A-1C are top views of scan patterns, in accordance with some embodiments of the present principles; [Figure 6] 1 is an isometric view of substrate support motion, in accordance with some embodiments of the present principles; [Figure 7] 1 is a cross-sectional view of a metrology system in communication with a hybrid bonder to enable alignment feedback, in accordance with some embodiments of the present principles; DETAILED DESCRIPTION OF THE INVENTION

[0011] To facilitate understanding, the same reference numerals have been used, where possible, to designate identical elements common to the figures. The figures are not drawn to scale and may be simplified for clarity. Elements and features of one embodiment may be beneficially incorporated in other embodiments without further description.

[0012] The method, apparatus, and system provide a fast, efficient alignment metrology solution for subsurface alignment of targets that meets the high demands found in packaging applications such as, but not limited to, hybrid bonding. Non-mechanical scanning techniques are used to enable faster data collection at each scan location while providing adjustable data collection scans. At critical locations, the metrology scan can be adjusted for higher accuracy (more data), while at less critical areas, it can be retuned to reduce data throughput and overall data processing requirements. Step-and-repeat metrology systems are typically based on mechanical mechanisms used to scan the surface of a substrate and acquire data from the substrate's surface. Acceleration and deceleration of the mechanical mechanisms impact the speed at which the step-and-repeat metrology system can acquire data, increasing overall processing time. The inventors of the present invention have discovered that by eliminating the mechanical aspect and using a non-mechanical scanning device, data collection speed can be significantly enhanced.

[0013] An increasingly pressing issue in hybrid bonding and packaging applications is the need to ensure precise alignment of the bond pads of the two objects to be connected. In absolute terms, this requirement is not as stringent as that required in typical IC manufacturing, but it nonetheless poses a significant problem because the bond pads are below the top silicon chip and therefore not visible in the visible light range. Infrared is used to enable the metrology station to see through the chip and below, revealing alignment marks on the substrate to which the chip is bonded. To achieve reasonable throughput, alignment metrology for packaging applications requires high-speed techniques approaching thousands of measurements per hour.

[0014] Conventional techniques for performing subsurface imaging and alignment tasks have been implemented using high-resolution short-wave IR (SWIR) microscopes in "step-and-repeat" imaging systems. As the number of inspection sites increases, the step-and-repeat approach becomes unacceptably slow. The present inventors have found that no suitable SWIR camera exists that can operate in smooth, continuous motion to achieve the required throughput needed in packaging applications. SWIR line-scan cameras exist. However, the present inventors have found that such sensors require focused line illumination, which requires laser radiation, and that such illumination cannot be used directly because it produces unacceptable levels of speckle in the image. Furthermore, the present inventors have found that attempts to "speckle-bust" laser radiation result in output light that is not tightly focused, thus negating the possibility of high-speed, high-resolution imaging.

[0015] The present inventors have discovered a solution to a challenging problem by using a SWIR spot scanning device in high-resolution, reflectance mode to generate subsurface images of targets (e.g., fiducials). Difficulties associated with suitable light sources or detectors exist that hinder the operation of such systems in field or line imaging modes in the SWIR regime. The present inventors have found that these difficulties for packaging applications can be overcome when used with a sufficiently powerful single-point detector and a laser source operating at wavelengths above about 1100 nm and capable of generating a small, diffraction-limited focal spot. By incorporating a non-mechanical scanning device, such as, but not limited to, an acousto-optic scanner operating in the near-IR, the speed of scanning can be dramatically increased.

[0016] The packaging process's need for subsurface imaging presents unique challenges compared to other metrology data collection systems. FIG. 1 illustrates an isometric view 100 of a first chip 102 bonded to a substrate 118, according to some embodiments. The first chip 102 typically originates from a component substrate (not shown). During hybrid bonding, the first chip 102 is removed from the component substrate, picked up, and flipped 108 upside down so that the top surface 104 of the first chip 102, bearing a first alignment mark 120 (or fiducial), becomes the bottom or chip-bonding surface, which is then bonded to the top surface 106 or substrate-bonding surface of the substrate 118. The top surface 106 of the substrate 118 typically bears a second alignment mark 122 that is used to align with the first alignment mark 120 of the first chip 102 during bonding 110. FIG. 2 shows a cross-sectional view 200B and a top view 200A of aligned fiducial marks, as well as a top view 200C of a first chip 102 bonded to a substrate 118, according to some embodiments. For the example shown in FIGS. 1 and 2 , and not intended to be limiting, the first alignment mark 120 on the first chip 102 is a circle, and the second alignment mark 122 on the substrate 118 is a ring. For this example, proper alignment is when the circle of the first alignment mark 120 is at the center of the ring of the second alignment mark 122 on the substrate 118, as shown in top views 200A and 200C. When the first chip 102 is bonded to the substrate 118, the first alignment mark 120 and the second alignment mark 122 are essentially coplanar along plane 128, as shown in cross-sectional view 200B.

[0017] To determine whether proper alignment between the first chip 102 and the substrate 118 has been achieved after bonding, the first illumination beam 124 is set to penetrate the first chip 102 to a depth 130 near the top surface 106 of the substrate 118 at a plane 128 and the bonding surface of the first chip 102. A detector (not shown, see FIG. 4 ) then captures the first reflected beam 126 and processes a set of image data from the first reflected beam 126. This detector or an associated image analyzer (alignment correlator 446 in FIG. 4 ) compares the image data of the first alignment mark 120 on the first chip 102 with the second alignment mark 122 on the substrate 118 to determine whether an alignment error has occurred. This image data may be compared to historical image / alignment error data to determine the total alignment error, including rotational error (angular alignment error caused by rotational skewing of the bonded chip relative to the substrate; see, for example, diagram 300C in FIG. 3 ).

[0018] Some packaging processes have stacked chips that require multi-planar subsurface imaging, which presents unique challenges. For example, the system should be able to image at multiple depths or focal points as needed. In some embodiments, as shown in FIG. 2 , an optional second chip 202 may be bonded to a first chip 102. A fourth alignment mark 222 on the optional second chip 202 is used to align with a third alignment mark 220 on the first chip 102 during bonding 110. For the example shown in FIG. 2 , and not intended to be limiting, the third alignment mark 220 on the optional second chip 202 is a circle, and the fourth alignment mark 222 on the first chip 102 is a ring. For this example, proper alignment occurs when the circle of the third alignment mark 220 is centered within the ring of the fourth alignment mark 222 on the first chip, as shown in top view 200C, as well as in top view 200A. When the optional second chip 202 is bonded to the first chip 102, the third alignment mark 220 and the fourth alignment mark 222 are essentially coplanar at the interface 228 of the bonding surfaces of the chips, as shown in cross-sectional view 200B.

[0019] To determine whether proper alignment between the first chip 102 and the second chip 202 has been achieved after bonding, the second illumination beam 224 is set to penetrate the optional second chip 202 to the depth of the chip interface 228. A detector (not shown, see FIG. 4 ) then captures the second reflected beam 226, which processes a set of image data from the second reflected beam 226. This detector or an associated image analyzer (alignment correlator 446 in FIG. 4 ) compares the image data of the third alignment mark 220 of the optional second chip 202 with the fourth alignment mark 222 of the first chip 102 to determine whether an alignment error has occurred. This image data may be compared to historical image / alignment error data to determine the total alignment error, including rotational error (angular alignment error caused by rotational skewing of the bonded chip relative to the substrate, see, e.g., view 300C in FIG. 3 ).

[0020] In more complex packaging processes, bonded single chips and bonded stacked chips may require multi-planar subsurface imaging to acquire fiducial alignment images when the fiducial alignment pairs are located in different planes. In cross-sectional view 300A of FIG. 3 , a first chip 304 is bonded to a substrate 302. The substrate 302 has a first set of fiducials or alignment marks 308 on a first surface 310. The first chip 304 has a second set of alignment marks 312 on a second surface 314 that are intended to align the first chip 304 with the substrate 302. To determine whether proper alignment between the first chip 304 and the substrate 302 is achieved after bonding, a first illumination beam 316 is set at a first position 322 on the substrate 302 to penetrate the first chip 304 to a first depth 318 near the first surface 310 of the substrate 302. The first reflected beam (not shown, see FIG. 4) is then captured by a detector (not shown, see FIG. 4), which processes a first set of image data from the first reflected beam. The second illumination beam 320 is adjusted to reflect off the second surface 314 and receive the second reflected beam (not shown) by the detector, which processes a second set of image data. The detector or an associated image analyzer (alignment correlator 446, FIG. 4) compares the sets of image data and determines whether an alignment error has occurred. The process then proceeds to a second location 324 on the substrate 302 and is repeated. The image data from the second location 324 may be compared with the image / alignment error data from the first location 322 to determine the total alignment error, including rotational error (angular alignment error caused by rotational skewing of the bonded chips).

[0021] 3, a second chip 306 is bonded to a first chip 304. The second chip 306 has a third set of alignment marks 326 on a third surface 330 of the second chip 306 that are intended to align the second chip 306 with the first chip 304. The third set of alignment marks 326 are aligned with a fourth set of alignment marks 328 on the second surface 314 of the first chip 304 at a third position 340, and are also aligned with one or more of the marks in the second set of alignment marks 312 on the second surface 314 of the first chip 304 at the first position 322. When alignment overlap occurs in multiple bonded stacks, previous image data can be used to compare with newly collected image data at different depths. Assuming that the first illumination beam 316 generated image data for the first set of alignment marks 308 at the first position 322, and the second illumination beam 320 generated image data for the second set of alignment marks 312 at the first position 322, then the third illumination beam 332 can be used to generate image data for the third set of alignment marks 326 at the first position 322. For the first position 322, image data from each of the surfaces of the substrate 302, the first chip 304, and the second chip 306 can be compared, without repeating the illumination beam at a different depth level (focus). For the third position 340, the illumination beam can be repeated at depths of the second surface 314 and the third surface 330 for the third position 340 to obtain alignment error data for the third set of alignment marks 326 relative to the fourth set of alignment marks 328 at the third position 340.

[0022] The top view 300C of FIG. 3 shows a substrate 302 with a first chip 304 bonded to it and a second chip 306 bonded to it. A third set of alignment marks 326 is visible on a third surface 330 of the second chip 306. A portion of a second set of alignment marks 312 is visible on a second surface 314 of the first chip 304. As an example, a desired orientation line 334 of the second chip 306 is used to compare with an actual (skewed) orientation line 336 having an error angle 338. The error angle 338 indicates a positive or negative rotation from the desired orientation. Therefore, both alignment and orientation errors can be determined from image data acquired from the alignment marks. The illumination source must be fast-focusing and adjustable to penetrate silicon materials and allow for multiple focal spots at different depths. In some situations, thousands of locations on the substrate 302 must be scanned in time. Despite the challenges presented, the present inventors have overcome these obstacles to create a high-speed scanning device with rapid focusing characteristics that allows for rapid and efficient acquisition of metrology data, such as bond alignment data, without slowing down the packaging process or significantly impacting yield. In addition, the device also accounts for inherent issues, such as optical aberrations (e.g., spherical aberration) that arise from illuminating a surface through a surface.

[0023] FIG. 4 shows a diagram of a metrology system 400 comprising an image detection device 442 including an illumination source 402 (e.g., an IR laser source), beam-forming optics 404 for shaping an illumination beam 428, a scanner 406 (e.g., an acousto-optic scanner, a polygon scanner, a galvanometer scanner, etc.), a focusing objective lens 410 for positioning and focusing the illumination beam 428 at a location on a substrate 412, and a detector 440 for generating a high-resolution subsurface image of the sample through the substrate 412 and / or silicon material, such as a chip, bonded to the substrate 412. In some embodiments, an acousto-optic scanner is used to enhance the speed and control of scanning without mechanical devices. The acousto-optic scanner may be implemented for maximum scanning speed in single or dual chirp modes. When maximum scanning speed is not required, some embodiments may use a mechanically based scanner, such as, but not limited to, a galvanometer scanner or a polygon scanner. For example, a mechanically based scanner may be used in conjunction with a metrology system having multiple illumination sources, etc. An image is generated by focusing an illumination beam 428 at a location 430 on a surface 432 of the substrate 412 or on the surface of a chip (not shown, see Figures 2 and 3) coupled to the substrate. In some embodiments, additional relay optics and / or a magnification device 436 may be used after the scanner 406 and before the splitter 408 (illumination beam splitter or reflective beam splitter).

[0024] The scanner 406 moves the illumination beam 428 in a fast scan in one direction, while the actuator 416 slowly moves the substrate support 414 in a raster fashion in a direction perpendicular to the fast scan. The metrology system 400 also includes a substrate motion device 444 to assist in scanning the substrate / chip. For example, as shown in diagram 500 of FIG. 5, the illumination beam 428 moves slowly in the X direction, while the fast scan 502 scans in a direction perpendicular to the X direction. The interface between two surfaces of the bonded piece is scanned in swath 508. Multiple swaths are performed, starting at the top and ending at the bottom of the image sample position. At the end of swath 508, the sample position is advanced in the Y direction 506 a distance 510 that is smaller than the width 512 of the swath 508 (for clarity, the actual width of the swath is slightly reduced in FIG. 5 because the orthogonal scan arrows for each swath would overlap each other, resulting in a loss of detail). This process is repeated by scanning the sample in the opposite direction, and so on. Thus, to ensure satisfactory imaging of the entire substrate 412, there is a few pixels of overlap 514 between any two adjacent swaths. In some embodiments, as shown in the isometric view 600 of FIG. 6, an actuator 416 (see FIG. 4) of the substrate support 414 slowly moves the substrate 412 in the X direction 504 so that the illumination beam 428 forms a swath 508 at a desired location on the substrate 412 (or a chip coupled to the substrate 412). When the swath 508 is completed, the actuator 416 moves the substrate support 414 in the Y direction 506 a distance less than the orthogonal scan width (width 512). The actuator 416 then moves in the opposite X direction, resulting in another swath.

[0025] During illumination of the target location by illumination beam 428, a reflected beam 434 is generated that is used to form image data. The reflected beam 434 is redirected by splitter 408 to allow it to be received by detector 440, which receives the reflected beam 434 and generates an image. Detector 440 may operate in conjunction with storage device 438, which stores the image in memory and can recall the image for later alignment comparison, and / or in conjunction with alignment correlator 446, which receives multiple images from the detector and uses an image overlay process to determine alignment error positions and values ​​(e.g., alignment shift, alignment rotation error, etc.). In some embodiments, to make the alignment statistics independent of the precise position of any target, detector 440 creates an image by imaging more samples than the Nyquist number according to the point spread function (PSF). Thus, when assembled, the image of the target is identical (in terms of the information it carries) to any non-scanned image acquired (such as an image taken with a digital camera). Thus, the algorithmic strategies used to perform measurements on the images (such as by alignment correlator 446) can use similar algorithmic strategies used for static images.

[0026] The illumination source used with the metrology system 400 should have a relatively narrow bandwidth for point scanning. Broadband light sources, such as those used with step-and-repeat metrology systems, are not compatible with the present metrology system. Broadband light sources distribute light energy over a "broad band" of light and do not produce enough light at any frequency to enable the efficient, intense light beam required for the present metrology system. In addition, broadband light sources tend to shift lateral resolution toward longer wavelengths. The inventors of the present invention have found that the selected light source should have a wavelength selected from a wavelength range longer than 1100 nm that has sufficient optical penetration into silicon (or other materials used in the chip or substrate, etc.) to produce easily detectable reflected light at the desired focal plane. A crucial factor is the amount of absorption of the wavelength in the material (e.g., the chip or substrate material), which is also affected by the thickness of the material. For example, shorter wavelengths may be used for thinner materials than for thicker materials of the same material. For example, if silicon is the primary material being bonded, a wavelength (e.g., an IR wavelength) that is not significantly absorbed by silicon should be selected. The inventors of the present invention have discovered that light sources including light emitting diodes or lasers or the like produce narrow band beams that are powerful and efficient enough to penetrate materials used in the semiconductor manufacturing arena.

[0027] In some embodiments, multiple illumination sources may be used simultaneously to increase throughput by using different wavelengths focused at different levels or planes. For example, FIG. 4 shows an optional second illumination source 450 in the metrology system 400. The second illumination source 450 generates a second illumination beam 452 that travels a similar path to the illumination beam 428 generated by the illumination source 402, except that it is focused at a different plane 460, such as the stack tip 462. When used with multiple illumination sources, the scanner 406 is typically a mechanically based scanner (e.g., a galvanometer scanner or a polygon scanner). Acousto-optic scanners are optimized for a single wavelength, and shifting the angle of the acousto-optic scanner to match the second wavelength requires time, which impacts throughput. When using an acousto-optic scanner, two different wavelengths also produce different scan speeds. Shifting from the Bragg angle for a particular frequency also reduces light reflection back to the detector. To avoid this angle adjustment problem, multiple acousto-optic scanners are required for the multiple illumination sources, increasing the cost and complexity of the metrology system 400. The second illumination beam 452 produces a second reflected beam 454, which is returned by a second splitter 456 to a second detector 458. Implementations of the metrology system 400 may use any number of multiple sets of illumination sources, detectors, splitters, sets of optics, or scanners.

[0028] The data stream acquired by the metrology system 400 may be analyzed in two stages. In the first stage, the raw data (i.e., the stream of photon counts coming from the detector 440) is reconstructed into an image and analyzed to extract overlay parameters. Standard machine vision fiducial recognition algorithms (e.g., image registration and template matching) can be used. The second analysis stage concerns how the extracted overlay measurements relate to various process tool parameters. For example, suboptimal parameter tuning in an upstream step may manifest as greater deviation of the coupon toward the wafer edge. As another example, position-dependent deviations may be correlated to the pressure profile used during the bonding step of a bonding process. Machine learning can also be used in the context of integrated packaging tools with on-board metrology capabilities capable of establishing these types of relationships.

[0029] In some embodiments, a first controller 420 may be used to enable data collection and feedback from each device in the measurement system 400 to optimize the performance of the measurement system 400 and the control of system devices (e.g., scan pattern, scan control, scan position, etc.). The first controller 420 generally includes a central processing unit (CPU) 422, memory 424, and support circuits 426. The CPU 422 may be any form of general-purpose computer processor that can be used in an industrial environment. The support circuits 426 are conventionally coupled to the CPU 422 and may include cache, clock circuits, input / output subsystems, power supplies, etc. The memory 424 may store software routines, such as the methods for controlling the measurement system 400 described above, which, when executed by the CPU 422, may transform the CPU 422 into a special-purpose computer (the first controller 420). The software routines may also be stored and / or executed by a second controller (not shown) located remotely from the measurement system 400.

[0030] In some embodiments, the first controller 420 may communicate with the illumination source 402 to change the illumination wavelength and / or power, etc., the beam-forming optics 404 to shape the illumination beam, the scanner 406 to increase or decrease the scan speed and / or scan width, the focusing objective lens 410 to position and focus the illumination beam to a location on the substrate / chip, and the detector 440 to generate high-resolution images by changing the algorithm type, etc. For hybrid bonding applications, precise positioning of pixels in the acquired images is very important. The first controller 420 can control the sweep of the illumination beam by the scanner 406 to obtain precise positioning. The first controller 420 can also communicate with the storage device 438 and / or the alignment correlator 446 to further enhance the metrology data collection process. The first controller 420 can also communicate with the actuator 416 of the substrate support 414 to enable movement in the X and Y directions to generate the scanned swaths discussed above. The first controller 420 may also change the scan position on the substrate / chip as needed by commanding the actuator 416 to move to a different position. The first controller 420 may also coordinately communicate with the scanner 406 and the actuator 416 to enable a desired scan pattern, etc. Those skilled in the art will appreciate that other ancillary devices (e.g., additional optics, power supplies, etc.) may be used with the metrology system 400, and the first controller 420 may also communicate with the ancillary devices.

[0031] The metrology system 400 has been described in terms of a single scanning spot architecture. However, a multi-spot configuration can also be used when a sample (e.g., a substrate, a chip, etc.) is interrogated simultaneously by many spots. The multi-spot configuration may use a chirped acoustic signal train in a long scan, or may use many short-span scanning spots enabled by diffractive optical elements. In such a system, a detector (e.g., detector 440) is used for each scanning spot. Although the electronic and computational requirements are more complex, the system throughput increases linearly with the number of scanning spots.

[0032] To improve the hybrid bonding alignment process, the metrology system 400 may be used in conjunction with a hybrid bonder 702, as shown in diagram 700 of FIG. 7 . The hybrid bonder 702 is typically controlled by a third controller 720. To optimize the performance of the hybrid bonder 702, the third controller 720 may be used to enable data collection and feedback / feedforward from the first controller 420 of the metrology system 400. The third controller 720 generally includes a central processing unit (CPU) 722, memory 724, and support circuits 726. The CPU 722 may be any form of general-purpose computer processor that can be used in an industrial environment. The support circuits 726 are conventionally coupled to the CPU 722 and may include cache, clock circuits, input / output subsystems, power supplies, etc. The memory 724 may store software routines, such as methods for controlling the alignment of bonded substrates and / or chips, which, when executed by the CPU 722, may transform the CPU 722 into a special-purpose computer (the third controller 720). The software routines may also be stored and / or executed by a fourth controller (not shown) located remotely from the hybrid bonder 702 .

[0033] In some embodiments, the metrology system 400 may communicate directly with the third controller 720 to relay alignment information to the hybrid bonder 702 so that subsequent bonding processes have improved alignment errors. Feedback / feedforward from the metrology system 400 may be performed in real time or at scheduled intervals. In some embodiments, the alignment correlator 446 of the metrology system 400 may transmit actual alignment error data and values, such as, but not limited to, alignment offsets for each fiducial, and / or angular rotation error values ​​for a given type of chip, and / or the position of one or more chips. To facilitate enhanced value of the alignment information, the metrology system 400 may incorporate machine learning to augment image data processing in, for example, but not limited to, the detector 440, the alignment correlator 446, and / or the first controller 420. In some embodiments, the alignment image data processor 704 may be used in conjunction with machine learning to interpret alignment information and / or data information from the metrology system 400 to correlate alignment errors with necessary changes in the hybrid bonder 702 to reduce or eliminate subsequent bond alignment errors. Thus, for example, but not limited to, the machine learning may consider alignment shift data, alignment rotation errors, etc., to determine, for example, but not limited to, the ability of the hybrid bonder 702 to reduce / eliminate errors. For example, the machine learning may consider the applied pressure of the hybrid bonder, the amount of vacuum used to pick up the chip, and / or the tolerances of the mechanical device used to place the chip on the substrate, etc., to reduce alignment errors due to the hybrid bonder.

[0034] Embodiments according to the present principles may be implemented in hardware, firmware, software, or any combination thereof. Embodiments may also be implemented as instructions stored using one or more computer-readable media, which may be read and executed by one or more processors. A computer-readable medium may include a mechanism for storing or transmitting information in a form readable by a machine (e.g., a computing platform or a “virtual machine” running on one or more computing platforms). For example, a computer-readable medium may include any suitable form of volatile or non-volatile memory. In some embodiments, a computer-readable medium may include a non-transitory computer-readable medium.

[0035] While the forgoing is directed to embodiments of the present principles, other and further embodiments of the present principles may be devised without departing from the basic scope thereof.

Claims

1. 1. An apparatus for detecting metrology data, comprising: a source using a laser configured to illuminate a focal spot through silicon, the wavelength of the source being selected from wavelengths longer than 1100 nm and configured to create a diffraction-limited focal spot for subsurface imaging; an optical lens configured to form an illumination beam when illuminated by the source; an acousto-optic scanner configured to move the illumination beam back and forth according to a scan pattern; a splitter configured to allow the illumination beam to be directed to a metrology sampling location while allowing a reflected beam generated by the illumination beam to pass through the splitter and reach a detector; a set of optics configured to focus the illumination beam to a focal point in a Z direction to acquire a subsurface image; a substrate platform configured to hold a substrate and move the substrate in an X and Y direction based on a metrology data collection pattern; Equipped with the apparatus is configured to acquire metrology data for a semiconductor packaging process; Device.

2. The apparatus of claim 1 , wherein the set of optical components is further configured to focus at two or more focal points in the Z direction to acquire subsurface images at multiple focal planes.

3. 2. The apparatus of claim 1, wherein the substrate platform is configured to move in an X direction to form a scanning swath while the acousto-optic scanner reciprocates the illumination beam, and is configured to move in the Y direction when the scanning swath is completed.

4. 10. The apparatus of claim 1, wherein the apparatus comprises an illumination beam splitter and a plurality of sets of acousto-optic scanners, splitters, and sets of optical components configured to acquire metrology data from a plurality of locations on the substrate simultaneously.

5. a controller in communication with the acousto-optic scanner and the set of optics, the controller configured to automatically adjust the scan pattern and focus based on an amount of metrology data for a particular location on the substrate. The apparatus of claim 1 further comprising:

6. The apparatus of claim 1 , wherein the semiconductor packaging process is a hybrid bonder.

7. The apparatus of claim 1 , wherein the set of optical components includes a correction for spherical aberration caused by the illumination beam transmitted through at least one surface plane.

8. The apparatus of claim 1 , wherein the acousto-optic scanner is configured to operate in a dual-chirp mode.

9. 1. A system for correcting packaging alignment errors, comprising: a device for detecting measurement data, at least one source using a laser configured to illuminate a focal point through silicon, the wavelength of the source configured to produce a diffraction-limited focal point for subsurface imaging; at least one optical lens configured to form at least one illumination beam when illuminated by the at least one source; at least one scanner configured to move the at least one illumination beam back and forth according to a scan pattern; at least one splitter configured to allow directing the at least one illumination beam to a metrology sampling location while allowing at least one reflected beam generated by the at least one illumination beam to pass through the at least one splitter and reach at least one detector; at least one set of optical components configured to focus the at least one illumination beam onto one or more focal planes in a Z direction to acquire a subsurface image; a substrate platform configured to hold a substrate and move the substrate in an X and Y direction based on a metrology data collection pattern; Including, the apparatus is configured to acquire metrology data for a semiconductor packaging process; The device, a first controller in communication with the at least one scanner and the at least one set of optical components, the first controller configured to automatically adjust the scan pattern and focus based on an amount of metrology data for a particular location on the substrate; the at least one detector configured to receive the at least one reflected beam and generate a subsurface image; an alignment correlator in communication with the at least one detector, the alignment correlator configured to determine an alignment error from the subsurface image from the at least one detector; a second controller in communication with the hybrid bonder, the second controller configured to adjust alignment of the chip on the substrate based on the alignment error from the alignment correlator; A system comprising:

10. The system of claim 9 , wherein the second controller is configured to adjust a pressure of the hybrid bonder based on the alignment error.

11. The system of claim 9 , wherein the alignment correlator is configured to determine the alignment error using machine learning.

12. The system of claim 9 , wherein the second controller is configured to use machine learning to determine adjustments to the hybrid bonder based on the alignment error.

13. 10. The apparatus of claim 9, wherein the wavelength of the source is selected from wavelengths greater than 1100 nm.

14. 10. The apparatus of claim 9, wherein the substrate platform is configured to move in an X direction to form a scanned swath while the scanner reciprocates the illumination beam, and is configured to move in the Y direction when a scanned swath is completed.

15. 10. The apparatus of claim 9, wherein the apparatus comprises at least one illumination beam splitter and a plurality of sets of scanners, splitters, and sets of optical components configured to acquire metrology data from a plurality of locations on the substrate simultaneously.

16. 10. The apparatus of claim 9, wherein at least one set of optical components includes a correction for spherical aberration caused by the at least one illumination beam transmitted through at least one surface plane.

17. The apparatus of claim 9 , wherein the at least one scanner is an acousto-optic scanner configured to operate in a dual-chirp mode.

18. 1. An apparatus for detecting metrology data, comprising: a source using a laser configured to illuminate a focal spot through silicon, the wavelength of the source being selected from wavelengths longer than 1100 nm and configured to create a diffraction-limited focal spot for subsurface imaging; an optical lens configured to form an illumination beam when illuminated by the source; an acousto-optic scanner configured to move the illumination beam back and forth according to a scan pattern; a splitter configured to allow the illumination beam to be directed to a metrology sampling location while allowing a reflected beam generated by the illumination beam to pass through the splitter and reach a detector; a set of optical components configured to focus the illumination beam to one or more focal points in a Z direction to acquire a subsurface image, the set of optical components including a correction for spherical aberration caused by the illumination beam passing through at least one surface plane; and a substrate platform configured to hold a substrate and move the substrate in an X and Y direction based on a metrology data collection pattern; a controller in communication with the acousto-optic scanner and the set of optics, the controller configured to automatically adjust the scan pattern and focus based on an amount of metrology data for a particular location on the substrate; Equipped with the apparatus is configured to acquire metrology data for a hybrid bonder in a semiconductor packaging process; Device.

19. 20. The apparatus of claim 18, wherein the substrate platform is configured to move in an X direction to form a scanned swath while the acousto-optic scanner reciprocates the illumination beam, and is configured to move in the Y direction when a scanned swath is completed.

20. 20. The apparatus of claim 18, wherein the apparatus is configured to include an illumination beam splitter and multiple sets of acousto-optic scanners, splitters, and sets of optical components configured to acquire metrology data from multiple locations on the substrate simultaneously, or the acousto-optic scanner is configured to operate in a dual-chirp mode.

Images