System, device, and method for compensating for shape variation when aligning bond pads using image-based alignment

Abstract

To provide a method and a system for precisely aligning chiplets with bonding sites.SOLUTION: A bonding system for aligning chiplets 22 to a substrate 25 includes acquiring an image of an area on the chiplet with an alignment microscope 145 or an upward microscope 141, measuring a deviation from planarity of the area on the chiplet, estimating a chiplet position error of the chiplet on the basis of at least the image of the area and the deviation from planarity of the area, and adjusting a relative position of the chiplet to the substrate based on the chiplet position error.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] This application relates generally to aligning chiplets with bonding sites. [Background technology]

[0002] The manufacture of devices (e.g., electronic devices) containing multiple components requires that these multiple components be bonded together. One method of bonding these components involves aligning a first set of electrical pads (interconnect contacts) on a first product (e.g., a chiplet) with a second set of electrical pads on a second product (e.g., a substrate). Accordingly, systems and methods for performing this alignment are designed to operate with an accuracy that is a fraction (1 / 10 to 1 / 4) of the diameter or narrowest width of the electrical pads. Summary of the Invention [Means for solving the problem]

[0003] Some embodiments of a method include obtaining an image of an area on a chiplet, measuring a deviation from planarity of the area on the chiplet, estimating a chiplet position error of the chiplet based on at least the image of the area and the deviation from planarity of the area, and adjusting a relative position of the chiplet and a substrate based on the chiplet position error.

[0004] Some embodiments of a system include one or more processors and one or more memories configured to: acquire an image of an area on a chiplet, measure a deviation from planarity of the area on the chiplet, calculate a chiplet position error for the chiplet based on at least the image of the area and the deviation from planarity of the area, and adjust a relative position of the chiplet and a substrate based on the chiplet position error.

[0005] In some embodiments of one or more computer-readable storage media storing instructions for execution by one or more computing devices, the operations to be performed by the one or more computing devices include acquiring an image of an area on a chiplet; measuring a surface relief of the area on the chiplet; calculating a chiplet position error of the chiplet based on at least the image of the area, the surface relief of the area, and a reference surface; and calculating a position adjustment value for the chiplet based on the chiplet position error. [Brief explanation of the drawings]

[0006] [Figure 1] FIG. 1 is a diagram of an exemplary embodiment of a bonding system configured to bond chiplets to a substrate. [Figure 2A] FIG. 2A illustrates an exemplary embodiment of a surface profile of a chiplet. [Figure 2B] FIG. 2B illustrates an exemplary embodiment of a surface profile of a chiplet. [Figure 2C] FIG. 2C illustrates an exemplary embodiment of a surface profile of a chiplet. [Figure 2D] FIG. 2D illustrates an exemplary embodiment of a surface profile of a chiplet. [Figure 3] FIG. 3 shows an example image of a chiplet taken by an alignment microscope. [Figure 4] FIG. 4 illustrates an exemplary embodiment of an operational flow for bonding a chiplet to a substrate. [Figure 5A] FIG. 5A shows an exemplary embodiment of a test area on a test chiplet. [Figure 5B] FIG. 5B shows an exemplary embodiment of a test area on a test chiplet. [Figure 6A] FIG. 6A shows an example of the surface profile of a reference chiplet and a test chiplet. [Figure 6B]FIG. 6B shows an example of the surface profile of a reference chiplet and a test chiplet. [Figure 7] FIG. 7 is a diagram of an exemplary embodiment of a bonding system configured to bond chiplets to a substrate. [Figure 8] FIG. 8 shows an exemplary embodiment of two upward-facing microscopes. [Figure 9] FIG. 9 illustrates an exemplary embodiment of an operational flow for bonding a chiplet to a substrate. [Figure 10] FIG. 10 illustrates an exemplary embodiment of an operational flow for bonding a chiplet to a substrate. [Figure 11] FIG. 11 illustrates an exemplary embodiment of an operational flow for bonding a chiplet to a substrate. [Figure 12] FIG. 12 illustrates an exemplary embodiment of an operational flow for bonding a chiplet to a substrate. [Figure 13] FIG. 13 illustrates an exemplary embodiment of an operational flow for bonding a chiplet to a substrate. [Figure 14] FIG. 14 illustrates an exemplary embodiment of an operational flow for bonding a chiplet to a substrate. [Figure 15] FIG. 15 is a schematic diagram of an exemplary embodiment of a control device. DETAILED DESCRIPTION OF THE INVENTION

[0007] The following paragraphs describe certain illustrative embodiments. Other embodiments may include alternatives, equivalents, and modifications. Also, the illustrative embodiments may include some novel features, and certain features may not be essential to some embodiments of the devices, systems, and methods described herein. Furthermore, some embodiments include features from two or more of the following illustrative embodiments. As such, features from various embodiments may be combined or substituted as appropriate.

[0008] Also, as used in this specification, the conjunction "or" generally refers to an inclusive "or," but also refers to an exclusive "or" when "or" is explicitly indicated or when the context indicates that the "or" must be an exclusive "or."

[0009] Furthermore, as used herein, terms such as "first," "second," "third," etc. do not necessarily denote any order, sequence, or priority, and may be used to more clearly distinguish one member, operation, element, group, collection, set, region, section, etc. from another without denoting any order, sequence, or priority. As such, a first member, operation, element, group, collection, set, region, section, etc. described below could be referred to as a second member, operation, element, group, collection, set, region, section, etc. without departing from the teachings herein.

[0010] Also, in the following description and drawings, like reference numerals designate the same or corresponding parts throughout the several views.

[0011] FIG. 1 is a diagram of an exemplary embodiment of a bonding system 100 configured to bond chiplets to a substrate. As shown in FIG. 1, bonding system 100 includes a chiplet source unit 110, a transfer-activation unit 120, and a chiplet bonding unit 130. Chiplet source unit 110 is the portion of bonding system 100 that stores or provides source chiplets 22 for use in the bonding process. Transfer-activation unit 120 activates source chiplets 22 so that they are ready for bonding, and transfers source chiplets 22 from chiplet source unit 110 to chiplet bonding unit 130. In some embodiments, one or both of chiplet source unit 110 and transfer-activation unit 120 are separate devices. Also, in some embodiments, transfer-activation unit 120 activates source chiplets 22 before they are placed in chiplet source unit 110. The chiplet bonding section 130 receives the chiplets 22 provided by the transfer-activation section 120 from the chiplet source section 110, and the chiplet bonding section 130 bonds the chiplets 22 to a product substrate 29. The product substrate 29 may include chiplets 22 stacked on top of each other. Also, in the following description, bonding a chiplet 22 to a product substrate 29 may include bonding a chiplet 22 to another chiplet that was previously bonded to the product substrate 29.

[0012] As used herein, a chiplet 22 is an integrated circuit, also referred to as a microchip, computer chip, etc. A chiplet 22 is also a component that includes a chiplet set of interconnect contacts. A chiplet 22 may be defined as a small block of semiconductor material on which a given functional circuit is fabricated. In the context of a substrate (e.g., a wafer) divided into individual chiplets 22, each chiplet 22 can be referred to as a die. A chiplet 22 typically carries a set of integrated electronic components and circuits formed thereon by patterning, coating, etching, doping, plating, singulation, etc. A chiplet 22 typically has electrical functionality such as memory, logic, field programmable gate arrays (FPGAs), accelerator circuits, application specific integrated circuits (ASICs), security coprocessors, graphics processing units (GPUs), machine learning circuits, specialized processors, controllers, devices, electrical circuits, arrays of passive elements, etc. Chiplets 22 may be or include microelectromechanical systems (MEMS) devices, optical devices, electro-optical devices, microfluidic devices, piezoelectric devices, thermoelectric devices, spintronic devices, superconducting devices, and the like.

[0013] In some embodiments, chiplets 22 have small geometric shapes (e.g., rectangular or other polygonal), and chiplets 22 may have planar dimensions that are between 0.5 mm and 30 mm, and may have a thickness of less than 1 mm (e.g., 0.8 to 0.01 mm). For example, chiplets 22 may have widths and heights on the order of 0.5 mm to 15 mm, and thicknesses of 10 to 800 μm. Chiplets 22 may be singulated from a larger substrate, such as a semiconductor wafer, which may be subjected to a thinning process.

[0014] The chiplet source portion 110 includes a source of one or more chiplets. In FIG. 1 , the chiplet source portion 110 includes a source substrate 25 and a source chuck 112 capable of holding the source substrate 25. A plurality of chiplets 22 are temporarily attached to the source substrate 36. In some embodiments, the source substrate 25 is a tape frame, and the source chuck 112 is configured to support mounting the tape frame and removing the chiplets 22 from the tape frame. In some embodiments, the source substrate 25 is a reel on which one or more chiplets 22 are mounted, and the source chuck 112 is a reel feeder. In some embodiments, the source substrate 25 is a tray with pockets for holding chiplets 22 in each pocket, and the source chuck 112 is a tray holder. Also, in some embodiments, the chiplet source portion 110 includes a front-opening integrated pod (“FOUP”) 114. The FOUP 114 may include a plurality of source substrates 25 and chiplets 22. Other chiplet sources known in the art may be used for chiplet source portion 110, such as a tray, adhesive tape on a frame, or an adhesive layer on a stiff substrate.

[0015] The transfer-activation section 120 includes a transfer robot 122 and an activation device 124. The transfer robot 122 can lift and transfer a substrate (e.g., source substrate 22) to a second substrate chuck 131 in the bonding section 130. As understood in the art, the transfer robot 122 typically includes a hand and a robot arm that provide a degree of motion for lifting, transferring, and placing a substrate from one location to another, such as from one substrate chuck to another or from a substrate storage location to a substrate chuck. Examples of the transfer robot 122 are commonly referred to as front-end equipment modules (EFEMs) and include robots for transferring various types of substrates between ultra-clean storage containers, such as FOUPs. The transfer robot 122 may be any suitable device known in the art, such as a wafer handling robot RR756L15 offered by Rorze Corporation of Fukuyama City, Hiroshima Prefecture, Japan.

[0016] The activation device 124 prepares the transferred chiplets 22 for hybrid bonding. Hybrid bonding is a chiplet bonding technique that contacts electrically insulating (silicon dioxide) chiplet surfaces with recessed metal pads (e.g., copper pads). The metal pads are aligned with each other, and the electrically insulating surfaces are bonded together via direct contact. An annealing process is then performed in which heat is applied to the bonded structure, causing the metal pads to expand relative to the electrically insulating material and contact each other, forming electrical connections between the chiplets 22 and the product substrate 29 (e.g., between the chiplets 22 and a chiplet previously bonded to the product substrate 29). For example, in some embodiments, when the chiplets 22 contact the product substrate 29, hydrogen bonds are formed between the chiplets 22 and the product substrate 29 (e.g., between the chiplets 22 and a chiplet previously bonded to the product substrate 29). The annealing then causes the metal pads on the chiplets 22 and the product substrate 29 to expand and fuse into covalent bonds. In some exemplary embodiments, activation device 124 includes a fluid source providing, for example, deionized water, and a plasma source to activate the surfaces of chiplets 22 before they are transferred by transfer robot 122 to chiplet bonding site 130. Due to the material (i.e., dielectric) of chiplets 22, when an activated chiplet 22 contacts another chiplet 22 (which may be bonded to product substrate 29), a fusion bond occurs between the dielectric surfaces of the two chiplets 22.

[0017] The bonding section 130 includes a second substrate chuck 131, a bridge 132, one or more bonding heads 133, a third substrate chuck 134 (also referred to as a product chuck 134), a substrate stage 135 (including a rotation stage 1351, an x-motion stage 1352, a y-motion stage 1353, and possibly other stages), a base 136, one or more transport heads 137, an upward alignment system 138, a downward alignment system 139, an upward microscope 141, a downward microscope 142, one or more shape measurement sensors 143, a reference wafer chuck 144, and an alignment microscope 145.

[0018] The second substrate chuck 131 receives the source substrate 25 from the transfer robot 122. The received source substrate 25 has chiplets 22 activated by the activation device 124. The second substrate chuck 131 is attached to a bridge 132. The second substrate chuck 131 is also referred to herein as a transfer chuck 131.

[0019] The third substrate chuck 134 is configured to hold the product substrate 29. The third substrate chuck 134 may be mounted on a substrate stage 135. The substrate stage 135 provides single-axis or multiple-axis (e.g., five-axis, six-axis) motion control over a limited range with millimeter to sub-millimeter accuracy. The substrate stage 135 may be a high-precision x-y-z-θ stage combined with an optical interferometry system so that absolute position is repeatedly achieved with high accuracy. In some embodiments, the substrate stage 135 includes a substrate rotation stage 1351, a substrate x-motion stage 1352, a substrate y-motion stage 1353, and possibly other stages.

[0020] Bonding section 130 also includes multiple bonding heads 133 attached to bridge 132. Each bonding head 133 can transfer chiplets 22 from one or more bonding heads 137 to a product substrate 29 held by third substrate chuck 134. Bonding heads 133 may be used in parallel. Bonding heads 133 are positioned opposite third substrate chuck 134. Each bonding head 133 includes a bonding head chiplet chuck 1331, and each bonding head 133 may include a bonding head chiplet stage 1332. Bonding head chiplet chuck 1331 is configured to hold chiplets 22 in a stable, fixed manner with positional stability within 1 μm. Bonding head chiplet chuck 1331 may be configured to hold the backside of chiplets 22. In some embodiments, bonding head chiplet chuck 1331 is configured to grip the edges of chiplets 22. The bonding head tiplet chuck 1331 may be, for example, a vacuum chuck, a latch-type chuck, an edge grip chuck, a pin-type chuck, a groove-type chuck, an electrostatic chuck, or an electromagnetic chuck.

[0021] The bonding head chiplet stage 1332 is a motion stage for controlling the position of the bonding head chiplet chuck 1331 relative to the bridge 132. The bonding head chiplet stage 1332 provides motion control in one or more directions, such as the z-axis direction, and other directions (e.g., one or more of the x-axis direction, y-axis direction, θ (rotation) axis direction, ψ (gradient) axis direction, and φ (tilt) axis direction). The bonding head chiplet stage 1332 may include one or more actuators or stages, such as a voice coil motor, a piezoelectric motor, a linear motor, a nut-and-screw motor, a piezo-actuated stage, a brushless DC motor stage, and a DC stepper motor. The positioning accuracy of the bonding head chiplet stage 1332 can be within 100 nm.

[0022] The bonding unit 130 includes a transfer head 137, and the bonding unit 130 may include multiple transfer heads 137 used in parallel. The transfer head 137 is used to transfer the chiplets 22 from the source substrate 25 held by the second substrate chuck 131 to the bonding head 133. In some embodiments, the transfer head 137 includes a tiplet chuck, which is a vacuum-type suction nozzle that can be moved by one or more actuators at least in a direction toward the second substrate chuck 131. The tip of the suction nozzle may be smaller than the chiplets 22. In some embodiments, the transfer head 137 includes a tiplet chuck for holding the chiplets 22, such as a Bernoulli chuck, a vacuum chuck, a pin-type chuck, a groove-type chuck, an electrostatic chuck, an electromagnetic chuck, a non-contact chuck, a PEEK plastic chuck, a suction cup, an edge grip chuck, or the like. The transfer head 137 may include one or more actuators or stages, such as a voice coil motor, a piezoelectric motor, a linear motor, a nut-and-screw motor, a piezo-actuated stage, a brushless DC motor stage, and a DC motor stage stepper motor, configured to move the chiplet chuck between the source substrate 25 and the bonding head 133, for example, in the z-axis direction and other directions (e.g., one or more of the x-axis, y-axis, θ-axis, ψ-axis, and φ-axis directions).

[0023] Bonding system 100 may include or be in communication with a chiplet pre-processing system (not shown). The pre-processing performed by the pre-processing system may include wet or dry chemical processes to prepare the surfaces of chiplets 22 prior to bonding the chiplets 22 to product substrate 29. Pre-processing of chiplets 22 may occur at any time before bonding the chiplets 22 to product substrate 29. For example, pre-processing may occur before the chiplets 22 are loaded onto product substrate 29. Also, for example, pre-processing may occur while the chiplets 22 are on source substrate 25, pre-processing may occur while the chiplets 22 are on transfer head 137, or pre-processing may occur while the chiplets 22 are on bonding head 133.

[0024] Chiplet bonding site 130 includes an upward-facing alignment system 138. Upward-facing alignment system 138 may be used to measure the positions of chiplets 22 on bonding head 133. Chiplet bonding site 130 also includes a downward-facing alignment system 139 that is used to measure bond sites on product substrate 29. In some embodiments, upward-facing alignment system 138 and downward-facing alignment system 139 are parts of a single system that can measure both chiplets 22 on bonding head 133 and bond sites on product substrate 29.

[0025] Chiplet bonding portion 130 may include at least one upward-looking microscope 141 for inspecting chiplets 22 on source substrate 25 held by second substrate chuck 131 and also for inspecting chiplets on bonding head 133. Bonding system 100 may include at least one downward-looking microscope 142 for inspecting product substrate 29.

[0026] The product substrate 29 includes a substrate set of interconnect contacts, and each chiplet 22 includes a chiplet set of interconnect contacts. The substrate set of interconnect contacts may be included on a chiplet 22 previously bonded to the product substrate 29. The chiplet set of interconnect contacts and the substrate set of interconnect contacts are generally referred to as interconnect contacts, which provide multiple connections between the chiplets 22 and the product substrate 29. In some embodiments, these multiple interconnect contacts provide electrical connections between the chiplets 22 and the product substrate 29. In some embodiments, the multiple interconnect contacts provide one or more of fluidic, optical, and electrical connections. During the bonding process, it is very important that the chiplet set interconnect contacts and the substrate set interconnect contacts are aligned with each other. This becomes more difficult as the size of each interconnect contact decreases and the density of multiple interconnect contacts increases. Each interconnect contact may be made of a conductive material, such as copper. The interconnect contacts may be flush with the surface of the chiplet 22 or product substrate 29, or the interconnect contacts may be recessed nanometers below the surface of the chiplet 22 or product substrate 29. The surface of the chiplet 22 or product substrate 29 may be a dielectric material such as silicon oxide or silicon nitride with the interconnect contacts recessed nanometers below the surface of the dielectric.

[0027] In some embodiments, product substrate 29 is a patterned semiconductor wafer having a substrate set of interconnect contacts. The substrate set of interconnect contacts may provide connections to components within or disposed on product substrate 29. Product substrate 29 may have multiple bonding sites for chiplet 22 and for other chiplets different from chiplet 22. Chiplets 22, which may be the same as or different from chiplet 22, may already be bonded to product substrate 29. As described above, the substrate set of interconnect contacts may be on chiplets 22 bonded to product substrate 29. In some embodiments, product substrate 29 is not a patterned semiconductor wafer, but has a substrate set of interconnect contacts.

[0028] One or more shape metrology sensors 143 can measure the height of chiplets 22 (chiplets 22 on source substrate 25 held by transfer chuck 131) relative to source substrate 25 or relative to a reference plane at multiple locations across chiplets 22 (or the height of chiplets 22 on bonding head 133 relative to bonding head 133). In this manner, one or more shape metrology sensors 143 can obtain shape measurements indicative of variations in flatness or non-flatness (non-planarity or deviation from planarity) of the surface of chiplets 22. Thus, the shape measurements are indicative of the surface topography (surface relief) of chiplets 22, and the surface topography is indicative of variations in non-flatness of the surface of chiplets 22.

[0029] For example, Figures 2A-2D show exemplary embodiments of surface profiles of a chiplet. In Figure 2A, the chiplet 22 is flat. However, particularly in a free-standing state, the chiplet 22 has significant curvature, which depends on the thickness of the chiplet 22 and the variations in stress in each layer within the chiplet 22. Generally, thinner chiplets 22 are more flexible and have greater curvature (e.g., due to their lower bending stiffness compared to thicker chiplets 22). In Figure 2B, the chiplet 22 with variations in non-flatness has a uniform spherical curvature.

[0030] 2C and 2D, the chiplets 22 have non-flatness variations (e.g., curvature) that vary across the x-axis, and the non-flatness variations of some chiplets 22 vary across the xy plane.

[0031] Examples of shape measurement sensors 143 include displacement measurement sensors such as optical sensors, air gauge sensors, interferometric sensors, capacitive sensors, and spectral interferometric sensors, two-dimensional (2D) digital image correlation (DIC) devices, three-dimensional (3D) digital image correlation (DIC) devices, white light interferometry devices, and confocal microscope devices.

[0032] The reference wafer chuck 144 may hold a reference wafer 31. The reference wafer 31 provides a reference point (a reference surface, such as a reference plane) for determining the flatness (non-flatness variation) or alignment of the chiplets 22 held by the bonding head 133. The chiplets 22 held by the bonding head 133 are compared to the reference wafer 31 (e.g., reference chiplets 33 on the reference wafer 31) to determine their respective non-flatness variation or alignment. The chiplets used as the reference chiplets 33 may have non-flatness variation that is within a threshold (e.g., may be as close to flat as possible according to the equipment and process used to fabricate them).

[0033] Alignment microscope 145 (or upward microscope 141) captures an image of reference wafer 31 (e.g., an image of an area on the reference wafer) and an image of a chiplet 22 held by one of bonding heads 133 (e.g., an image of an area on a chiplet 22). Alignment microscope 145 may also be aligned with reference wafer chuck 144 (and, via reference wafer chuck 144, reference wafer 31) and bonding head 133 (and, via bonding head 133, a test chiplet 22—a test chiplet 22 being a chiplet 22 held by a member of bonding portion 130). For example, alignment microscope 145 may capture an image of an area of ​​reference chiplet 33 and an image of a corresponding area of ​​a chiplet 22 held by bonding head 133. The corresponding areas of the chiplets are located at the same location (e.g., coordinates) as measured from their edges or centers. These regions may also include one or more distinct or unique patterns, marks, features, or edges that make the regions easier to identify. If the reference chiplets 33 and 22 are identical and perfectly aligned, the corresponding regions will be identical, and therefore the images of the regions will be identical. However, if they are not identical or not perfectly aligned, the images of the regions will not be identical.

[0034] For example, FIG. 3 shows an example of an image of a chiplet captured by alignment microscope 145 or upward microscope 141. Reference image 42 is an image of reference area 34 of reference chiplet 33 on reference wafer 31, and test image 43 is an image of corresponding test area 23 of test chiplet 22 (chiplet 22 held by members of bonding portion 130). Alignment microscope 145 is aligned with reference chiplet 33, and the field of view (imaging field of view) of reference image 42 is centered on x, y coordinates Cx, Cy in this example. Thus, bonding system 100 can use reference image 42 as an alignment template. Also, for example, if reference chiplet 33 has a width of 10 mm, the width of the imaging field of view may be 500 μm. Thus, reference area 34 of reference chiplet 33 in reference image 42 is 500 μm wide. The field of view of test image 43 is centered on Cx, Cy. When the reference chiplets 33 and test chiplets 22 are identical and aligned, and consequently the reference areas 34 and corresponding test areas 23 are identical and aligned, the reference image 42 and test image 43 are identical.

[0035] However, in this example, test chiplet 22 is rotated relative to reference chiplet 33. Thus, test image 43 is not identical to reference image 42, reference area 34 and corresponding test area 23 are not aligned, and structures on test chiplet 22 (e.g., copper pads) are rotated relative to structures on reference chiplet 33.

[0036] Additionally, bonding system 100 (e.g., controller 150) maps (e.g., calibrates, aligns) bonding system 100 components, subsystems, motion, metrology, etc. to a global (i.e., bridge 132 and / or base 136 world) coordinate system. Test chiplets 22 aligned to reference chiplets 33 need not be physically superimposed on the reference chiplets 33. Instead, bonding head chiplet chuck 1331 may position test chiplets 22 such that they hold one or more upward-looking microscopes 141, one or more shape measurement sensors 143, or alignment microscope 145 in the same positional relationship as the reference chiplets 33 on the reference wafer 31, as indicated by the position of reference wafer chuck 144.

[0037] Bonding system 100 also includes one or more processors 151 and one or more computer-readable storage media 152. The one or more processors 151 and the one or more computer-readable storage media 152 may be components of a controller 150. Bonding system 100 is coordinated, controlled, or directed by the one or more processors 151, which are in communication with one or more components or subsystems of bonding system 100, such as source chuck 112, transfer robot 122, activation device 124, second substrate chuck 131, third substrate chuck 134, substrate stage 135, transfer head 137, bonding head 133, upward-facing alignment system 138, downward-facing alignment system 139, upward-facing microscope 141, downward-facing microscope 142, shape measurement sensor 143, reference wafer chuck 144, and alignment microscope 145.

[0038] The one or more processors 151 may be or include one or more central processing units (CPUs), such as a microprocessor (e.g., a single-core microprocessor, a multi-core microprocessor), one or more graphics processing units (GPUs), one or more application-specific integrated circuits (ASICs), one or more field-programmable gate arrays (FPGAs), one or more digital signal processors (DSPs), or other electronic circuits (e.g., other integrated circuits). Furthermore, the processor 151 may be a dedicated controller or a general-purpose controller. The one or more processors 151 may include multiple processors, including both (i) a processor included in the control device 150 and (ii) a processor that communicates with the bonding system 100 but is not included in the control device 150. Furthermore, the one or more processors 151 are an example of a processing unit.

[0039] The one or more processors 151 may operate based on computer-readable instructions in one or more programs stored on one or more computer-readable storage media 152. As used herein, computer-readable storage media 152 refers to computer-readable media including, for example, magnetic disks (e.g., floppy disks, hard disks), optical disks (e.g., CDs, DVDs, Blu-rays), magneto-optical disks, magnetic tapes, and articles of manufacture such as semiconductor memory (e.g., non-volatile memory cards, flash memory, solid-state drives, SRAM, DRAM, EPROM, EEPROM). Examples of the one or more computer-readable storage media 152 include network-attached storage (NAS) devices, intranet-attached storage devices, and internet-attached storage devices. The one or more computer-readable storage media 152, which may include both ROM and RAM, can store computer-readable data or computer-executable instructions. Furthermore, in embodiments in which the one or more computer-readable storage media 152 include RAM, the one or more processors 151 can use the RAM as a work area. Additionally, when the controller 150 or one or more processors 151 are described as obtaining, recording, generating, storing, operating on, or processing information or data, the information or data is stored on one or more computer-readable storage media 152. Additionally, one or more computer-readable storage media 152 is an example of a storage unit. Additionally, non-transitory computer-readable storage media 152 may be distributed among multiple processors 151.

[0040] Controller 150 also includes I / O components 153. I / O components 153 include physical interface and communication components (e.g., GPU, network interface controller) that enable communication (wired or wireless) with other components of bonding system 100 (e.g., source chuck 112, transfer robot 122, activation device 124, second substrate chuck 131, third substrate chuck 134, substrate stage 135, transfer head 137, bonding head 133, upward-facing alignment system 138, downward-facing alignment system 139, upward-facing microscope 141, downward-facing microscope 142, shape measurement sensor 143, reference wafer chuck 144, and alignment microscope 145), other computing devices (e.g., network computer 160), and input or output devices that may include display device 155, network device, keyboard 156, mouse, printing device, light pen, optical storage device, scanner, microphone, drive, joystick, and control pad.

[0041] Additionally, the hardware components of controller 150 communicate via one or more buses 154 or other electrical connections. Examples of buses 154 include a Universal Serial Bus (USB), an IEEE 1394 bus, a PCI bus, an Accelerated Graphics Port (AGP) bus, a Serial AT Attachment (SATA) bus, and a Small Computer System Interface (SCSI) bus.

[0042] Network computer 160 may perform analysis and provide information, such as information about the substrate and chiplets. In some embodiments, there are one or more graphical user interfaces (GUIs) 162 on one or both of network computer 160 and display device 155 that communicate directly with controller 151 that are presented to an operator or user.

[0043] 4 illustrates an exemplary embodiment of an operational flow for bonding chiplets to a substrate. While this and other operational flows described herein are each presented in a particular individual order, some embodiments of these operational flows perform at least some of the operations in an order different from the order presented. Examples of different orders include simultaneous, parallel, overlapping, reordered, synchronous, incremental, and interleaved orders. Also, some embodiments of these operational flows include operations (e.g., blocks) from two or more of the operational flows described herein. As such, some embodiments of the operational flows may omit blocks, add blocks (e.g., include blocks from other operational flows described herein), change the order of blocks, combine blocks, or divide blocks into more blocks than the exemplary embodiments of the operational flows described herein.

[0044] This and other operational flows described herein are performed by bonding system 100, and one or more controllers 150 can control bonding system 100 to perform the operations described in the operational flows.

[0045] In FIG. 4, the first flow begins at block B405 and the second flow begins at block B430.

[0046] In block B405, bonding system 100 loads source substrate 25 onto transfer chuck 131. The loading in blocks B405 and B430 may be performed by one or more robots. Examples of one or more robots include a wafer handling robot, a chiplet feed system, and a tray handler.

[0047] Next, in block B410, bonding system 100 inspects source substrate 25 loaded onto transfer chuck 131. The inspection includes acquiring images of source substrate 25 using upward-looking microscope 141. Controller 150 can analyze these images to identify the location of test chiplets 22 (source chiplets held by members of bonding portion 130) on source substrate 25.

[0048] The first flow then proceeds to block B415, where one or more transfer heads 137 transfer the test chiplets 22 from the source substrate 25 to one or more bonding heads 133. The transfer can be performed with a precision that is within the confines of the bonding head chiplet stage 1332. In some embodiments, the one or more bonding heads 133 directly remove the one or more test chiplets 22 from the source substrate 25. The approximate position of the test chiplets 22 on the bonding head 133 may be measured using an upward-looking microscope 141. The upward-looking microscope 141 may be a high numerical aperture (NA) microscope capable of resolving approximately 3 μm marks or interconnect contacts at the corners of the test chiplets 22.

[0049] Next, in block B420, alignment microscope 145 captures one or more test images of test chiplet 22. The one or more test images may be images of individual test areas 23 on test chiplet 22, where test areas 23 correspond to individual reference areas 34 on reference chiplet 33. For example, FIGS. 5A and 5B show exemplary embodiments of test areas 23 on test chiplet 22. FIG. 5A shows both a view along the z-axis and a view along the y-axis, illustrating the surface profile (showing the surface topography) of test chiplet 22. FIG. 5B shows a close-up view of one test area 23. Test chiplet 22 is not flat; rather, test chiplet 22 has variations in non-flatness (surface topography) including various curvatures or irregularities. Also, in FIG. 5, the test areas have variations in non-flatness.

[0050] Next, in block B425, one or more shape measurement sensors 143 acquire shape measurements of the surface of the test chiplet 22. For example, one or more shape measurement sensors 143 may acquire individual shape measurements of each test area 23 on the test chiplet 22. The shape measurements indicate variations in the non-flatness (surface topography) of the surface of the test area 23, and thus the surface measurements indicate any deviations from planarity of the test area 23. For example, the shape measurements may include individual 3D coordinates for multiple locations on the surface of the test area 23.

[0051] Then, in block B450, the first flow is connected to the second flow.

[0052] The second flow begins at block B430, where bonding system 100 loads product substrate 29 onto product chuck 134. Loading can include standard lithographic alignment techniques using alignment marks such as moiré interference marks, two-dimensional gratings, crosses, box-in-box, bar-in-bar, bull's-eye, edge marks, serpentine marks, arrays of interconnect contacts, etc.

[0053] Next, in block B435, bonding system 100 inspects product substrate 29 loaded on product chuck 134. The inspection includes acquiring images of product substrate 29 using downward-looking microscope 142. Controller 150 can analyze these images to inspect product substrate 29 and identify one or more bonding sites on product substrate 29.

[0054] The second flow then proceeds to block B440, where the alignment microscope 145 captures one or more reference images 42 of the reference chiplet 33 to which the test chiplet 22 is aligned. For example, the alignment microscope 145 may capture individual reference images 42 of one or more reference regions 34 on the reference chiplet 33. The one or more reference regions 34 may include several unique or distinct features that facilitate identification of the one or more reference regions 34.

[0055] Next, in block B445, one or more shape measurement sensors 143 obtain shape measurements of the surface of the reference chiplet 33. For example, the one or more shape measurement sensors 143 may obtain individual shape measurements of each reference region 34 on the reference chiplet 33. Also, for example, the shape measurements may include individual 3D coordinates for multiple locations on the surface of the reference region 34. The second flow then moves to block B450, where the second flow connects with the first flow.

[0056] In block B450, the bonding system 100 obtains an observed alignment error (measured alignment error), which is the alignment difference between the test chiplet 22 and the reference chiplet 33 (e.g., between the reference area 34 on the reference chiplet 33 and the test area 23 on the test chiplet 22). Then, in the flow of FIG. 4 , the bonding system 100 performs a position adjustment based on the observed alignment error. The position adjustment compensates for (e.g., removes) the alignment error, and the position adjustment includes one or more processors 151 sending movement commands to one or more bonding head chiplet stages 1332 or substrate stage 135. To obtain the observed alignment error, the one or more processors can perform digital image correlation (DIC). For example, some embodiments of digital image correlation are described by the following:

[0057]

number

[0058] where g a is the reference image 42 of the reference region 34, and g b is the test image 43 of the test area 23, and G a is the 2D Fourier transform of the reference image 42, and G b is the 2D Fourier transform of the test image 43, and R is the a and the complex conjugate G b * is the Fourier transform of the correlation curve, obtained by taking the element-wise normalized Hadamard product of jk is G a,jk Each element of and the complex conjugate of G b * ,jk where r is the cross-correlation, calculated by performing the inverse Fourier transform of R, and (Δx, Δy) is an integer shift. Δx and Δy are the x and y coordinates at the maximum value of r, which are obtained by determining the x and y locations of the maximum value of r. The result of digital image correlation may also be a correlation image showing the observed alignment error.

[0059] Observed alignment error err ob are all seen to be in-plane alignment errors (translation, in-plane rotation). However, the observed alignment error err ob is the component due to the in-plane alignment error (the in-plane error component err ip ") and a component due to curvature (out-of-plane variations) of the chiplet 22 ("curvature-induced component cc"). In this way, the observed alignment error err ob is the in-plane error component err ip and the curvature component cc (err ob =err ip + cc). Furthermore, the curvature-induced component cc may be calculated by subtracting the observed alignment error err obAlthough it appears to be an in-plane alignment error in , the curvature-induced component cc may not actually be due to alignment errors that include in-plane alignment errors. When chiplet 22 is pressed against product substrate 29, non-flatness variations in chiplet 22 are planarized (reduced or eliminated), thus reducing or eliminating any in-plane alignment errors that may be present due to the curvature-induced component cc.

[0060] As a result, the curvature-induced component cc may not actually indicate an alignment error (because the curvature-induced component cc is reduced or eliminated as the chiplet 22 is pressed against the product substrate 29), and the curvature-induced component cc may actually indicate an alignment error err ob The curvature-induced component cc may cause alignment measurement errors because it may appear to be one or more in-plane alignment errors in . Thus, because the curvature-induced component cc may not actually indicate an in-plane alignment error, the curvature-induced component cc causes or acts as an in-plane error in the alignment measurement, and is also referred to as an alignment measurement error.

[0061] As noted above, the variation in non-flatness (e.g., curvature) depends on the thickness t of the chiplet, with thinner chiplets 22 tending to be more flexible and less rigid. For example, a chiplet 40 μm thick and 5×5 to 6×6 mm in size has a spherical curvature of 20 μm. In this situation, for a 500×500 μm imaging field of view (FOV), the maximum shape variation across the field of view is approximately 4 μm.

[0062] Also, as mentioned above, variations in the non-flatness of the chiplets 22 can contribute to the observed alignment error err ob For example, in some embodiments, the curvature-induced component cc due to variations in the non-flatness of the tiplets 22 may be described by:

[0063]

number

[0064] where t av is the average thickness of the chiplet 22 in the image field of view, and θ is the slope of the surface of the chiplet 22. The slope is calculated relative to a reference plane, which in this embodiment is the reference chiplet 33, although some embodiments use a flat surface as the reference plane. Thus, in some embodiments, the slope θ is calculated based on shape measurements of the test chiplet 22 and the reference chiplet 33. For example, the shape measurements of the test chiplet 22 and the reference chiplet 33 are used to calculate the non-flatness variation nfv of the test chiplet 22, which can be used to calculate the slope θ.

[0065] Another source of the curvature-induced component cc is variation in the thickness t of the chiplet 22 (chiplet thickness t). Thus, the tiplet thickness t may be measured by the shape measurement sensor 143 (e.g., in block B425). Typically, a laser displacement sensor or other similar sensor measures the distance D to the surface of the bonding head chiplet chuck 1331 in the z-axis direction. z (excluding chiplets 22), and then the distance (D z +t+Δt) and calculate the chiplet thickness t(t+Δt=D z +t+Δt-D z ) can be obtained. This assumes that the sensor has a sensing range of t + Δt. Spectral interference sensors, interferometers, capacitive sensors, or air gauges can similarly be used for this precise chiplet thickness measurement.

[0066] Also, variations in chiplet non-flatness nfv are introduced in both in-plane (x and y) directions, for example as described by:

[0067]

number

[0068] where M is the non-flatness angle θ of the chiplet relative to an orthogonal in-plane axis used to project the vector defined by the non-flatness angle onto a reference plane. x and θ y is a linear matrix or system of equations consisting of trigonometric geometry transformation terms that depend on

[0069] For example, Figures 6A and 6B show example surface profiles of a reference chiplet and a test chiplet. Figure 6A shows a test area 23A on test chiplet 22 and a corresponding reference area 34A on reference chiplet 33. Also, while both test area 23A and reference area 34A contain variations in non-planarity, reference chiplet 33 is generally flatter than test chiplet 22 (and in particular, reference area 34A). Figure 6B shows the slope θ between the surface of chiplet 22 in test area 23A and the surface of reference chiplet 33 in the corresponding reference area 34A.

[0070] The slope θ indicates the angle between the surface of the chiplet 22 and the corresponding surface of the reference chiplet 33. The slope θ may be the average slope of the chiplet 22 in the imaged field of view, the maximum slope of the chiplet 22 in the imaged field of view, the angle in each of the x-axis and y-axis directions of a plane fitted to the topography of the chiplet 22 in the field of view, or the angle in each of the x-axis and y-axis directions of a plane fitted to the edge of the chiplet 22 in the imaged field of view.

[0071] For example, in an imaging field of view of 500 x 500 μm, if the maximum non-flatness variation nfv (deviation from planarity) is 4 μm and the average thickness t is 100 μm, the curvature-induced component cc due to non-flatness variation can be calculated as 800 nm (100 / 2*4 / 250 μm). Furthermore, if the target is, for example, a bonding with an alignment error that is less than 50-100 nm, such a curvature-induced component cc can be calculated as 800 nm (100 / 2*4 / 250 μm) as shown in this example. ob , or potentially orders of magnitude larger than the in-plane alignment error. Furthermore, if the goal is an alignment measurement error of less than 20 nm, such curvature-induced component cc will prevent the goal from being achieved. For example, in some embodiments, the metal pads on chiplet 22 and product substrate 29 each have a radius of 1 μm to 2 μm, and the goal is to align the metal pads on chiplet 22 and product substrate 29 with an alignment accuracy of 10 nm.

[0072] To meet the alignment measurement error goal of 10 nm or less, the chiplets 22 must be flatter, for example, with a 100 μm thick chiplet, having a non-flatness variation of 50 nm or less across a 500 μm imaging field. Thicker chiplets 22, such as 300 μm thick chiplets 22, must be even flatter, with a variation of 17 nm or less across a 500 μm imaging field. However, fabricating such flat chiplets 22 is challenging for both thick and thin chiplets 22. Typically, after chuck-mounting the chiplets 22, the chiplets 22 still have a non-flatness variation of hundreds to thousands of nanometers (microns) across the chiplet 22 or across the imaging field. For example, a non-flatness variation of 1 μm across a 500 μm imaging field can result in an in-plane error of 25 nm in the alignment measurement. Furthermore, in some embodiments, many chiplets 22 are stacked on top of each other on product substrate 29, and chiplets 22 tend to be thinner.

[0073] Thus, in block B455, bonding system 100 calculates the observed alignment error err based on the shape measurements of test chiplet 22 and the shape measurements of reference chiplet 33. ob In some embodiments, at least a portion of the calculation of the curvature-induced component cc is described by equation (6).

[0074] Next, in block B460, bonding system 100 calculates the curvature-induced component cc and the observed alignment error err ob Based on this, the in-plane error component err ip For example, the bonding system 100 may calculate the in-plane alignment error (chiplet position error) as follows: ob By subtracting the curvature-induced component cc from ip(In-plane alignment error) may be calculated (err ip =err ob -cc). As a result, in blocks B455-B460, bonding system 100 calculates an in-plane alignment error based on the image of test area 23 on chiplet 22 and the deviation from planarity (non-flatness variation) of test area 23. Furthermore, the in-plane alignment error (chiplet position error) accounts for the alignment difference between test chiplet 22 and reference chiplet 33 (e.g., between reference area 34 on reference chiplet 33 and test area 23 on test chiplet 22).

[0075] Flow then proceeds to block B465, where bonding system 100 aligns test chiplets 22 with bonding sites on product substrate 29. For example, bridge 132 and product chuck 134 (or substrate stage 135) may have alignment marks that are used to align up-looking microscope 141 and alignment microscope 145 with bridge 132, and down-looking microscope 142 with product chuck 134 (or substrate stage 135). Such alignment aligns up-looking microscope 141, alignment microscope 145, and down-looking microscope 142 with a fixed reference frame. Product substrate 29 also includes features or global alignment marks that can be identified from the image of product substrate 29 acquired in block B435, and these features can be used to align product substrate 29 to a global coordinate system and then align reference chiplets 33 to product substrate 29, such as so that structures on reference chiplets 33 (e.g., electrical pads such as metal pads, interconnect contacts) are aligned with respective corresponding structures on product substrate 29 (e.g., electrical pads such as metal pads, interconnect contacts). Bonding head 133 may also be aligned to reference wafer chuck 144, which enables bonding system 100 to align test chiplets 22 held by bonding head 133 to bonding sites on product substrate 29 that are aligned with reference chiplets 33.

[0076] Flow then moves to block B470, where bonding system 100 adjusts the position of test chiplet 22 relative to product substrate 29 based on the in-plane alignment error (adjusted alignment difference). For example, in some embodiments, the adjusted position ap is described by:

[0077]

number

[0078] where g(x) is a correction or compensation function (eg, a matrix).

[0079] The substrate stage 135 may include stage position encoders that enable precise positioning of the product chuck 134 and product substrate 29 relative to the bonding head 133. Examples of stage position encoders include laser light interferometer position measurement systems, ultrasonic distance measurement systems, capacitive displacement measurement systems, and optical position encoders. Some embodiments also use other methods of measuring the position of an object with sub-micron resolution. Each of the bonding head chiplet stages 1332 may include a bonding head position encoder similar to the stage position encoder.

[0080] Next, in block B475, bonding system 100 bonds the test chiplets 22 to the product substrate 29. In block B475, the bonding head 133 holding the test chiplets 22 is moved toward the product chuck 134. The bonding head chiplet stage 1332 may include one or more actuators that can move the bonding head chiplet chucks 1331 toward the product chuck 134 until the test chiplets 22 held by the respective bonding head chiplet chucks 1331 contact the bonding sites on the product substrate 29.

[0081] If bonding system 100 is performing a hybrid bonding method, the bonding surfaces of test chiplets 22 to be bonded to product substrate 29 are activated before block B475. For example, the bonding surfaces of test chiplets 22 may be activated while the test chiplets 22 are on bonding head 133, or while the test chiplets 22 are on transfer head 137, or while the test chiplets 22 are on source substrate 25 held by transfer chuck 131, or the bonding surfaces of test chiplets 22 may be activated before block B405. The bonding surfaces of product substrate 29 may be activated while product substrate 29 is held by product chuck 134, or the bonding surfaces of product substrate 29 may be activated before block B430. Activating the bonding surfaces may include rinsing the bonding surfaces with deionized water and exposing the bonding surfaces to plasma. Other known methods may be used to prepare the bonding surface so that dangling bonds are created on the bonding surface.

[0082] After block B475, the first flow proceeds to block B480, where bonding system 100 checks source substrate 25 held by transfer chuck 131 to determine whether the source substrate 25 is empty (i.e., has no chiplets 22 to be bonded to product substrate 29). This may include one or more processors 151 checking information about the source substrate 25 in non-transitory computer-readable storage medium 152 to determine how many chiplets 22 to be bonded were on source substrate 25 and how many have already been removed. If source substrate 25 is not empty of chiplets 22 to be bonded (B480=No), the first flow returns to block B415.

[0083] If the source substrate 25 is empty of chiplets 22 to be bonded (B480=Yes), the first flow proceeds to block B485, where one or more processors 151 send instructions for the source substrate 25 to be removed from the bonding system 100. Also, in some embodiments, the source substrate 25 (e.g., an array of chiplets on a source frame) may have more chiplets 22 than the number of bonding sites on the product substrate 29 (e.g., the source substrate 25 has enough chiplets 22 to supply chiplets 22 to more than one product substrate 29). For example, the source substrate 25 may have more Type 1 chiplets 22 than the number of bonding sites for Type 1 chiplets 22 on the product substrate 29. In this case, the bonding system 100 may unload the source substrate 25 and reload it with another source substrate 25 having Type 2 chiplets 22 to be bonded to the product substrate 29, even if the chiplets 22 are not completely empty. The first flow then returns to block B405 unless there are no more chiplets 22 to bond, in which case the first and second flows end.

[0084] After block B475, the second flow proceeds to block B490, where bonding system 100 checks production substrate 29 to determine whether all bonding sites on production substrate 29, including bonding sites for different chiplet types, are populated (bonded to chiplets 22). This may include one or more processors 151 checking information about production substrate 29 in non-transitory computer-readable storage medium 152 to determine how many bonding sites (sites to be bonded) were on production substrate 29 and how many of those bonding sites already have chiplets 22 bonded to them after completion of block B475. If production substrate 29 is not complete (B490=No) and has bonding sites that have not yet been bonded to chiplets 22, the second flow returns to block B435. If production substrate 29 is complete (B490=Yes) and there are no vacant bonding sites on production substrate 29, the second flow proceeds to block B493.

[0085] In block B493, bonding system 100 unloads product substrate 29. For example, one or more processors 151 may send instructions for product substrate 29 to be unloaded from product chuck 134 and removed from bonding system 100. The second flow then returns to block B430 unless there are no more chiplets 22 to bond, in which case the second flow ends.

[0086] Alternatively, if the product substrate 29 is bonded using a hybrid bonding method, the third flow moves from block B493 to block B495. In block B495, the product substrate 29 is heated to expand the metal interconnect contacts toward the bonding surface and form electrical connections between the chiplets 22 and the product substrate 29. Heating may be performed at 200°C to 300°C in an inert environment under an external pressure to prevent oxidation of the metal interconnects. In some embodiments, the heating in block B495 is performed at a higher temperature than the bonding in block B475. The product substrate 29 may then undergo one or more additional bonding methods to attach additional chiplets 22 to the product substrate 29. After block B495, the product substrate 29 may also undergo additional semiconductor process steps, such as singulation, testing, and encapsulation, used to fabricate multiple articles from the product substrate 29.

[0087] Figure 7 is a diagram of an exemplary embodiment of a bonding system 100 configured to bond chiplets to a substrate. This embodiment is similar to the embodiment shown in Figure 1, except that the bonding system 100 in Figure 7 includes two or more alignment microscopes 145.

[0088] For example, FIG. 8 shows an exemplary embodiment of two alignment microscopes 145. The two alignment microscopes 145 form a stereoscopic system. The alignment microscopes 145 capture images that bonding system 100 (e.g., controller 150) can use to generate a three-dimensional (3D) image of the imaged area of ​​the chiplet's surface (e.g., test area 23 of test chiplet 22, reference area 34 of reference chiplet 33). The 3D image contains both image information and shape measurements (non-flatness variations). Thus, a pixel in the 3D image represents both a visual appearance (visual value of the pixel) and a three-dimensional coordinate, and the 3D image defines both the visual appearance and the surface topography of the imaged area of ​​the chiplet.

[0089] The bonding system may also include two or more upward-looking microscopes 141 that form a stereoscopic system. The two or more upward-looking microscopes 141 can capture images of the entire chiplet (test chiplet 22, reference chiplet 33) and can be used to generate a 3D global image of the chiplet.

[0090] 9 shows an exemplary embodiment of an operational flow for bonding chiplets to a substrate. The flow begins in block B900, where bonding system 100 inspects product substrate 29 held by product chuck 134. The inspection includes acquiring images of product substrate 29 using downward-looking microscope 142, and controller 150 can analyze these images to inspect product substrate 29 and identify one or more bonding sites on product substrate 29.

[0091] Next, in block B905, bonding system 100 acquires a three-dimensional (3D) reference image 42 of reference chiplet 33. The 3D reference image 42 may be a 3D image of a reference area 34 on the surface of reference chiplet 33. Controller 150 uses the captured images from two or more alignment microscopes 145 to generate a 3D reference image 42 of the front surface of reference chiplet 33 in a global (world) coordinate system (the coordinate system to which all other subsystems, motion, metrology, etc. are referenced, mapped, calibrated, or aligned). As described above, a 3D image, such as 3D reference image 42, contains both image information and shape measurements. Typically, the surface of a reference chiplet is flat, with very little non-flatness variation in the global coordinate system.

[0092] Flow then moves to block B910, where bonding system 100 acquires a 3D test image 43 of test chiplet 22 aligned with reference chiplet 33. 3D test image 43 may be a 3D image of test area 23 on test chiplet 22. To acquire 3D test image 43 of test chiplet 22, controller 150 uses captured images from two or more alignment microscopes 145 to generate 3D test image 43 of the front surface of test chiplet 22 in a global coordinate system.

[0093] Then, in block B915, bonding system 100 obtains (e.g., calculates) the observed alignment error and shape difference between 3D reference image 42 and 3D test image 43. For example, controller 150 may perform 3D digital image correlation using reference image 42 and 3D test image 43, and the results of the 3D digital image correlation are decomposed to indicate both the observed alignment error and shape difference between 3D reference image 42 and 3D test image 43. The shape difference indicates the variation in non-flatness of test chiplet 22 relative to reference chiplet 33 (e.g., the variation in non-flatness of test area 23 of test chiplet 22 relative to reference area 34 of reference chiplet 33).

[0094] Flow then proceeds to block B920, where bonding system 100 calculates a curvature-induced component of the observed alignment error based on the shape differences (e.g., variation in non-flatness of the test region 23 of the test chiplet 22 relative to the reference region 34 of the reference chiplet 33). As described above, calculating the curvature-induced component may include calculating a slope θ between a surface of the test chiplet 22 and a corresponding surface of the reference chiplet 33. Also, as described above, the slope θ may be the average slope of the test chiplet 22 in the imaged field of view, the maximum slope of the test chiplet 22 in the imaged field of view, the angle in each of the x-axis and y-axis directions of a plane fitted to the topography of the test chiplet 22 in the field of view, or the angle in each of the x-axis and y-axis directions of a plane fitted to the edge of the chiplet 22 in the imaged field of view.

[0095] Then, in block B925, bonding system 100 calculates an in-plane alignment error based on the curvature-induced component and the observed alignment error, for example, by subtracting the curvature-induced component from the observed alignment error. Flow then proceeds to block B930, where bonding system 100 adjusts the in-plane position of test chiplet 22 based on the in-plane alignment error. Then, in block B935, bonding system 100 bonds test chiplet 22 to product substrate 29.

[0096] 10 illustrates an exemplary embodiment of an operational flow for bonding a chiplet to a substrate. The flow begins in block B1000, where bonding system 100 acquires a reference image 42 of a reference region 34 of a reference chiplet 33 using one or more alignment microscopes 145. Next, in block B1005, bonding system 100 acquires shape measurements of the reference region 34. For example, bonding system 100 may use one or more shape measurement sensors 143 to measure individual z-axis coordinates of multiple locations (e.g., pixels) on the surface of the reference region 34. Also, in some embodiments, bonding system 100 acquires a 3D reference image 42 of the reference region 34 in block B1000 using two or more alignment microscopes 145, and block B1005 is omitted in such embodiments because the 3D reference image 42 includes shape measurements of the reference region 34.

[0097] In block B1010, bonding system 100 acquires measurements of the global tilt of a test chiplet 22 held by bonding head chiplet chuck 1331 of bonding head 133. Tilt, in the context of this application, refers to the tip, tilt, or tip and tilt of the bonding surface of the chiplet. For example, bonding system 100 may acquire z-axis measurements across the surface of the test chiplet 22 using one or more shape measurement sensors 143 (e.g., displacement measurement sensors), such as optical sensors, air gauge sensors, interferometric sensors, capacitive sensors, and spectral interferometric sensors. The sampling frequency of the z-axis measurements may be low (e.g., 5-10 positions on the surface of the test chiplet 22). Bonding system 100 then calculates the global tilt based on the acquired measurements. Bonding system 100 may also acquire a 3D global image of the test chiplet 22 and calculate the global tilt based on the 3D global image of the test chiplet 22. For example, global tilt may be calculated by fitting a plane to the test chiplet 22 or by calculating the average or median of individual surface normals at multiple locations on the chiplet (e.g., test chiplet 22, reference chiplet 33).

[0098] Next, in block B1015, bonding system 100 adjusts the global tilt of test chiplet 22 based on the global tilt of test chiplet 22. After block B1015, test chiplet 22 has no global tilt or negligible global tilt (tilt within an acceptable range). Block B1015 may also include adjusting (translating) the position of test chiplet 22 on the z-axis.

[0099] Flow then proceeds to block B1020 where bonding system 100 aligns the tilt-adjusted test chiplet 22 to the reference chiplet 33.

[0100] Flow then moves to block B1025, where bonding system 100 acquires a test image 43 of test area 23 of test chiplet 22. Then, in block B1030, bonding system 100 acquires shape measurements of test area 23. For example, bonding system 100 may measure individual z-axis coordinates of multiple locations (e.g., pixels) on the surface of test area 23. Furthermore, in some embodiments, bonding system 100 acquires a 3D test image 43 of test area 23 in block B1025, and block B1030 is omitted in such embodiments because 3D test image 43 includes shape measurements of test area 23.

[0101] Next, in block B1035, bonding system 100 obtains (e.g., calculates) an observed alignment error between reference image 42 and test image 43. Then, in block B1040, bonding system 100 calculates a curvature-induced component of the observed alignment error based on the shape measurements of reference region 34 and the shape measurements of test region 23. For example, calculating the curvature-induced component may include calculating a shape difference indicative of a variation in non-flatness of test region 23 relative to reference region 34. Similarly, calculating the curvature-induced component may include calculating a slope θ between a surface of test chiplet 22 and a corresponding surface of reference chiplet 33.

[0102] Then, in block B1045, bonding system 100 calculates an in-plane alignment error based on the curvature-induced component and the observed alignment error, for example, by subtracting the curvature-induced component from the observed alignment error. Flow then moves to block B1050, where bonding system 100 adjusts the in-plane position of test chiplet 22 based on the in-plane alignment error by translating test chiplet 22 on one or more of the x-axis and y-axis or by rotating test chiplet 22 about the z-axis.

[0103] Next, in block B1055, bonding system 100 measures the tilt of product substrate 29 at the bonding site (the bonding site of test chiplet 22). The tilt of product substrate 29 may be measured using downward-looking alignment system 139 or downward-looking microscope 142, or the tilt of product substrate 29 may be received by processor 151 based on a previous measurement of product substrate 29 outside of bonding system 100. Then, in block B1060, bonding system 100 adjusts the global tilt of test chiplet 22 to match the tilt of product substrate 29 at the bonding site.

[0104] Next, in block B1065, bonding system 100 bonds test chiplets 22 to product substrate 29.

[0105] 11 illustrates an exemplary embodiment of an operational flow for bonding a chiplet to a substrate. The flow begins in block B1100, where bonding system 100 acquires 3D reference image 42 of reference area 34 of reference chiplet 33.

[0106] Flow then moves to block B1105, where bonding system 100 acquires a 3D global image of test chiplet 22 held by bonding head chiplet chuck 1331 of bonding head 133. Controller 150 uses the captured images from two or more upward-looking microscopes 141 to generate the 3D global image of test chiplet 22.

[0107] Flow then moves to block B1110, where bonding system 100 calculates a global tilt of test chiplet 22 relative to reference chiplet 33 based on the 3D global image of test chiplet 22, which is assumed to be level. For example, in some embodiments, bonding system 100 acquires a 3D global image of reference chiplet 33 and calculates the global tilt of test chiplet 22 by comparing the 3D global image of test chiplet 22 with the 3D global image of reference chiplet 33, e.g., using 3D DIC. To calibrate bonding system 100 to perform 3D DIC, multiple images of a planar dot pattern with spacing can be used to acquire one or more of the intrinsic characteristics of a stereoscopic vision system (e.g., two or more upward-looking microscopes 141, two or more alignment microscopes 145). Examples of unique characteristics include the image plane center position of each imaging device (e.g., upward microscope 141, alignment microscope 145), the horizontal and vertical scale factors of each imaging device, the skew imaging distortion of each imaging device, the radial distortion factor of each imaging device, the relative attitude of each imaging device to other imaging devices, and the relative position of each imaging device to other imaging devices.

[0108] Also, for example, global tilt may be calculated as the average out-of-plane difference between a level reference plane and the 3D global image of the test chiplet 22.

[0109] Next, in block B1115, bonding system 100 adjusts the global tilt of test chiplet 22. Adjusting the global tilt reduces or eliminates the global tilt of test chiplet 22 relative to a reference plane (e.g., relative to reference chiplet 33).

[0110] Flow then moves to block B1120, where bonding system 100 aligns the tilt-adjusted test chiplet 22 with the reference chiplet 33. Then, in block B1125, bonding system 100 acquires a 3D test image 43 of the test area 23 of the test chiplet 22.

[0111] At block B1130, bonding system 100 obtains, for example, using 3D DIC, an observed alignment error at the adjusted tilt based on 3D test image 43 of test area 23 of test chiplet 22 and 3D reference image 42 of reference area 34 of reference chiplet 33. Flow then proceeds to block B1135, where bonding system 100 calculates, for example, using 3D DIC, a curvature-induced component of the observed alignment error at the adjusted tilt based on 3D test image 43 of test area 23 of test chiplet 22 and 3D reference image 42 of reference area 34 of reference chiplet 33.

[0112] Then, in block B1140, bonding system 100 calculates the in-plane alignment error at the adjusted tilt based on the curvature-induced component and the observed alignment error, for example, by subtracting the curvature-induced component from the observed alignment error. Flow then moves to block B1145, where bonding system 100 adjusts the in-plane position of test chiplet 22 based on the in-plane alignment error.

[0113] Next, in block B1150, bonding system 100 measures the tilt of product substrate 29 at the bonding site where test chiplet 22 is bonded. Next, in block B1155, bonding system 100 adjusts the global tilt of test chiplet 22 to match the tilt of product substrate 29 at the bonding site.

[0114] Then, in block B1160, bonding system 100 bonds test chiplets 22 to product substrate 29.

[0115] 10 omits block B1015, and some embodiments of FIG. 11 omit block B1115. In such embodiments (block B1045 in FIG. 10 and block B1140 in FIG. 11), the calculation of the in-plane alignment error is further based on the global tilt of the test chiplet 22. For example, in some embodiments, the relationship between the in-plane alignment error, the observed alignment error, the curvature-induced component, and the global tilt is described by the following:

[0116]

number

[0117] where err ip is the in-plane alignment error, and err ob where σ is the observed alignment error, cc is the curvature-induced component, α is the global tilt of the test chiplet 22, and r is the distance from the center of the test chiplet 22 to the test area 23. Additionally, the magnification or scale of the field of view of the test image 43 of the test area 23 can be adjusted, for example, as described by the following:

[0118]

number

[0119] Here, FOV new is the adjusted field of view of the test image 43, and FOV old is the field of view of the captured test image 43 and α is the global tilt of the test chiplet 22.

[0120] 12 shows an example embodiment of an operational flow for bonding chiplets to a substrate. The flow begins in block B1200, where bonding system 100 identifies bonding sites on product substrate 29 held by product chuck 134. For example, downward-looking microscope 142 may capture an image of product substrate 29, and controller 150 may identify one or more bonding sites on product substrate 29 in the image.

[0121] Next, in block B1205, the bonding system 100 uses one or more upward-looking microscopes 141 to acquire individual test images 43 of one or more test areas 23 of the test chiplet 22 held by the bonding head 133.

[0122] Flow then moves to block B1210, where bonding system 100 acquires shape measurements (e.g., surface profile) of one or more test areas 23 on test chiplet 22. For example, one or more shape measurement sensors 143 may acquire individual measurements of the flatness or shape of each of one or more test areas 23 on test chiplet 22.

[0123] Next, in block B1215, bonding system 100 determines the alignment of test chiplet 22 relative to the bonding site on product substrate 29. For example, controller 150 and face-down alignment system 139 can measure the position of the substrate set of interconnect contacts on product substrate 29 relative to product chuck 134, and controller 150 and face-up alignment system 138 can measure the position of the tiplet set of interconnect contacts on test chiplet 22 relative to transfer chuck 131. Then, because product chuck 134 and test chuck 131 are aligned with one another, controller 150 can determine the alignment of test chiplet 22 relative to the bonding site. The output of block B1215 indicates the relative position of the substrate set of interconnect contacts on product substrate 29 relative to the tiplet set of interconnect contacts on test chiplet 22 for the current alignment of product chuck 134 and bonding head 133 holding test chiplet 22.

[0124] In block B1220, the bonding system 100 determines an observed alignment error in the current alignment between the product chuck 134 and the bonding head 133 holding the test chiplet 22 based on the relative position of the substrate set of interconnect contacts on the product substrate 29 to the chiplet set of interconnect contacts on the test chiplet 22.

[0125] The flow then proceeds to block B1225, where the bonding system 100 calculates the curvature-induced component of the observed alignment error based on shape measurements of one or more test areas 23 of the test chiplet 22 and a reference surface (e.g., a plane) used as a reference point for determining non-flatness variations (e.g., slope) on the surface of the test chiplet 22 in one or more test areas 23.

[0126] Then, in block B1230, bonding system 100 calculates an in-plane alignment error based on the curvature-induced component and the observed alignment error, for example, by subtracting the curvature-induced component from the observed alignment error. The flow then moves to block B1235, where bonding system 100 adjusts the in-plane position of test chiplet 22 based on the in-plane alignment error, for example, by controlling and moving (rotating or translating) bonding head chiplet stage 1332 of bonding head 133 holding test chiplet 22. Then, in block B1240, bonding system 100 bonds test chiplet 22 to a bonding site on product substrate 29.

[0127] 13 illustrates an exemplary embodiment of an operational flow for bonding chiplets to a substrate. The flow begins in block B1300, where bonding system 100 identifies bonding sites on product substrate 29 held by product chuck 134.

[0128] Next, in block B1305, the bonding system 100 uses two or more alignment microscopes 145 to acquire individual 3D test images 43 of one or more test areas 23 of the test chiplet 22 held by the bonding head 133.

[0129] Next, in block B1310, bonding system 100 determines the alignment of test chiplets 22 relative to bond sites on product substrate 29. The output of block B1310 indicates the relative positions of the substrate sets of interconnect contacts on product substrate 29 relative to the chiplet sets of interconnect contacts on test chiplets 22 for the current alignment of product chuck 134 and bonding head 133 holding test chiplets 22.

[0130] In block B1315, the bonding system 100 determines an observed alignment error based on the relative position of the substrate set of interconnect contacts on the product substrate 29 to the chiplet set of interconnect contacts on the test chiplet 22 in the current alignment between the product chuck 134 and the bonding head 133 holding the test chiplet 22.

[0131] The flow then proceeds to block B1320, where the bonding system 100 calculates the curvature-induced component of the observed alignment error based on a 3D image of one or more test areas 23 of the test chiplet 22 and a reference surface (e.g., a plane) used as a reference point for determining the variation in non-flatness (e.g., slope) on the surface of the test chiplet 22 in the one or more test areas 23.

[0132] Then, in block B1325, bonding system 100 calculates an in-plane alignment error based on the curvature-induced component and the observed alignment error, for example, by subtracting the curvature-induced component from the observed alignment error. Flow then moves to block B1330, where bonding system 100 adjusts the in-plane position of test chiplet 22 based on the in-plane alignment error, for example, by controlling and moving (rotating or translating) bonding head chiplet stage 1332 of bonding head 133 holding test chiplet 22. Then, in block B1335, bonding system 100 bonds test chiplet 22 to product substrate 29.

[0133] 14 illustrates an exemplary embodiment of an operational flow for bonding chiplets to a substrate. The flow begins in block B1400, where bonding system 100 identifies bonding sites on product substrate 29 held by product chuck 134.

[0134] Next, in block B1405, bonding system 100 measures the tilt of product substrate 29 at the bonding site. The tilt is measured relative to a reference plane (for example, the xy plane).

[0135] At block B1410, bonding system 100 obtains a measurement of the global tilt of test chiplet 22 held by bonding head 133 relative to a reference plane. Flow then moves to block B1415, where bonding system 100 adjusts the tilt of test chiplet 22 based on the global tilt of test chiplet 22 and the tilt at the bonding site on product substrate 29. The tilt of test chiplet 22 is adjusted to minimize the tilt difference between the global tilt of test chiplet 22 and the tilt at the bonding site on product substrate 29.

[0136] Next, in block B1420, bonding system 100 aligns tilt-adjusted test chiplet 22 to the bonding site. In block B1425, bonding system 100 acquires test image 43 of test area 23 of tilt-adjusted test chiplet 22.

[0137] Then, in block B1430, bonding system 100 acquires shape measurements of test area 23. For example, bonding system 100 may measure individual z-axis coordinates of multiple locations (e.g., pixels) in test area 23. Additionally, in some embodiments, bonding system 100 acquires a 3D test image 43 of test area 23 in block B1425, and block B1430 is omitted in such embodiments because 3D test image 43 includes shape measurements of test area 23.

[0138] Next, in block B1435, bonding system 100 obtains (e.g., calculates) an observed alignment error between reference image 42 and test image 43. Block B1435 may include determining the relative position of the substrate set of interconnect contacts on product substrate 29 to the chiplet set of interconnect contacts on tilt-adjusted test chiplet 22 in the current alignment of product chuck 134 and bonding head 133 holding tilt-adjusted test chiplet 22. Also, in block B1440, bonding system 100 calculates a curvature-induced component of the observed alignment error based on shape measurements of test area 23 and a reference surface (e.g., a flat surface) used as a reference point for determining non-flatness variations on the surface of test chiplet 22 in test area 23.

[0139] Then, in block B1445, bonding system 100 calculates an in-plane alignment error based on the curvature-induced component and the observed alignment error, for example, by subtracting the curvature-induced component from the observed alignment error. Next, flow moves to block B1450, where bonding system 100 adjusts the in-plane position of test chiplet 22 based on the in-plane alignment error. Then, in block B1450, bonding system 100 bonds test chiplet 22 to a bonding site on product substrate 29.

[0140] 15 is a schematic diagram of an exemplary embodiment of a controller 150. The controller 150 includes one or more processors 151, one or more computer-readable storage media 152, one or more I / O components 153, and a bus 154.

[0141] The controller 150 further includes a system control module 1521, a communications module 1522, an image acquisition module 1523, a shape acquisition module 1524, and an alignment error calculation module 1525. As used herein, a module includes logic, computer-readable data, or computer-executable instructions. In the embodiment shown in FIG. 15 , the modules are implemented in software (e.g., Assembly, C, C++, C#, Java, JavaScript, BASIC, Perl, Visual Basic, Python, PHP). However, in some embodiments, the modules are implemented in hardware (e.g., customized circuitry) or, alternatively, a combination of software and hardware. When a module is implemented at least partially in software, the software is stored on one or more computer-readable storage media 152. Also, in some embodiments, the controller 150 includes additional or fewer modules, the modules are combined into fewer modules, or the modules are divided into more modules. Modules can use (e.g., call) other modules. Controller 150 also includes a data repository 1526 that stores information about the chiplets and substrates, such as the number of chiplets on the substrate, the number of bonding sites on the substrate, and the shape and location of alignment marks on the chiplets and substrate.

[0142] System control module 1521 includes instructions that cause applicable components of controller 150 (e.g., one or more processors 151, storage medium 152, I / O components 153) to communicate with and control other components of bonding system 100. The applicable components operating in accordance with system control module 1521 implement an example of a system controller.

[0143] Communications module 1522 includes instructions that cause applicable components of control device 150 (e.g., one or more processors 151, storage medium 152, I / O components 153) to communicate with one or more other devices, such as other computing devices (e.g., network computer 160 of FIG. 1) and input or output devices (e.g., display device 155 and keyboard 156 of FIG. 1). Together, the applicable components operating in accordance with communications module 1522 implement an example of a communications unit.

[0144] Image acquisition module 1523 includes instructions that cause applicable components of control device 150 (e.g., one or more processors 151, storage medium 152, I / O components 153) to control components of bonding system 100 to capture images of test chiplets 22 (e.g., images of test area 23, global images) and images of reference chiplets 33 (e.g., images of reference area 34, global images). For example, some embodiments of image acquisition module 1523 include instructions that cause applicable components of control device 150 to cause bonding system 100 to perform at least some of the operations set forth in blocks B420 and B440 of FIG. 4, blocks B905 and B910 of FIG. 9, blocks B1000 and B1025 of FIG. 10, blocks B1100-B1105 and B1125 of FIG. 11, block B1205 of FIG. 12, block B1305 of FIG. 13, and block B1425 of FIG. 14. Thus, the applicable components operating in accordance with image acquisition module 1523 implement an example of an image acquisition unit.

[0145] Shape acquisition module 1524 includes instructions that cause applicable components of control device 150 (e.g., one or more processors 151, storage medium 152, I / O components 153) to control components of bonding system 100 to obtain shape and tilt measurements of test chiplets 22, reference chiplets 33, and bonding sites on product substrate 29. This may include obtaining information from one or more shape measurement sensors 143, or generating 3D images from images taken using two or more upward-looking microscopes 141 that comprise a stereo vision system, or from two or more alignment microscopes 145 that comprise a stereo vision system. For example, some embodiments of geometry acquisition module 1524 include instructions that cause applicable components of controller 150 to control bonding system 100 to perform at least some of the operations set forth in blocks B425 and B445 of Figure 4, blocks B905 and B910 of Figure 9, blocks B1005, B1010, B1030, and B1055 of Figure 10, blocks B1105 and B1110 of Figure 11, block B1210 of Figure 12, block B1305 of Figure 13, and blocks B1405-B1410 and B1430 of Figure 14. The applicable components operating in accordance with geometry acquisition module 1524 implement an example of a geometry acquisition unit.

[0146] Alignment error calculation module 1525 includes instructions that cause applicable components of controller 150 (e.g., one or more processors 151, storage media 152, I / O components 153) to control components of bonding system 100 to calculate observed alignment errors, calculate curvature-induced components and in-plane alignment errors from the observed alignment errors, adjust the in-plane position of test chiplet 22, and adjust the global tilt of test chiplet 22. For example, some embodiments of alignment error calculation module 1525 include instructions that cause applicable components of controller 150 to control bonding system 100 to perform at least some of the operations set forth in blocks B450-B470 of Figure 4, blocks B915-B930 of Figure 9, blocks B1035-B1050 and B1060 of Figure 10, blocks B1110-B1115, B1130-B1145 and B1155 of Figure 11, blocks B1220-B1235 of Figure 12, blocks B1315-B1330 of Figure 13, and blocks B1415 and B1435-B1450 of Figure 14. The applicable components operating in accordance with alignment error calculation module 1525 then implement an example of an alignment error calculation unit.

[0147] At least some of the devices, systems, and methods described above may be implemented, at least in part, by providing one or more computer-readable media containing computer-executable instructions for performing the operations described above to one or more computing devices configured to read and execute the computer-executable instructions. The systems or devices perform the operations of the described embodiments when executing the computer-executable instructions. Additionally, an operating system on one or more systems or devices may implement at least some of the operations of the described embodiments.

[0148] Furthermore, some embodiments implement the above-described devices, systems, and methods using one or more functional units, which may be implemented solely in hardware (e.g., customized circuitry) or in a combination of software and hardware (e.g., a microprocessor executing software).

[0149] In the description, specific details are set forth to provide a thorough understanding of the disclosed embodiments. However, well-known methods, procedures, components, and circuits may not be described in detail to avoid unnecessarily lengthening the disclosure.

[0150] Additionally, when an element (e.g., an element, part, or component) is referred to herein as being "on," "on," "connected to," or "coupled to" another element, the element may be directly on, on, connected to, or coupled to the other element, but intervening elements may exist between the element and the other element. In contrast, when an element is referred to as being "directly on," "directly on," "directly connected to," or "directly coupled to" another element, there are no intervening elements between the element and the other element.

[0151] Furthermore, the terms "comprising," "having," "includes," "including," and "containing" are to be construed as open-ended terms unless otherwise specified. Thus, when these terms are used herein, they specify the presence of stated features, integers, steps, operations, elements, materials, or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, materials, or components not expressly stated.

Claims

1. 1. A method for aligning a chiplet to a substrate, comprising: acquiring an image of an area on a chiplet; measuring deviation from planarity of the region on the chiplet; estimating a tiplet position error for the chiplet based on at least the image of the region and the deviation from planarity of the region; adjusting a relative position between the chiplet and the substrate based on the chiplet position error; 10. A method comprising:

2. The method of claim 1 , wherein estimating the tiplet position error of the chiplet is further based on a reference plane.

3. The method of claim 2 , wherein the reference surface is an image of a region of a reference chiplet.

4. estimating the chiplet position error calculating an alignment difference between the image of the area on the chiplet and the reference surface; The alignment difference includes one or both of a translation difference and a rotation difference.

3. The method of claim 2.

5. estimating the chiplet position error adjusting the alignment difference based on the deviation from planarity of the region; 5. The method of claim 4, comprising:

6. 6. The method of claim 5, wherein adjusting the alignment difference based on the deviation from planarity of the region comprises subtracting a component of the alignment difference attributable to the deviation from planarity of the region.

7. The method of claim 1 , wherein the chiplet is held by a bonding head.

8. 1. A system comprising: one or more processors; one or more memories; and The one or more processors and the one or more memories acquiring an image of an area on the chiplet; measuring deviation from planarity of the region on the chiplet; calculating a tiplet position error for the chiplet based on at least the image of the region and the deviation from planarity of the region; adjusting a relative position between the chiplet and a substrate based on the chiplet position error; The system is configured as follows.

9. 10. The system of claim 8, wherein the tiplet position error of the chiplet is calculated further based on a reference plane.

10. The system of claim 9 , wherein the reference surface is an image of an area of ​​a reference chiplet.

11. To calculate the tiplet position error, the one or more processors and the one or more memories: calculating an alignment difference between the image of the region on the chiplet and the reference surface; It is configured as follows: The alignment difference includes one or both of a translation difference and a rotation difference.

10. The system of claim 9.

12. To calculate the tiplet position error, the one or more processors and the one or more memories: adjusting the alignment difference based on the deviation from planarity of the region.

12. The system of claim 11, further configured to:

13. 13. The system of claim 12, wherein to adjust the alignment difference based on the deviation from planarity of the region, the one or more processors and the one or more memories are further configured to subtract a component of the alignment difference that is attributable to the deviation from planarity of the region.

14. Further comprising a bonding head; the chiplet is held by the bonding head; 9. The system of claim 8.

15. One or more computer-readable storage media storing instructions for execution by one or more computing devices, the operations causing the one or more computing devices to perform include: acquiring an image of an area on a chiplet; measuring the surface relief of the region on the chiplet; calculating a tiplet position error for the chiplet based on at least the image of the region, the surface relief of the region, and a reference surface; calculating a position adjustment value for the chiplet based on the chiplet position error; 1. One or more computer-readable storage media, comprising:

16. The operation is controlling the bonding head to move in accordance with the position adjustment value; 16. The one or more computer-readable storage media of claim 15, further comprising:

17. 16. The one or more computer-readable storage media of claim 15, wherein calculating the position adjustment value for the chiplet includes calculating a surface relief component of the tiplet position error based on the surface relief of the region, the surface relief component resulting from a difference between the surface relief of the region and the reference surface.

18. Calculating the position adjustment value of the chiplet includes: calculating in-plane translational and rotational differences of the region; calculating the in-plane translational and rotational differences of the region includes subtracting the surface relief component from the tiplet position error.

20. One or more computer-readable storage media as recited in claim 17.

19. 20. The one or more computer-readable storage media of claim 18, wherein the in-plane translation difference and the in-plane rotation difference define the alignment value.

20. 20. The one or more computer-readable storage media of claim 17, wherein the reference surface is a surface of a reference chiplet or an ideal plane.

Images