Joining method and joining apparatus

Abstract

A joining method for joining a substrate (W1) and a substrate (W2) that is vibrating relative to the substrate (W1), the method comprising: a vibration waveform identification step for identifying, from a time course of a displacement amount, a vibration waveform of the vibration of the substrate (W2) relative to the substrate (W1); a timing estimation step for estimating a target timing at which the displacement amount of the substrate (W2) relative to the substrate (W1) becomes a target amount, on the basis of the vibration waveform; and a contact step for bringing the substrate (W2) into contact with the substrate (W1), on the basis of the estimated target timing.

Description

Bonding method and bonding device

[0001] The present invention relates to a joining method and a joining device.

[0002] A bonding device has been proposed in which, while one of two objects to be bonded is held on a stage and the other is held on a head, the amount of positional deviation between the two objects is measured, the objects are aligned based on the amount of positional deviation, and then the objects are bonded together (see, for example, Patent Document 1).

[0003] JP 2011-066287 A

[0004] However, depending on the installation location of the bonding device, vibrations may occur in the bonding device. In this case, in the bonding device described in Patent Document 1, the head vibrates relative to the stage, and there is a risk that the amount of relative positional deviation between the two workpieces cannot be measured with high accuracy. If the measurement accuracy of the amount of positional deviation between the two workpieces is low, it becomes difficult to bond the two workpieces with high positional accuracy.

[0005] The present invention has been made in view of the above-mentioned circumstances, and has an object to provide a joining method and a joining device that can join objects to be joined with high positional accuracy.

[0006] In order to achieve the above object, the bonding method according to the present invention is a bonding method for bonding a first object to a second object vibrating relatively to the first object, and includes: a vibration waveform specifying step of specifying a vibration waveform of the second object vibrating relative to the first object from a time progression of a positional deviation amount of the second object relative to the first object; a timing estimation step of estimating a target timing at which the positional deviation amount of the second object relative to the first object will reach a target amount based on the vibration waveform; and a contacting step of contacting the second object with the first object based on the estimated target timing.

[0007] A bonding device according to another aspect of the present invention is a bonding device for bonding a first object to be welded and a second object to be welded, comprising: a first object to be welded holding part for holding the first object to be welded; a second object to be welded holding part which vibrates relative to the first object to be welded holding part and holds the second object to be welded; and a holder drive part which moves at least one of the first object to be welded holding part and the second object to be welded holding part in a first direction in which the first object to be welded holding part and the second object to be welded holding part approach each other or in a second direction in which the first object to be welded holding part and the second object to be welded holding part move away from each other. a control unit that repeatedly measures, at predetermined time intervals while the first object and the second object are spaced apart, an amount of misalignment of the first object relative to the second object in the first direction and a direction perpendicular to the second direction, identifies a vibration waveform of the vibration of the second object relative to the first object from the time progression of the amount of misalignment, estimates a target timing at which the amount of misalignment of the second object relative to the first object will reach a target amount based on the vibration waveform, and controls the holding unit drive unit to move at least one of the first object holding unit and the second object holding unit in the first direction based on the estimated target timing, so as to bring the second object into contact with the first object.

[0008] According to the present invention, the vibration waveform of the vibration of the second object relative to the first object is identified from the time transition of the positional deviation of the second object relative to the first object, and the target timing at which the positional deviation of the second object relative to the first object will reach the target amount is estimated based on the identified vibration waveform. Then, the second object is brought into contact with the first object based on the estimated target timing. This makes it possible to join the second object to the first object with high positional accuracy even when the second object is vibrating relative to the first object.

[0009] 1 is a schematic configuration diagram of a bonding apparatus according to a first embodiment of the present invention; FIG. 1 is a schematic perspective view showing the vicinity of a stage and a head according to the first embodiment; FIG. 2 is a diagram explaining a method for fine-tuning a head according to the first embodiment; FIG. 3 is a schematic plan view of a stage and a head according to the first embodiment; FIG. 4 is a schematic cross-sectional view of a stage and a head according to the first embodiment; FIG. 5 is a diagram showing two alignment marks provided on one of two substrates to be bonded; FIG. 6 is a diagram showing two alignment marks provided on the other of two substrates to be bonded; FIG. 7 is a schematic view showing a captured image of alignment marks; FIG. 8 is a schematic view showing a state in which alignment marks are misaligned with each other; FIG. 9 is a schematic view of a part of a bonding apparatus according to the first embodiment; FIG. 10 is a flowchart showing the flow of a bonding method performed by the bonding apparatus according to the first embodiment; FIG. 11 is a schematic cross-sectional view showing a state in which substrates are held by a stage and a head according to the first embodiment; FIG. 12 is a diagram showing the transition of the amount of misalignment of the head relative to the stage according to the first embodiment; FIG. 13 is a schematic cross-sectional view showing a state in which central portions of bonding surfaces of substrates held by a stage and a head according to the first embodiment are brought into contact with each other; FIG. 14 is a schematic cross-sectional view showing a state in which substrates held by a stage and a head according to the first embodiment are brought close to each other; FIG. 15 is a schematic cross-sectional view showing a state in which peripheral portions of bonding surfaces of substrates held by a stage and a head according to the first embodiment are brought into contact with each other. FIG. 1 is a schematic cross-sectional view showing a state in which the head according to embodiment 1 is removed from the stage. FIG. 2 is a diagram showing the frequency dependency of the vibration transmissibility of the pedestal according to embodiment 1. FIG. 3 is a diagram showing an example of the frequency spectrum of the vibration amplitude when the bonding apparatus is placed on a pedestal having a vibration isolation function and when the bonding apparatus is placed on the floor, with respect to vibrations transmitted to the bonding apparatus according to embodiment 1. FIG. 4 is a schematic diagram of a part of the bonding apparatus according to comparative example 1. FIG. 5 is a schematic configuration diagram of a chip bonding system according to embodiment 2 of the present invention. FIG. 6 is a plan view of a chip holding unit according to embodiment 2. FIG. 7 is a cross-sectional view showing a part of the chip transport apparatus according to embodiment 2. FIG. 8 is a schematic configuration diagram of the bonding apparatus according to embodiment 2. FIG. 9 is a cross-sectional view of a head according to embodiment 2. FIG. 10 is a plan view of the head according to embodiment 2. FIG. 11 is a diagram showing the positional relationship between the alignment mark of the chip and the hollow part of the head according to embodiment 2. FIG. 12 is a schematic perspective view showing a part of the bonding unit according to embodiment 2.16 of a bonding apparatus according to a second embodiment. FIG. 17 is a cross-sectional view taken along the line A-A in FIG. 16 of the bonding apparatus according to the second embodiment. FIG. 18 is a view showing alignment marks provided on a chip. FIG. 19 is a view showing alignment marks provided on a substrate. FIG. 20 is a view showing relative positional deviation of alignment marks. FIG. 21 is a view showing details of a head according to the second embodiment. FIG. 22 is a plan view of a stage unit according to the second embodiment. FIG. 23 is a side view of a stage unit according to the second embodiment. FIG. 24 is a schematic side view showing how chips are supplied from a chip supply unit in the chip bonding system according to the second embodiment. FIG. 25 is a schematic side view showing how chips are delivered from a chip transport device to a head in the chip bonding system according to the second embodiment. FIG. 26 is a flowchart showing an example of the flow of a chip bonding process executed by the chip bonding system according to the second embodiment. FIG. 27 is a view for explaining the operation of the chip bonding system according to the second embodiment, showing a state in which the chip is separated from the substrate. FIG. 28 is a view for explaining the operation of the chip bonding system according to the second embodiment, showing a state in which the chip is in contact with the substrate. FIG. 29 is a view for explaining the operation of the chip bonding system according to the second embodiment, showing how the chip is being released from the substrate. FIG. 29 is a view for explaining the operation of the chip bonding system according to the second embodiment, showing how one set of two sets of alignment marks is imaged. FIG. 10 is a diagram for explaining the operation of the chip bonding system according to Comparative Example 2, and is a diagram showing how the other set is imaged. FIG. 11 is a cross-sectional view of a head according to a modified example. FIG. 12 is a plan view of a head according to a modified example. FIG. 13 is a diagram for explaining the operation of a head according to a modified example. FIG. 14 is a cross-sectional view of a head according to a modified example.

[0010] (Embodiment 1) A bonding apparatus according to an embodiment of the present invention will be described below with reference to the drawings. The bonding apparatus according to this embodiment bonds substrates W1 and W2 together by bringing the bonding surfaces of the activated substrates into contact with each other. As shown in FIG. 1 , the bonding apparatus 1 according to this embodiment includes a chamber 120, a stage 401, a head 402, a stage driver 403, a head driver 404, substrate heaters 481 and 482, an imaging unit 500, and a vibration isolation unit 160. Generally, vibrations transmitted through the floor F and vibrations generated in the bonding apparatus are transmitted to at least one of the head 402 and the stage 401, causing the head 402 holding the substrate W2 and the stage 401 holding the substrate W1 to vibrate relative to each other. This causes the substrates held by the head 402 and the stage 401 to vibrate relative to each other. In contrast, the bonding apparatus 1 according to this embodiment reduces the influence of the vibrations on the positional misalignment between the substrates W1 and W2 held by the head 402 and the stage 401 when bonding the substrates W1 and W2. Examples of the substrates W1 and W2 include Si substrates, glass substrates, and sapphire substrates. The bonding apparatus 1 also includes a distance measurement unit 490 that measures the distance between the stage 401 and the head 402. The bonding apparatus 1 is supported by a stand 41 placed on the floor F where the bonding apparatus 1 is installed. The stand 41 has a vibration-isolating function that suppresses transmission of vibrations transmitted through the floor F to the bonding apparatus 1, and is, for example, an earthquake-resistant stand with a grating installed vertically above it. In the following description, the ±Z direction in FIG. 1 will be referred to as the up-down direction, and the XY direction as the horizontal direction.

[0011] The chamber 120 maintains the area where the substrates W1 and W2 are placed at a vacuum level equal to or higher than a predetermined reference vacuum level. The chamber 120 is connected to a vacuum pump 121a via an exhaust pipe 121b and an exhaust valve 121c. When the exhaust valve 121c is opened and the vacuum pump 121a is operated, gas within the chamber 120 is exhausted to the outside of the chamber 120 through the exhaust pipe 121b, maintaining a reduced pressure atmosphere within the chamber 120. The air pressure (vacuum level) within the chamber 120 can be adjusted by adjusting the amount of exhaust by varying the opening / closing amount of the exhaust valve 121c. A window 120a is provided in part of the chamber 120 and is used to measure the relative positions of the substrates W1 and W2 using the imaging unit 500. The air pressure within the chamber 120 can be set within a range of 1 Pa to 1000 Pa.

[0012] The stage driver 403 is a holder driver that can move the stage 401 in the X and Y directions and rotate it around the Z axis. The head driver 404 includes an elevation driver 406 that raises and lowers the head 402 vertically upward or downward (see arrow AR1 in FIG. 1 ), an XY direction driver 405 that moves the head 402 in the X and Y directions, and a rotation driver 407 that rotates the head 402 in a rotational direction around the Z axis (see arrow AR2 in FIG. 3 ). The XY direction driver 405 and the rotation driver 407 constitute a holder driver that moves the head 402 in directions perpendicular to the vertical direction (X and Y directions, rotational direction around the Z axis). The head driver 404 also includes a piezo actuator 411 that adjusts the tilt of the head 402 relative to the stage 401 and a pressure sensor 412 that measures the pressure applied to the head 402. The XY direction drive unit 405 and the rotation drive unit 407 move the head 402 relative to the stage 401 in the X direction, the Y direction, and the rotation direction around the Z axis, thereby enabling alignment of the substrate W1 held on the stage 401 and the substrate W2 held on the head 402.

[0013] The lifting / lowering driver 406 moves the head 402 in the vertical direction to bring the stage 401 and the head 402 closer to each other or move the head 402 away from the stage 401. When the lifting / lowering driver 406 moves the head 402 vertically downward, the substrate W1 held on the stage 401 and the substrate W2 held on the head 402 come into contact with each other. When the lifting / lowering driver 406 applies a driving force to the head 402 in a direction that moves it closer to the stage 401 while the substrates W1 and W2 are in contact with each other, the substrate W2 is pressed against the substrate W1. The lifting / lowering driver 406 is also provided with a pressure sensor 408 that measures the driving force that the lifting / lowering driver 406 applies to the head 402 in a direction that moves it closer to the stage 401. From the measurement value of the pressure sensor 408, the pressure acting on the bonding surface of the substrates W1 and W2 when the lifting / lowering driver 406 presses the substrate W2 against the substrate W1 can be detected. The pressure sensor 408 is composed of, for example, a load cell.

[0014] As shown in FIG. 2A , there are three piezo actuators 411 and three pressure sensors 412. The three piezo actuators 411 and three pressure sensors 412 are disposed between the head 402 and the XY-direction drive unit 405. The three piezo actuators 411 are attitude adjustment units fixed at three non-collinear positions on the top surface of the head 402, i.e., at three positions arranged at approximately equal intervals around the periphery of the top surface of the head 402, which is generally circular in plan view. The three pressure sensors 412 connect the upper ends of the piezo actuators 411 to the lower surface of the XY-direction drive unit 405. Each of the three piezo actuators 411 can extend and retract in the vertical direction. The extension and retraction of the three piezo actuators 411 finely adjusts the tilt of the head 402 around the X-axis and the Y-axis and the vertical position of the head 402. For example, if the head 402 is tilted with respect to the stage 401 as shown by the dashed line in Fig. 2B , the lower surface of the head 402 can be made substantially parallel to the upper surface of the stage 401 by extending one of the three piezo actuators 411 (see arrow AR3 in Fig. 2B ) to finely adjust the attitude of the head 402. Furthermore, the three pressure sensors 412 measure the pressure forces at three positions on the lower surface of the head 402. Then, by driving each of the three piezo actuators 411 so that the pressure forces measured by the three pressure sensors 412 are equal, the substrates W1 and W2 can be brought into contact with each other while maintaining the lower surface of the head 402 and the upper surface of the stage 401 substantially parallel to each other.

[0015] The stage 401 and the head 402 are disposed in the chamber 120 so as to face each other in the vertical direction and so that the stage 401 is positioned vertically below the head 402. The stage 401 is a first object holder that supports the substrate W1 on its upper surface 401a, and the head 402 is a second object holder that supports the substrate W2 on its lower surface 402a. Here, the stage 401 supports the substrate W1 with its upper surface 401a in surface contact with the entire substrate W1, and the head 402 supports the substrate W2 with its lower surface 402a in surface contact with the entire substrate W2. The stage 401 and the head 402 are formed from a light-transmitting material such as light-transmitting glass. 3A and 3B, the stage 401 and the head 402 are provided with electrostatic chucks 441 and 442 for holding the substrates W1 and W2, a pressing mechanism 431 for pressing the central portion of the substrate W1, and a pressing mechanism 432 for pressing the central portion of the substrate W2. The electrostatic chucks 441 and 442 hold the peripheral portions of the substrates W1 and W2. Furthermore, the center portions of the stage 401 and the head 402 are provided with through-holes 401b and 402b that are circular in plan view.

[0016] The electrostatic chucks 441, 442 are provided in a first region A1 of the stage 401 and the head 402 facing the peripheries of the substrates W1, W2 when the substrates W1, W2 are supported by the stage 401 and the head 402. Each of the electrostatic chucks 441, 442 has an annular shape and includes a terminal electrode arranged along the circumferential direction outside the second region A2 inside the first region A1 of the stage 401 and the head 402, and a plurality of linear electrode elements electrically connected to the terminal electrode at their base ends. The terminal electrode and the plurality of electrode elements are formed of a transparent conductive film containing a transparent conductive material such as ITO. The electrostatic chucks 441, 442 attract and hold the substrates W1, W2 when a voltage is applied by a chuck driver (not shown).

[0017] Furthermore, recesses 401c and 402c are provided in the second area A2 of the stage 401 and the head 402. The depth of the recesses 401c and 402c is set to a depth such that the bottoms of the recesses 401c and 402c do not come into contact with the substrates W1 and W2 when the substrates W1 and W2 are held therein, and is set to, for example, 1 μm or more.

[0018] 3B , the pressing mechanism 431 is provided in the center of the stage 401, and the pressing mechanism 432 is provided in the center of the head 402. The pressing mechanism 431 has a pressing unit 431a that can be extended and retracted toward the head 402 through a through-hole 401b in the stage 401, and a pressing drive unit 431b that drives the pressing unit 431a. The pressing mechanism 431 also has a stopper 431c that prevents the pressing unit 431a from moving beyond a preset immersion amount. The pressing mechanism 432 has a pressing unit 432a that can be extended and retracted toward the stage 401 through a through-hole 402b in the head 402, and a pressing drive unit 432b that drives the pressing unit 432a. The pressing mechanism 432 also has a stopper 432c that prevents the pressing unit 432a from moving beyond a preset immersion amount. The pressing drive unit 431b and the pressing drive unit 432b have, for example, voice coil motors. Furthermore, the pressing unit 431a and the pressing unit 432a perform either pressure control, which controls the pressure applied to the substrates W1, W2 to be kept constant, or position control, which controls the contact position of the substrates W1, W2 to be kept constant. For example, by position-controlling the pressing unit 431a and pressure-controlling the pressing unit 432a, the substrates W1, W2 are pressed at a constant position with a constant pressure.

[0019] Returning to FIG. 1 , the distance measurement unit 490 is, for example, a laser rangefinder, and measures the distance between the stage 401 and the head 402 without contacting the stage 401 or the head 402. The distance measurement unit 490 measures the distance between the stage 401 and the head 402 from the difference between the light reflected from the upper surface of the stage 401 and the light reflected from the lower surface of the head 402 when a laser beam is irradiated from above the head 402, which is made of a light-transmitting material, toward the stage 401. As shown in FIG. 2A , the distance measurement unit 490 measures the distance between three positions P11, P12, and P13 on the upper surface of the stage 401 and three positions P21, P22, and P23 on the lower surface of the head 402 that face the positions P11, P12, and P13 in the Z direction. As described above, because the head 402 or the stage 401 is made of a light-transmitting material, it is possible to measure the distance by transmitting a laser beam through the head 402 or the stage 401. Therefore, there is an advantage that the distance measurement unit 490 can be configured to be disposed on the side of the head 402 and stage 401 opposite to the side of the substrates W1 and W2.

[0020] 1 , the imaging unit 500 includes imaging sections 501 and 502 and mirrors 504 and 505. The imaging sections 501 and 502 are disposed on the side of the stage 401 opposite to the side where the substrate W1 is held. The imaging sections 501 and 502 each include an imaging element (not shown) and a coaxial illumination system (not shown). As a light source for the coaxial illumination system, a light source that emits light (e.g., infrared light) that transmits through the substrates W1 and W2, the stage 401, and a window 120 a provided in the chamber 120 is used.

[0021] 4A and 4B, for example, two alignment marks (first alignment marks) MK1a and MK1b are provided on the substrate W1, and two alignment marks (second alignment marks) MK2a and MK2b are provided on the substrate W2. The bonding apparatus 1 performs an alignment operation (alignment operation) for the two substrates W1 and W2 while recognizing the positions of the alignment marks MK1a, MK1b, MK2a, and MK2b provided on the substrates W1 and W2 imaged by the imaging unit 500. More specifically, the bonding apparatus 1 first performs a rough alignment operation (rough alignment operation) for the substrates W1 and W2 to face each other while recognizing the alignment marks MK1a, MK1b, MK2a, and MK2b provided on the substrates W1 and W2 imaged by the imaging unit 500. Thereafter, the bonding apparatus 1 performs an even more precise alignment operation (fine alignment operation) while simultaneously recognizing the alignment marks MK1a, MK2a, MK1b, and MK2b provided on the two substrates W1 and W2 imaged by the imaging unit 500.

[0022] 1, light emitted from the light source of the coaxial illumination system of the imaging unit 501 is reflected by the mirror 504 and travels upward, and then passes through the window 120a and part or all of the substrates W1 and W2. The light that has passed through part or all of the substrates W1 and W2 is reflected by the alignment marks MK1a and MK2a of the substrates W1 and W2, travels downward, passes through the window 120a, and is reflected by the mirror 504 to enter the image sensor of the imaging unit 501. Meanwhile, light emitted from the light source of the coaxial illumination system of the imaging unit 502 is reflected by the mirror 505 and travels upward, and then passes through the window 120a and part or all of the substrates W1 and W2. Light that has transmitted through part or all of the substrates W1, W2 is reflected by the alignment marks MK1a, MK2a on the substrates W1, W2, travels downward, passes through the window 120a, is reflected by the mirror 505, and is incident on the image sensor of the image sensor 502. Then, the image sensors 501, 502 of the image sensor 500 each use the light incident on their image sensors to simultaneously capture, within one field of view, a captured image GAa including the alignment marks MK1a, MK2a on the two substrates W1, W2, and a captured image GAb including the alignment marks MK1b, MK2b on the two substrates W1, W2, as shown in Figures 5A and 5B. The image sensor 501 captures the image GAa, and the image sensor 502 captures the image GAb, simultaneously.

[0023] Returning to FIG. 1 , the substrate heating units 481 and 482 are, for example, electric heaters, and are provided on the stage 401 and head 402, respectively, as shown in FIG. 3B . The substrate heating units 481 and 482 heat the substrates W1 and W2 held on the stage 401 and head 402 by transferring heat to the substrates W1 and W2. The temperatures of the substrates W1 and W2 and their bonding surfaces can be adjusted by adjusting the heat generation amounts of the substrate heating units 481 and 482. The substrate heating units 481 and 482 are connected to a heater driver (not shown), which supplies current to the substrate heating units 481 and 482 based on control signals input from the controller 9 shown in FIG. 1 to cause the substrate heating units 481 and 482 to generate heat.

[0024] The vibration isolation unit 160 is a so-called active vibration isolation table, and collectively supports the chamber 120, stage 401, head 402, stage driver 403, head driver 404, substrate heating units 481 and 482, and imaging unit 500. As shown in Fig. 6, the vibration isolation unit 160 has a top plate 161, a base plate 165 arranged vertically below the top plate 161, i.e., on the -Z direction side, and a plate support unit 162 having a vibration isolation mechanism, fixed to the base plate 165, and supporting the top plate 161 on the +Z direction side so that it can move vertically and horizontally. Here, the chamber 120, stage 401, head 402, stage driver 403, head driver 404, substrate heating units 481 and 482, and imaging unit 500 are installed vertically above the top plate 161, i.e., on the +Z direction side. The plate support 162 has a vibration isolation mechanism using, for example, an air spring, a coil spring, or the like, and supports the top plate 161 so that it can move freely in the vertical and horizontal directions. The vibration isolation unit 160 further includes a vibration detection unit 164 that detects vibrations transmitted to the top plate 161, a plate driver 163 that moves the top plate 161 relative to the plate support 162, and a vibration isolation control unit 169 that controls the plate driver 163 to reduce vibrations transmitted to the top plate 161. The vibration detection unit 164 detects vibrations applied to the top plate 161 in three-dimensional directions, i.e., the X, Y, and Z directions. The plate driver 423 is composed of a hydraulic actuator, an electromagnetic actuator, a pneumatic actuator, a piezoelectric actuator, a linear actuator, or the like, and applies a force acting on the top plate 161 in the Z-axis direction or horizontal direction. The vibration isolation control unit 169 controls the plate driving unit 163 based on the vibrations detected by the vibration detection unit 164 so that the top plate 161 moves to cancel out the vibrations. The vibration isolation control unit 169 is specialized for controlling the plate driving unit 163, and executes processing independently of the control unit 9. This vibration isolation unit 160 removes vibration components of frequencies higher than 10 Hz that are transmitted to the stage 401 or the head 402 from the floor on which the bonding apparatus 1 is installed, etc.It is preferable that the vibration isolation unit 160 has a frequency band of vibrations that can be isolated extending into the lower frequency range, and preferably removes vibration components with frequencies higher than 4 Hz, more preferably removes vibration components with frequencies higher than 2 Hz, and even more preferably removes vibration components with frequencies higher than 1 Hz.

[0025] Returning to FIG. 1 , the control unit 9 is a control system including, for example, a personal computer, and includes a CPU (Central Processing Unit) and memory. The memory stores programs executed by the CPU. The memory also stores preset positional deviation thresholds Δxth, Δyth, and Δθth for the relative positional deviations Δx, Δy, and Δθ of the substrates W1 and W2 (described later). The control unit 9 converts measurement signals input from the pressure sensor 412, the pressure sensor 408, and the distance measurement unit 490 into measurement information and acquires the measurement information. The control unit 9 also converts captured image signals input from the imaging units 501 and 502 into captured image information and acquires the captured image information. Furthermore, the control unit 9 controls the operations of the holding unit driver, the piezo actuator 411, the pressure driver 431b, the pressure driver 432b, the heating unit driver, the stage driver 403, and the head driver 404 by outputting control signals to each of these units.

[0026] As shown in FIG. 5B , the control unit 9 calculates the misalignment amounts Δxa and Δya between a set of alignment marks MK1a and MK2a provided on the substrates W1 and W2 based on the captured image GAa acquired from the imaging unit 501. Note that FIG. 5B shows a state in which the set of alignment marks MK1a and MK2a are misaligned with respect to each other. Similarly, the control unit 9 calculates the misalignment amounts Δxb and Δyb between another set of alignment marks MK1b and MK2b provided on the substrates W1 and W2 based on the captured image GAb acquired from the imaging unit 502. The control unit 9 then calculates the relative misalignment amounts Δx, Δy, and Δθ between the two substrates W1 and W2 in the X direction, Y direction, and rotational direction around the Z axis based on the misalignment amounts Δxa, Δya, Δxb, and Δyb of these two sets of alignment marks and the geometric relationship between the two sets of marks. Then, the control unit 9 moves the head 402 in the X direction and Y direction and rotates it around the Z axis so as to reduce the calculated misalignment amounts Δx, Δy, and Δθ. This reduces the relative misalignment amounts Δx, Δy, and Δθ between the two substrates W1 and W2. In this way, the bonding apparatus 1 performs an alignment operation to correct the misalignment amounts Δx, Δy, and Δθ in the horizontal direction between the two substrates W1 and W2.

[0027] Furthermore, the control unit 9 repeatedly measures the positional misalignment amount of the substrate W2 relative to the substrate W1 during a predetermined waveform measurement period while the substrates W1 and W2 are spaced apart. The length of the waveform measurement period is set to a period at least longer than the vibration period of the vibration component of the vibration of the substrate W2 relative to the substrate W1. For example, if the positional misalignment measurement process can be performed at a period of several tens of milliseconds, the control unit 9 can handle a case in which the vibration of the substrate W2 relative to the substrate W1 includes a vibration component with a period of 10 Hz. However, since the greater the number of samples per period, the higher the accuracy, the vibration component included in the vibration of the substrate W2 relative to the substrate W1 is preferably 5 Hz or less, and more preferably 2 Hz or less. Here, since the vibration component with a period of 2 Hz is in a range that cannot be removed by the stand 41 having a vibration isolation function, the length of the waveform measurement period during which the positional misalignment amount is repeatedly measured is preferably set to, for example, 1 second or more. The control unit 9 then repeatedly calculates the relative positional deviations Δx, Δy, and Δθ between the two substrates W1 and W2 at predetermined time intervals shorter than the vibration period of the substrate W2 relative to the substrate W1. The control unit 9 also identifies the vibration waveform of the vibration of the substrate W2 relative to the substrate W1 from the time progression of the positional deviations Δx, Δy, and Δθ measured during the waveform measurement period. Here, the control unit 9 identifies, for example, the positional deviation, vibration amplitude, and vibration period corresponding to the vibration center of the vibration waveform. The control unit 9 then estimates the target timing at which the positional deviations Δx, Δy, and Δθ of the substrate W2 relative to the substrate W1 will reach the target amounts, based on the identified vibration waveform. For example, the control unit 9 measures the positional deviations Δx, Δy, and Δθ multiple times to identify the vibration waveform of the substrate W2 relative to the substrate W1, and estimates the time at which the positional deviations Δx, Δy, and Δθ measured at any timing within one vibration period will reach the target amounts, based on the identified vibration waveform. Furthermore, when specifying the vibration waveform, the control unit 9 does not necessarily need to measure the maximum vibration amplitude of the positional deviation amount, but only needs to estimate the time required for the positional deviation amounts Δx, Δy, and Δθ measured at any timing within one vibration cycle to reach the target amounts. Here, the target amounts are set to positional deviation amounts corresponding to the amplitude centers of the vibration components of the positional deviation amounts Δx, Δy, and Δθ, for example.The target amount may also be set to an amount offset by the amount of misalignment of the substrate W2 relative to the substrate W1 caused by contact between the substrates W1 and W2. In this case, the relative alignment of the substrates W1 and W2 is performed so that the amounts of misalignment Δx, Δy, and Δθ when the substrates W1 and W2 are brought into contact with each other become zero.

[0028] Furthermore, the control unit 9 pre-stores in memory required time information indicating the required time required for the substrates W1 and W2 to move from a spaced-apart state to a state in which they are in contact with each other. This required time corresponds to, for example, the time required for the head 402 to be positioned so that the gap G1 between the substrates W1 and W2 is large enough to bring the central portions W1c and W2c of the substrates W1 and W2 into contact with each other by protruding the pressing portions 431a and 432a of the pressing mechanisms 431 and 432, respectively, while the head 402 is positioned so that the gap G1 between the substrates W1 and W2 is large enough to bring the central portions W1c and W2c of the substrates W1 and W2 into contact with each other simply by bending the substrates W1 and W2. The control unit 9 then starts the operation for bringing the substrates W1 and W2 into contact with each other at a time point that is the required time before the estimated target timing. Specifically, the control unit 9 controls the pressure drive units 431b and 432b so that the protruding operations of the pressure units 432a and 432b begin at a point in time that is the required time before the estimated target timing.

[0029] The control unit 9 also controls the distance measurement unit 490 to measure the distances between three sites P11, P12, and P13 on the stage 401 and sites P21, P22, and P23 on the head 402 that correspond to the sites P11, P12, and P13, respectively. Then, based on the distances measured by the distance measurement unit 490, the control unit 9 controls the three piezo actuators 411 described above so that the substrate W2 held by the head 402 is parallel to the substrate W1 held on the stage 401.

[0030] Next, a bonding method performed by the bonding apparatus 1 according to this embodiment will be described with reference to FIGS. 7 to 9 . In FIG. 7 , the bonding apparatus 1 is assumed to have completed measurement of the distance between the upper surface of the stage 401 and the lower surface of the head 402 using the distance measurement unit 490 without the substrates W1 and W2 being held by the stage 401 and the head 402, and to have stored the measurement results in memory. Furthermore, it is assumed that the thickness measurement results of the substrates W1 and W2 have already been stored in memory. It is assumed that the substrates W1 and W2 are vibrating in a direction perpendicular to the vertical direction, i.e., in the horizontal direction, and that the imaging units 501 and 502 are vibrating in the vertical direction. It is also assumed that the substrates W1 and W2 are vibrating with a vibration waveform having a frequency of 10 Hz or less or a vibration amplitude of 1 μm or less.

[0031] First, the bonding apparatus 1 causes the stage 401 to hold only the peripheral portion of the substrate W1, and causes the head 402 to hold only the peripheral portion of the substrate W2 with the bonding surfaces of the substrates W1 and W2 facing each other (step S101). Here, for example, with the substrate W1 placed on the stage 401, the control unit 9 drives the electrostatic chuck 441 disposed in the first region A1 of the stage 401 to hold only the peripheral portion of the substrate W1 on the stage 401. Also, for example, with the head 402 in contact with the side of the substrate W2 opposite to the bonding surface side, which is disposed vertically below the head 402, the control unit 9 drives the electrostatic chuck 442 disposed in the first region A1 of the head 402 to hold only the peripheral portion of the substrate W2 on the head 402.

[0032] Next, the bonding apparatus 1 calculates the distance between the bonding surfaces of the substrates W1 and W2 based on the thicknesses of the substrates W1 and W2 and the distance between the upper surface 401a of the stage 401 and the lower surface 402a of the head 402 when the substrates W1 and W2 are not held on the stage 401 and the head 402. Then, based on the calculated distance, the bonding apparatus 1 moves the head 402 vertically downward to bring the substrates W1 and W2 closer to each other (step S102).

[0033] Next, the bonding apparatus 1 measures the misalignment amount of the substrate W1 relative to the substrate W2 while the substrates W1 and W2 are spaced apart (step S103). Here, the control unit 9 first captures images GAa and GAb (see FIG. 5A ) of the two substrates W1 and W2 in a non-contact state using the imaging units 501 and 502 of the imaging unit 500. Then, the control unit 9 calculates the misalignment amounts Δx, Δy, and Δθ of the two substrates W1 and W2 in the X direction, the Y direction, and the rotational direction around the Z axis, respectively, based on the two captured images GAa and GAb. Specifically, the control unit 9 calculates the misalignment amounts Δxa and Δya (see FIG. 5B ) using a vector correlation method based on the captured image GAa obtained by simultaneously reading alignment marks MK1a and MK2a spaced apart in the Z direction, for example. Similarly, the controller 9 calculates the positional deviations Δxb and Δyb using the vector correlation method based on the captured image GAb obtained by simultaneously reading the alignment marks MK1b and MK2b spaced apart in the Z direction. Then, the controller 9 calculates the positional deviations Δx, Δy, and Δθ of the two substrates W1 and W2 in the horizontal direction based on the positional deviations Δxa, Δya, Δxb, and Δyb.

[0034] 7 , the bonding apparatus 1 then aligns the substrates W1 and W2 by moving the substrate W2 relative to the substrate W1 so as to correct the calculated misalignments Δx, Δy, and Δθ (step S104). Here, the bonding apparatus 1 moves the head 402 in the X direction, the Y direction, and a rotational direction around the Z axis while keeping the stage 401 fixed so as to eliminate the misalignments Δx, Δy, and Δθ. At this time, the bonding apparatus 1 also measures the distances between three positions P11, P12, and P13 on the stage 401 and positions P21, P22, and P23 on the head 402 corresponding to the positions P11, P12, and P13, respectively, and performs an attitude adjustment step of adjusting the attitude of the substrate W2 held by the head 402 relative to the substrate W1 held on the stage 401 based on the measured distances.

[0035] Next, the bonding apparatus 1 moves the head 402 even closer to the stage 401 (step S105). Here, as shown in FIG. 8A, the bonding apparatus 1 positions the head 402 at a position where the gap G1 between the substrates W1 and W2 is large enough that the central portions W1c and W2c of the substrates W1 and W2 come into contact with each other simply by bending the substrates W1 and W2. Next, as shown in FIG. 7, the bonding apparatus 1 executes a misalignment measurement process in which, with the substrates W1 and W2 spaced apart from each other, the bonding apparatus 1 repeatedly measures the misalignment amount of the substrate W2 relative to the substrate W1 during a predetermined waveform measurement period (step S106). The length of the waveform measurement period is set to be at least longer than the vibration period dT1 of the substrate W2 relative to the substrate W1 shown in FIG. 8B, for example, 1 sec or longer. Furthermore, the bonding apparatus 1 repeatedly measures the misalignment amount at a predetermined time interval dT2 shorter than the vibration period dT1 of the substrate W2 relative to the substrate W1. In the bonding apparatus 1, the alignment marks MK1a and MK2a and the alignment marks MK1b and MK2b are spaced apart by a distance that falls within the depth of field of the imaging units 501 and 502, and the imaging units 501 and 502 are positioned so that they can capture images of the pair of alignment marks MK1a and MK2a and the pair of alignment marks MK1b and MK2b, respectively. The bonding apparatus 1 controls the imaging units 501 and 502 so that the imaging units 501 and 502 simultaneously capture the corresponding pair of alignment marks MK1a and MK2a and the corresponding pair of alignment marks MK1b and MK2b in a single image capture at the same timing during each repeated measurement. The bonding apparatus 1 then calculates the amount of misalignment of the substrate W2 with respect to the substrate W1 based on the images captured by the imaging units 501 and 502.

[0036] 7 , the bonding apparatus 1 then executes a vibration waveform identification step (step S107) to identify the vibration waveform of the vibration of the substrate W2 relative to the substrate W1 from the time transition of the misalignment amount measured during the waveform measurement period. Here, the bonding apparatus 1 identifies the vibration waveform of the vibration of the substrate W2 in the horizontal and rotational directions relative to the substrate W1 from the time transition of the misalignment amount in the horizontal and rotational directions. Then, the bonding apparatus 1 identifies the misalignment amounts Δx, Δy, Δθ, vibration amplitude, and vibration period corresponding to the vibration center of the vibration waveform.

[0037] Next, the bonding apparatus 1 determines whether all of the positional deviations Δx, Δy, and Δθ corresponding to the vibration centers of the identified vibration waveforms are equal to or less than the preset positional deviation thresholds Δxth, Δyth, and Δθth (step S108). Assume that the bonding apparatus 1 determines that any of the positional deviations Δx, Δy, and Δθ corresponding to the vibration centers of the identified vibration waveforms is greater than the preset positional deviation thresholds Δxth, Δyth, and Δθth (step S108: No). In this case, the bonding apparatus 1 calculates a correction movement amount for the substrate W2 relative to the substrate W1 so that all of the positional deviations Δx, Δy, and Δθ corresponding to the vibration centers of the identified vibration waveforms are equal to or less than the positional deviation thresholds Δxth, Δyth, and Δθth (step S109). The control unit 9 then calculates a correction movement amount for moving the substrate W2 by the positional deviations Δx, Δy, and Δθ corresponding to the vibration centers of the identified vibration waveforms in the direction opposite to the positional deviation direction.

[0038] Next, the bonding apparatus 1 performs alignment to correct the relative positional deviations Δx, Δy, and Δθ between the two substrates W1 and W2 (step S110). Here, the bonding apparatus 1 moves the head 402 in the X direction, the Y direction, and the rotational direction around the Z axis by the correction movement amount calculated in step S109, while the stage 401 is fixed. In this way, the bonding apparatus 1 adjusts the relative position of the substrate W2 with respect to the substrate W1 so that the positional deviations Δx, Δy, and Δθ are reduced, while the substrates W1 and W2 are spaced apart from each other. Then, the bonding apparatus 1 performs the process of step S106 again.

[0039] On the other hand, suppose the bonding apparatus 1 determines that all of the calculated positional deviations Δx, Δy, and Δθ are equal to or less than the preset positional deviation thresholds Δxth, Δyth, and Δθth (step S108: Yes). In this case, the bonding apparatus 1 executes a timing estimation step to estimate a target timing at which the positional deviation of the substrate W2 relative to the substrate W1 will reach a target amount, i.e., a positional deviation amount corresponding to the vibration center of the vibration waveform, based on the identified vibration waveform (step S111). Here, the bonding apparatus 1 estimates a target timing Ts1c (or Ts2c) at which the positional deviation Δ of the substrate W2 relative to the substrate W1 will reach a positional deviation Δc corresponding to the vibration center of the vibration waveform, as shown in FIG. 8B .

[0040] Returning to FIG. 7 , the bonding apparatus 1 then executes a contacting step (step S112) in which the substrates W1 and W2 are bent based on the estimated target timing to bring the central portion W1c of the substrate W1 into contact with the central portion W2c of the substrate W2. Here, as shown in FIG. 8B , the bonding apparatus 1 starts the operation for bringing the substrates W1 and W2 into contact with each other at a time point that is the aforementioned required time dT3 before the estimated target timing Ts1c (or Ts2c). Here, the bonding apparatus 1 bends the substrate W1 so that the central portion W1c protrudes toward the substrate W2 relative to the peripheral portion W1s of the substrate W1, as shown in FIG. 9A , for example. At this time, the bonding apparatus 1 applies a voltage to the electrostatic chuck 441 from the holder driving unit to hold the substrate W1 on the electrostatic chuck 441, and then presses the central portion of the substrate W1 toward the substrate W2 with the pressing unit 431a. As a result, the substrate W1 is bent so that its central portion W1c protrudes toward the substrate W2. Then, the central portions W1c and W2c of the substrates W1 and W2 come into contact with each other. The bonding apparatus 1 also bends the substrate W2 so that the central portion W2c protrudes toward the substrate W1 relative to the peripheral portion W2s of the substrate W2. At this time, the bonding apparatus 1 applies a voltage from the holder drive unit to the electrostatic chuck 442 to hold the substrate W2 on the electrostatic chuck 442, and then presses the central portion W2c of the substrate W2 toward the substrate W1 with the pressing unit 432a. As a result, the substrate W2 bends so that the central portion W2c protrudes toward the substrate W1.

[0041] The bonding apparatus 1 then moves the head 402 downward using the elevation drive unit 406, thereby expanding the contact areas of the substrates W1 and W2 from the central portions W1c and W2c of the substrates W1 and W2 toward the peripheral portions W1s and W2s. While the central portions W1c and W2c of the substrates W1 and W2 are abutted against each other and the distance between the peripheral portions W1s and W2s of the substrates W1 and W2 is maintained constant, the contact areas of the substrates W1 and W2 expand from the central portions W1c and W2c of the substrates W1 and W2 toward the peripheral portions W1s and W2s due to intermolecular forces (van der Waals forces) generated between the substrates W1 and W2. Here, the bonding apparatus 1 applies point pressure to the central portions W1c and W2c of the substrates W1 and W2 using the pressing units 431a and 432a while separating the substrates W1 and W2 by approximately 50 μm. This point pressure triggers a so-called bonding wave to naturally spread toward the peripheries of the substrates W1 and W2 without external pressure being applied to the substrates W1 and W2 in the direction that brings them closer to each other. Due to the bonding force between the bonding surfaces of the substrates W1 and W2, this bonding wave spreads toward the peripheries of the substrates W1 and W2 without external pressure being applied to the substrates W1 and W2 in the direction that brings them closer to each other. Here, the bonding apparatus 1 moves the pressing unit 431a in a direction that immerses it in the stage 401, and moves the pressing unit 432a in a direction that immerses it in the head 402, as indicated by arrow AR12 in FIG. 9B . At the same time, the bonding apparatus 1 moves the head 402 in a direction that brings it closer to the stage 401, as indicated by arrow AR13. Here, the pressing unit driver 431b sinks the pressing unit 431a as the head 402 descends while controlling the position so that the tip of the pressing unit 431a is maintained at a preset position, and the pressing unit driver 432b controls the pressure applied to the pressing unit 432a so that the pressure is constant, thereby maintaining the contact points of the substrates W1 and W2 at the center in the opposing direction of the substrates W1 and W2. This makes it possible to prevent warping of the substrates W1 and W2 when the substrates W1 and W2 are bonded. The bonding apparatus 1 then reduces the distance between the peripheral portions W1s and W2s of the substrates W1 and W2 while the central portions W1c and W2c of the substrates W1 and W2 are butted against each other.As a result, as shown by arrow AR11 in FIG. 9B , the contact area between the substrates W1 and W2 further expands from the central portions W1c and W2c of the substrates W1 and W2 toward the peripheral portions W1s and W2s. The bonding apparatus 1 then further expands the contact area between the substrates W1 and W2 from the central portions W1c and W2c of the substrates W1 and W2 toward the peripheral portions W1s and W2s of the substrates W1 and W2, bringing the entire surfaces of the substrates W1 and W2 into contact with each other. Here, the bonding apparatus 1 moves the pressing unit 431a in a direction to retract the pressing unit 431a into the stage 401 and moves the pressing unit 432a in a direction to retract the head 402, while simultaneously moving the head 402 toward the stage 401, thereby reducing the distance between the peripheral portions of the substrates W1 and W2. In this way, the bonding apparatus 1 brings the peripheral portion of the substrate W1 into contact with the peripheral portion of the substrate W2, bringing the bonding surfaces of the substrates W1 and W2 into contact with each other over their entire surfaces, as shown in FIG. 10A .

[0042] Thereafter, the bonding apparatus 1 presses only the peripheral portion W1s of the substrate W1 against the peripheral portion W2s of the substrate W2 while the entire surfaces of the substrates W1 and W2 are in contact with each other, thereby pressing the peripheral portions W1s and W2s of the substrates W1 and W2 together (step S113). Next, the bonding apparatus 1 stops the electrostatic chuck 442 of the head 402 to release the substrate W2 (step S114). Next, the bonding apparatus 1 lifts the head 402 to separate it from the substrate W2, as shown by arrow AR14 in FIG. 10B .

[0043] By placing the joining device 1 on a stand 41 having a vibration isolation function, a part of the vibration components transmitted to the floor F is reduced in vibration transmitted to the joining device 1, as shown in FIG. 11. In FIG. 11, f indicates the frequency of the vibration transmitted to the floor F, and f ０ indicates the resonance frequency specific to the base 41. The vertical axis indicates the vibration transmissibility of the vibration transmitted from the floor F to the joining device 1 via the base 41, and the horizontal axis indicates the frequency, which is the resonance frequency f ０ Furthermore, f ｔ indicates the lower limit frequency of the frequency range in which the vibration isolation effect can be obtained by the stand 41. However, as shown in FIG. 11, when the frequency of the vibration transmitted to the floor F is ｔThe vibration transmissibility is reduced only for vibration components higher than the lower limit frequency f ｔ The vibration transmissibility cannot be reduced for the following vibration components: ０ is about 1.8 Hz for a stand 41 with a normal vibration isolation function, and f ｔ is about 2 Hz. In this case, if the vibration transmitted to the floor F includes a vibration component with a frequency of 2 Hz or less, this vibration component cannot be reduced by the stand 41 and the vibration isolation unit 160. In fact, as shown in FIG. 12, the frequency spectrum SPE1 when the joining apparatus 1 is placed on the stand 41 with vibration isolation function has a vibration amplitude reduced to less than 0.1 μm, particularly around 6 Hz, compared to the frequency spectrum SPE2 when the joining apparatus 1 is not placed on the stand 41. However, the vibration amplitude at frequencies of 2 Hz or less is still 0.1 μm or more, as shown in the area surrounded by the dashed line. In FIG. 12, the dashed line indicates a vibration amplitude of 0.1 μm. Therefore, in the joining apparatus 1 according to this embodiment, the lower limit frequency f of the frequency range in which the stand 41 can provide a vibration isolation effect is ｔ The vibration waveforms of the vibration components of the following relatively low frequencies are identified, and based on the identified vibration waveforms, the target timing at which the positional deviation amount of the substrates W1 and W2 will be the positional deviation amount corresponding to the vibration center is estimated. Then, the bonding apparatus 1 protrudes the pressing portion 431a of the pressing mechanism 431 and the pressing portion 432a of the pressing mechanism 432 so that the substrates W1 and W2 come into contact with each other at the estimated target timing. As a result, the substrate W2 is pressed against the substrate W1 at the aforementioned lower limit frequency f ｔ Even if the substrates W1 and W2 are vibrating at a relatively low frequency below this, it is possible to bring the substrates W1 and W2 into contact with each other with high positional accuracy.

[0044] As described above, the bonding apparatus 1 according to this embodiment identifies the vibration waveform of the vibration of the substrate W2 relative to the substrate W1 from the time progression of the misalignment amount of the substrate W2 relative to the substrate W1, and estimates the target timing at which the misalignment amount of the substrate W2 relative to the substrate W1 will reach the target amount based on the identified vibration waveform. The bonding apparatus 1 then brings the substrates W1 and W2 into contact with each other based on the estimated target timing. This allows the substrate W2 to be bonded to the substrate W1 with high positional accuracy even when the substrate W2 is vibrating relative to the substrate W1.

[0045] Conventionally, bonding methods have been proposed in which the substrates W1 and W2 are aligned by repeatedly contacting, separating, and aligning the substrates W1 and W2 with each other. However, with this bonding method, if the substrate W2 is vibrating relative to the substrate W1, the contact position between the substrates W1 and W2 fluctuates depending on the timing of contacting the substrates W1 and W2 with each other. In contrast, the bonding method of the present embodiment identifies the vibration waveform of the substrate W2 relative to the substrate W1, estimates the timing of contacting the substrates W1 and W2 with each other based on the identified vibration waveform, and contacts the substrates W1 and W2 with each other at the estimated timing. This reduces the relative positional deviation of the substrate W2 with respect to the substrate W1, which is caused by variations in the timing of contacting the substrates W1 and W2 with each other. This has the advantage of eliminating the need to repeatedly contact and separate the substrates W1 and W2 with each other.

[0046] Furthermore, according to the bonding apparatus 1 of this embodiment, in the first contacting step described above, the central portions of the bonding surfaces of the substrates W1 and W2 are brought into contact with each other while the substrates W1 and W2 are bent so that the central portions of the bonding surfaces of the substrates W1 and W2 protrude toward each other relative to the peripheral portions. Therefore, for example, by bringing the head 402 closer to the stage 402, the time required for the substrates W1 and W2 to come into contact with each other from a spaced-apart state can be reduced compared to when the entire bonding surfaces of the substrates W1 and W2 are brought into contact with each other. This has the advantage of making it easier to bring the substrates W1 and W2 into contact with each other at the target timing described above.

[0047] Furthermore, the bonding apparatus 1 according to this embodiment measures the distances between three positions P11, P12, and P13 on the stage 401 and positions P21, P22, and P23 on the head 402 corresponding to the positions P11, P12, and P13, respectively, and controls the three piezo actuators 411 based on the measured distances so that the substrate W2 held by the head 402 is parallel to the substrate W1 held on the stage 401. This makes it possible to improve the relative positional accuracy of the substrates W1 and W2 even when the substrates W1 and W2 are aligned while spaced apart from each other and then bonded together.

[0048] Furthermore, in the bonding apparatus 1 according to the present embodiment, the alignment marks MK1a, MK1b, MK2a, and MK2b provided on the substrates W1 and W2 are simultaneously imaged by the imaging units 501 and 502, and therefore the bonding apparatus 1 is not affected by misalignment of the imaging units 501 and 502. Furthermore, in the bonding apparatus 1, the imaging units 501 and 502 are arranged corresponding to the set of alignment marks MK2a and MK1a and the set of alignment marks MK2b and MK1b, respectively, and images the two sets simultaneously, so that the effects of vibration of the imaging units 501 and 502 can be canceled out.

[0049] In the bonding apparatus according to the present embodiment, due to the structure for aligning the relative positions of the head 402 and the stage 401 in the horizontal direction, the vibration amplitude in the vertical direction tends to be relatively small and the vibration amplitude in the direction perpendicular to the vertical direction tends to be relatively large. In this case, for example, as in Comparative Example 1 shown in FIG. 13 , when the image capture units 501 and 502 are oriented with their optical axes perpendicular to the vertical direction, i.e., aligned along the X-axis direction, the horizontal vibrations of the image capture units 501 and 502, which have relatively large vibration amplitudes as indicated by arrows AR23 and AR24, affect the measurement of the misalignment between the alignment marks MK1a and MK2a and the alignment marks MK1b and MK2b. Therefore, in the configuration according to this comparative example, the horizontal vibrations of the image capture units 501 and 502, as well as the horizontal vibrations of the head 402 and the stage 401 as indicated by arrows AR21 and AR22, affect the alignment between the substrates W1 and W2. In contrast, the imaging units 501 and 502 according to this embodiment are disposed with their optical axes perpendicular to the vertical direction, i.e., aligned along the X-axis. In this state, the imaging units 501 and 502 receive light traveling downward from the alignment marks MK1a and MK2a and the alignment marks MK1b and MK2b via a mirror 504 that converts the light to a direction perpendicular to the vertical direction, thereby capturing images of the alignment marks MK1a and MK2a and the alignment marks MK1b and MK2b with the imaging units 501 and 502. This allows horizontal vibrations of the imaging units 501 and 502, which have a relatively large amplitude, to primarily affect the focus position of the imaging units 501 and 502, but to have almost no effect on the misalignment between the alignment marks MK1a and MK2a and the alignment marks MK1b and MK2b captured by the imaging units 501 and 502. Therefore, it is possible to prevent horizontal vibrations of the imaging units 501 and 502, which have a relatively large vibration amplitude, from affecting the amount of positional deviation between the alignment marks MK1a and MK2a and the alignment marks MK1b and MK2b.

[0050] The bonding apparatus 1 according to this embodiment also includes the vibration isolation unit 160. This eliminates vibration components with frequencies higher than 10 Hz that are transmitted to the stage 401 or head 402 from the floor on which the bonding apparatus 1 is installed. Conventionally, vibration components transmitted to the stage 401 or head 402 were limited to those that could be eliminated using the vibration isolation unit 160, which is an active vibration isolation table. As a result, vibration components below 10 Hz that could not be eliminated by the vibration isolation unit 160 remained. In contrast, the bonding apparatus 1 according to this embodiment identifies the vibration waveform of the vibration of the substrate W2 relative to the substrate W1 and, based on the identified vibration waveform, estimates the target timing at which the misalignment amount of the substrate W2 relative to the substrate W1 reaches a target amount. This reduces the influence of vibration components below 10 Hz on the misalignment of the bonded substrates W1 and W2, thereby effectively eliminating vibration components below 10 Hz. Furthermore, in the bonding apparatus 1 according to this embodiment, the vibration velocity of vibration components in the relatively high frequency range higher than 10 Hz is faster than the image processing speed, making it difficult to estimate the target timing based on the vibration waveform. However, by using the vibration isolation unit 160, which is a so-called active vibration isolation table, in combination, it is possible to reduce vibration components in the entire frequency range.

[0051] (Embodiment 2) A chip bonding system according to this embodiment is an apparatus for bonding a semiconductor chip (hereinafter simply referred to as a "chip") onto a substrate. The semiconductor chip is supplied, for example, from a diced substrate. This chip bonding system performs activation processing on the surface of the substrate to which the chip is bonded and the bonding surface of the electronic chip, and then bonds the chip to the substrate by bringing the chip into contact with the substrate and applying pressure.

[0052] 14 , the chip bonding system 2 according to this embodiment includes a chip supply device 2010, a bonding device 2030, a chip transport device 2039, and a control unit 2009. The chip supply device 2010 cuts out one chip CP from a plurality of chips CP produced by dicing a substrate, and supplies the chip CP to the bonding device 2030. Here, dicing refers to a process of cutting a substrate, on which a plurality of electronic chips are built, in the vertical and horizontal directions to form chips. The chip supply device 2010 includes a chip supply unit 2011 and a supplied chip imaging unit 2015.

[0053] The chip supply unit 2011 includes a sheet holding frame 2112 that holds a sheet TE to which multiple chips CP are attached, a frame holding unit 2119 that holds the sheet holding frame 2112, a pickup mechanism 2111 that picks up one chip CP from the multiple chips CP, and a cover 2114. The chip supply unit 2011 also includes a holding frame drive unit 2113 that drives the sheet holding frame 2112 in the XY directions or in a direction rotating around the Z axis. The frame holding unit 2119 holds the sheet holding frame 2112 in an orientation where the surface of the sheet TE to which the multiple chips CP are attached faces vertically upward (+Z direction). The sheet holding frame 2112 and the frame holding unit 2119 form a sheet holding unit that holds the sheet TE to which the multiple chips CP are attached on the side opposite the bonding surface CPf of each chip CP, with the bonding surface CPf facing vertically upward.

[0054] The pickup mechanism 2111 separates one of the chips CP from the sheet TE by cutting out the chip CP from the side opposite the chips CP on the sheet TE. Here, the pickup mechanism 2111 holds a peripheral portion of the chip CP opposite the bonding surface CPf, which is different from the central portion held by the head 2033H (described later), and cuts out the chip CP. The pickup mechanism 2111 has a needle 2111a and is movable in the vertical direction as indicated by arrow AR14. The cover 2114 is disposed so as to cover the chips CP vertically above, and has a hole 2114a in the portion facing the pickup mechanism 2111. There are, for example, four needles 2111a. However, the number of needles 2111a may be three, five, or more. The pickup mechanism 2111 supplies chips CP by piercing the sheet TE with a needle 2111a from vertically below (-Z direction) the sheet TE and lifting the chips CP vertically upward (+Z direction). Then, each chip CP attached to the sheet TE is pushed upward by the needle 2111a through a hole 2114a in the cover 2114 one by one, and is delivered to the chip transport device 2039. The holding frame drive unit 2113 drives the sheet holding frame 2112 in the XY direction or in a direction rotating around the Z axis, thereby changing the position of the chip CP located vertically below the needle 2111a.

[0055] The supplied chip imaging unit 2015 is disposed above (in the +Z direction) the chip supply unit 2011 in the chip supply device 2010. The supplied chip imaging unit 2015 photographs the chip CP protruded above the cover 2114 by the pickup mechanism 2111.

[0056] The chip transport device 2039 transports chips CP supplied from the chip supply unit 2011 to a transfer position Pos1 where the chips CP are transferred to the head 2033H of the bonding unit 2033 of the bonding device 2030. The chip transport device 2039 has a long plate 2391, an arm 2394, a chip holder 2393 provided at the tip of the arm 2394, and a plate driver 2392 that rotates the plate 2391. The plate 2391 is long and cylindrical, and one end of the plate 2391 rotates around a rotation axis AX extending vertically, with the other end located between the chip supply unit 2011 and the head 2033H as the base point. The number of plates 2391 may be multiple or may be one.

[0057] As shown in FIG. 15A , the chip holding unit 2393 is provided at the tip of an arm 2394 and has two legs 2393 a for holding a chip CP. As shown in FIG. 15B , the plate 2391 is capable of housing a long arm 2394 inside. An arm drive unit 2395 that drives the arm 2394 along the longitudinal direction of the plate 2391 is provided inside the plate 2391. This allows the chip transport device 2039 to use the arm drive unit 2395 to place the tip of the arm 2394 in a state where it protrudes outside the plate 2391 or a state where it is retracted inside the plate 2391. When the chip transport device 2039 rotates the plate 2391, the arm 2394 is retracted into the plate 2391, as shown by arrow AR15, to store the chip holding unit 2393 inside the plate 2391. This prevents particles from adhering to the chip CP during transport. The two legs 2393a may be provided with suction grooves (not shown). In this case, the chip CP is held by suction on the legs 2392a, allowing the chip CP to be transported without misalignment. Furthermore, a protrusion (not shown) may be provided at the tip of the legs 2393a to prevent the chip CP from flying out due to centrifugal force generated when the plate 2391 rotates. When the chip transport device 2039 receives the chip CP from the pickup mechanism 2111, it rotates the plate 2391 around the axis AX to transport the chip CP to a transfer position Pos1 where the chip CP overlaps with the head 2033H in the vertical direction.

[0058] 16 , the bonding device 2030 has a stage unit 2031, a bonding section 2033 having a head 2033H, a head driving section 2036 that drives the head 2033H, image capturing sections 2035a and 2035b, an image capturing section 2041, a camera F-direction driving section 2365, and a camera Z-direction driving section 2363. The bonding section 2033 has a Z-axis direction moving member 2331, a first disk member 2332, a piezo actuator 2333, a second disk member 2334, a mirror fixing member 2336, a mirror 2337, and the head 2033H.

[0059] A first disk member 2332 is fixed to the upper end of the Z-axis direction moving member 2331. A second disk member 2334 is disposed above the first disk member 2332. The first disk member 2332 and the second disk member 2334 are connected via a piezo actuator 2333. Furthermore, a head 2033H is fixed to the upper surface side of the second disk member 2334. The head 2033H adsorbs and holds the chip CP.

[0060] The head 2033H holds the tip CP from vertically below (-Z direction). As shown in FIGS. 17A and 17B, the head 2033H includes a tip tool 2411, a head main body 2413, a tip support 2432a, and a support drive unit 2432b. The tip tool 2411 is formed from a material (e.g., silicon (Si)) that transmits imaging light (e.g., infrared light). The head main body 2413 also includes a built-in ceramic heater, coil heater, or the like. The head main body 2413 also includes hollow portions 2415 and 2416 for transmitting (passing through) the imaging light. Each hollow portion 2415 and 2416 is a transparent portion that transmits imaging light and is provided to penetrate the head main body 2413 in the vertical direction (Z-axis direction). As shown in FIG. 18, each hollow portion 2415 and 2416 has an elliptical shape when viewed from above. The two hollow portions 2415, 2416 are arranged point-symmetrically about the axis BX at diagonal corners of the head main body portion 2413, which has a generally square shape in top view. As shown in FIG. 16 , holes 2334a, 2334b are also provided in the second disk member 2334 at portions corresponding to the hollow portions 2415, 2416 to allow imaging light to pass through. Returning to FIGS. 17A and 17B , the head main body portion 2413 has a holding mechanism 2440 having an adsorption portion for adsorbing and holding the tip CP to the tip tool 2411. The head main body portion 2413 also has an adsorption portion (not shown) for fixing the tip tool 2411 to the head main body portion 2413 by vacuum adsorption. The tip tool 2411 has a through-hole 2411a formed at a position corresponding to the holding mechanism 2440 of the head main body portion 2413, and a through-hole 2411b through which the tip support portion 2432a is inserted.

[0061] The chip support portion 2432a is, for example, a cylindrical suction post that is provided at the tip of the head 2033H and is movable in the vertical direction. The chip support portion 2432a supports the side of the chip CP opposite to the bonding surface CPf. One chip support portion 2432a is provided, for example, in the center.

[0062] The support part driver 2432b drives the chip support part 2432a in the vertical direction and, with the chip CP placed on the tip of the chip support part 2432a, depressurizes the inside of the chip support part 2432a, thereby adsorbing the chip CP to the tip of the chip support part 2432a. The support part driver 2432b positions the chip holding part 2393 of the chip transport device 2039, holding the chip CP, at a transfer position for the head 2033H (see Pos1 in FIG. 14 ), and, with the tip of the chip support part 2432a supporting the center of the chip CP, moves the chip support part 2432a vertically upward from the chip holding part 2393. As a result, the chip CP is transferred from the chip holding part 2393 of the chip transport device 2039 to the head 2033H.

[0063] The piezo actuator 2333 is an attitude adjustment unit that adjusts at least one of the distance between the bonding surface WTf of the substrate WT and the bonding surface CPf of the chip CP and the inclination of the chip CP with respect to the bonding surface WTf of the substrate WT. As shown in FIG. 19A , three piezo actuators 2333 are located between the first disk member 2332 and the second disk member 2334, and each is expandable and contractible in the Z direction. By controlling the degree of expansion and contraction of each of the three piezo actuators 2333, the inclination angle of the second disk member 2334 and, ultimately, the head 2033H with respect to the horizontal plane is adjusted. At least one of the distance between the bonding surface CPf of the chip CP held by the head 2033H and the bonding surface WTf of the substrate WT and the inclination of the bonding surface CPf of the chip CP held by the head 2033H with respect to the bonding surface WTf of the substrate WT is adjusted. The three piezo actuators 2333 are arranged at positions (planar positions) that do not block illumination light (including reflected light) related to the imaging units 2035a and 2035b.

[0064] 16 , the mirror 2337 is fixed to the first disk member 2332 via a mirror fixing member 2336, and is disposed in the gap between the first disk member 2332 and the second disk member 2334. The mirror 2337 has inclined surfaces 2337a and 2337b that are inclined diagonally downward at an angle of 45 degrees. Photographing light that is incident on the inclined surfaces 2337a and 2337b of the mirror 2337 from the imaging units 2035a and 2035b is reflected upward.

[0065] The head driving unit 2036 moves the head 2033H holding the chip CP vertically upward (in the +Z direction) to bring the head 2033H closer to the stage 2031, thereby bonding the chip CP to the bonding surface WTf of the substrate WT. More specifically, the head driving unit 2036 moves the head 2033H holding the chip CP vertically upward (in the +Z direction) to bring the head 2033H closer to the stage 2031, thereby bringing the chip CP into contact with the bonding surface WTf of the substrate WT and surface-bonding the chip CP to the substrate WT. Here, the bonding surface WTf of the substrate WT and the bonding surface CPf of the chip CP to be bonded to the substrate WT have been previously subjected to activation treatment such as exposure to plasma or irradiation with a particle beam. Therefore, the chip CP is bonded to the substrate WT by bringing the bonding surface CPf of the chip CP into contact with the bonding surface WTf of the substrate WT. The bonding surface CPf of the chip CP may be, for example, a surface having at least a partially exposed flat metal portion.

[0066] The head driver 2036 includes a Z-direction driver 2034, a rotating member 2361, and a θ-direction driver 2037. The Z-direction driver 2034 includes a servo motor, a ball screw, and the like. The Z-direction driver 2034 is provided on the lower end side of the rotating member 2361 (described later), and drives the Z-axis direction moving member 2331 of the bonding unit 2033 in the Z-axis direction, as indicated by arrow AR211 in FIG. 14 . When the Z-direction driver 2034 moves the Z-axis direction moving member 2331 in the Z direction, the head 2033H provided at the upper end of the bonding unit 2033 moves in the Z direction accordingly. That is, the head 2033H is driven in the Z direction by the Z-direction driver 2034.

[0067] The rotating member 2361 is cylindrical, and as shown in FIG. 19B , the cross section of the hollow interior is octagonal. Meanwhile, the Z-axis moving member 2331 has a rod-shaped portion with an octagonal cross section, and is inserted inside the rotating member 2361. Furthermore, linear guides 2038 are provided between four of the eight side surfaces of the Z-axis moving member 2331 and the inner surface of the rotating member 2361, such that the Z-axis moving member 2331 slides relative to the rotating member 2361 in the Z-axis direction. When the rotating member 2361 rotates around the axis BX, the Z-axis moving member 2331 rotates in conjunction with the rotating member 2361. That is, the bonding portion 2033 and the rotating member 2361 rotate in conjunction with each other around the axis BX, as indicated by arrow AR212 in FIG. 14 .

[0068] The θ-direction driving unit 2037 has a servo motor, a reducer, and the like, and is fixed to a fixed member 2301 provided in the welding device 2030, as shown in Fig. 16. The θ-direction driving unit 2037 supports a rotating member 2361 rotatably around an axis BX. The θ-direction driving unit 2037 rotates the rotating member 2361 around the axis BX in response to a control signal input from the control unit 2009.

[0069] The imaging units 2035a and 2035b capture images of the alignment marks of the chip CP from vertically below (in the -Z direction) the chip CP, with the chip CP positioned at a position on the substrate WT where the chip CP will be bonded. The imaging unit 2035a is fixed to the rotating member 2361 via a camera Z-direction driver 2363 and a camera F-direction driver 2365. The imaging unit 2035b is also fixed to the rotating member 2361 via a camera Z-direction driver 2363 and a camera F-direction driver 2365. This causes the imaging units 2035a and 2035b to rotate together with the rotating member 2361. Here, as described above, the mirror 2337 is fixed to the Z-axis moving member 2331, and the rotating member 2361 and the Z-axis moving member 2331 rotate in conjunction with each other. Therefore, the relative positional relationship between the imaging units 2035a, 2035b and the mirror 2337 remains unchanged, and therefore, regardless of the rotation of the rotating member 2361, the imaging light reflected by the mirror 2337 is guided to the imaging units 2035a, 2035b. Furthermore, the hollow portions 2415, 2416 of the head 2033H rotate around the axis BX in conjunction with the rotation of the rotating member 2361. For example, as shown in FIG. 18 , assume that alignment marks MC2a, MC2b are provided at opposite corners of a square-shaped chip CP, sandwiching the center of the chip. In this case, when the imaging units 2035a, 2035b are positioned on the diagonal line connecting the two corners of the chip CP where the alignment marks MC2a, MC2b are provided, the imaging units 35a, 35b can acquire images of the alignment marks MC2a, MC2b through the hollow portions 415, 416.

[0070] Returning to FIG. 16 , the imaging units 2035a and 2035b each include an image sensor 2351a, 2351b, an optical system 2352a, 2352b, and a coaxial illumination system (not shown). The imaging units 2035a and 2035b each capture images using reflected light of illumination light (e.g., infrared light) emitted from a light source (not shown) of the coaxial illumination system. The illumination light emitted horizontally from the coaxial illumination system of the imaging units 2035a and 2035b is reflected by the inclined surfaces 2337a and 2337b of the mirror 2337, changing its direction of travel vertically upward. The light reflected by the mirror 2337 then travels toward an imaging target area, including the chip CP held by the head 2033H and the substrate WT disposed opposite the chip CP, and is reflected by each imaging target area. The light reflected from the imaging target portions of the chip CP and the substrate WT travels vertically downward, is reflected again by the inclined surfaces 2337a and 2337b of the mirror 2337, and its direction of travel is changed horizontally, reaching the imaging units 2035a and 2035b. The imaging units 2035a and 2035b then use the reflected light that has reached the imaging units 2035a and 2035b to image the imaging target portions of the chip CP and the substrate WT. Furthermore, when the imaging units 2035a and 2035b acquire captured images including images of the alignment marks provided on the chip CP and the alignment marks provided on the substrate WT, they output captured image signals indicating the acquired captured images to the control unit 2009. For example, as shown in FIGS. 20A and 20B , the substrate WT is provided with two alignment marks MC1a and MC1b, and the chip CP is also provided with two alignment marks MC2a and MC2b. Then, the control unit 2009 recognizes the relative position of each chip CP with respect to the substrate WT in a direction parallel to the surface of the substrate WT to which the chip CP is bonded, based on the captured image including images of alignment marks MC1a, MC1b, MC2a, and MC2b captured by the imaging units 2035a and 2035b.

[0071] 21 , a portion of the light emitted from the imaging unit 2035a, reflected by the mirror 2337, and passing through the hollow portion 2415 of the head 2033H is transmitted through the tip tool 2411 and the chip CP. A portion of the light that passed through the chip CP is reflected by a portion of the substrate WT where the alignment mark MC1a is provided. The remaining portion of the light that passed through the hollow portion 2415 of the head 2033H is reflected by a portion of the chip CP where the alignment mark MC2a is provided. The light reflected by the portion of the substrate WT where the alignment mark MC1a is provided or the portion of the chip CP where the alignment mark MC2a is provided passes through the tip tool 2411 and passes through the hollow portion 2415 of the head 2033H. The light that passed through the hollow portion 2415 of the head 2033H is then reflected by the mirror 2337 and enters the image sensor of the imaging unit 2035a. The imaging unit 2035a then uses light incident on the imaging element to acquire a captured image Ga including an image of the alignment mark MC2a provided on the chip CP and an image of the alignment mark MC1a provided on the substrate WT. The bonding device 2030 then uses the same imaging unit 2035a to simultaneously recognize a pair of the alignment mark MC1a on the chip CP and the alignment mark MC2a on the substrate WT by capturing images once without moving the focus axis.

[0072] Furthermore, a portion of the light emitted from the imaging unit 2035b, reflected by the mirror 2337, and passing through the hollow portion 2416 of the head 2033H also passes through the tip tool 2411 and the chip CP. A portion of the light that passes through the chip CP is reflected by the portion of the substrate WT where the alignment mark MC2b is provided. A portion of the remaining light that passes through the hollow portion 2416 of the head 2033H is reflected by the portion of the chip CP where the alignment mark MC2b is provided. The light reflected by the portion of the substrate WT where the alignment mark MC1b is provided or the portion of the chip CP where the alignment mark MC2b is provided passes through the tip tool 2411 and passes through the hollow portion 2416 of the head 2033H. Then, this light that passes through the hollow portion 2416 of the head 2033H is reflected by the mirror 2337 and enters the image sensor of the imaging unit 2035b. The imaging unit 2035b then captures a captured image including an image of the alignment mark MC2b provided on the chip CP and an image of the alignment mark MC1b provided on the substrate WT, using light incident on the imaging element. Here, the bonding device 2030 simultaneously recognizes a pair of the alignment mark MC1b on the chip CP and the alignment mark MC2b on the substrate WT by capturing images once, without moving the focus axis, using the same imaging unit 2035b.

[0073] Returning to FIG. 16 , the camera F-direction driver 2365 adjusts the focal positions of the image capture units 2035a and 2035b by driving the image capture units 2035a and 2035b in the focus direction as indicated by arrow AR221. The camera Z-direction driver 2363 drives the image capture units 2035a and 2035b in the Z-axis direction as indicated by arrow AR222. Here, the camera Z-direction driver 2363 typically moves the image capture units 2035a and 2035b so that the amount of movement of the Z-axis direction moving member 2331 in the Z-axis direction is the same as the amount of movement of the image capture units 2035a and 2035b in the Z-axis direction. In this way, when the head 2033H moves in the Z-axis direction, the image capture target areas of the image capture units 2035a and 2035b do not change before and after the movement. However, the camera Z-direction driving unit 2363 may move the imaging units 2035a and 2035b so that the movement amount in the Z-axis direction of the imaging units 2035a and 2035b differs from the movement amount in the Z-axis direction of the Z-axis direction moving member 2331. In this case, the relative positions in the Z direction of the imaging units 2035a and 2035b and the mirror 2337 change, and therefore the portions of the chip CP and the substrate WT that are to be imaged by the imaging units 2035a and 2035b change.

[0074] The stage unit 2031 has a stage 2315 that holds the substrate WT in an orientation in which the bonding surface WTf of the substrate WT, to which the chips CP are bonded, faces vertically downward (in the -Z direction), and a stage drive unit 2320 that drives the stage 2315. The stage 2315 is a substrate holder that can move in the X direction, Y direction, and rotational direction. This makes it possible to change the relative positional relationship between the bonding unit 2033 and the stage 2315, and to adjust the bonding positions of the chips CP on the substrate WT.

[0075] As shown in FIGS. 22A and 22B , the stage driver 2320 is a substrate holder driver having an X-direction moving unit 2311, a Y-direction moving unit 2313, an X-direction driver 2321, and a Y-direction driver 2323. The X-direction moving unit 2311 is fixed to the base member 2302 of the bonding device 2030 via two X-direction drivers 2321. The two X-direction drivers 2321 each extend in the X direction and are spaced apart in the Y direction. The X-direction driver 2321 has a linear motor and a slide rail, and moves the X-direction moving unit 2311 in the X direction relative to the fixed member 2301. The Y-direction moving unit 2313 is arranged below the X-direction moving unit 2311 (in the −Z direction) via two Y-direction drivers 2323. The two Y-direction drivers 2323 each extend in the Y direction and are spaced apart in the X direction. The Y-direction driving unit 2323 has a linear motor and a slide rail, and moves the Y-direction moving unit 2313 in the Y direction relative to the X-direction moving unit 2311. The stage 2315 is fixed to the Y-direction moving unit 2313.

[0076] The stage 2315 moves in the X and Y directions in accordance with the movements of the X-direction drive unit 2321 and the Y-direction drive unit 2323. An opening 2312 that is rectangular in plan view is provided in the center of the X-direction movement unit 2311, and an opening 2314 that is also rectangular in plan view is provided in the center of the Y-direction movement unit 2313. An opening 2316 that is circular in plan view is provided in the center of the stage 2315. The marks on the substrate WT are recognized by the imaging unit 2041 through these openings 2312, 2314, and 2316. Note that an infrared irradiator (not shown) may be provided to irradiate the substrate WT with infrared rays, thereby heating the substrate WT.

[0077] The imaging unit 2041 is, for example, an infrared camera, and is disposed above the stage 2315 as shown in FIGS. 14 and 16 . The imaging unit 2041 captures an image of the alignment marks (MC1a and MC1b in FIG. 20B ) on the substrate WT from vertically above (in the +Z direction) the substrate WT, with the chip CP positioned at the position on the substrate WT where the chip CP will be bonded. The imaging unit 2041 also generates a captured image signal indicating a captured image including an image of the alignment marks on the substrate WT, and outputs the captured image signal to the control unit 2009. Based on the captured image captured by the imaging unit 2041, the control unit 2009 recognizes the relative position of the bonding position of the chip CP with respect to the head 2033H in a direction parallel to the surface of the substrate WT where the chip CP will be bonded. The imaging unit 2041 includes an image sensor 2418, an optical system 2419, and a coaxial illumination system (not shown). The image capturing unit 2041 captures an image using reflected light of illumination light (for example, infrared light) emitted from a light source (not shown) of a coaxial illumination system.

[0078] The control unit 2009 is a control system having, for example, a personal computer, and includes a CPU and memory. The memory stores programs executed by the CPU. The memory also stores information indicating the first distance and the second distance, which will be described later. The control unit 2009 is connected to the supply tip imaging unit 2015, imaging units 2035a and 2035b, imaging unit 2041, Z-direction driving unit 2034, θ-direction driving unit 2037, piezo actuator 2333, support unit driving unit 2432b, X-direction driving unit 2321, Y-direction driving unit 2323, plate driving unit 2392, arm driving unit 2395, pickup mechanism 2111, and holding frame driving unit 2113. The control unit 2009 converts the captured image signals input from the supply tip imaging unit 2015, imaging units 2035a and 2035b, and imaging unit 2041 into captured image information and acquires it. In addition, the control unit 2009 executes the program stored in the memory, thereby outputting control signals to the Z-direction drive unit 2034, the θ-direction drive unit 2037, the piezoelectric actuator 2333, the support unit drive unit 2432b, the X-direction drive unit 2321, the Y-direction drive unit 2323, the plate drive unit 2392, the arm drive unit 2395, the pickup mechanism 2111, and the holding frame drive unit 2113, thereby controlling their operation.

[0079] The control unit 2009 calculates the relative position error between the substrate WT and the chip CP from the captured image of the alignment marks MC1a, MC1b, MC2a, and MC2b while the substrate WT and the chip CP are in contact with each other. For example, as shown in FIG. 20C , the control unit 2009 recognizes the positions of a set of alignment marks MC1a and MC2a provided on the chip CP and the substrate WT based on the captured image Ga, and calculates the misalignment amounts Δxa and Δya between the alignment marks MC1a and MC2a by the vector correlation method. Similarly to the above, the control unit 2009 recognizes the positions of a set of alignment marks MC1b and MC2b provided on the chip CP and the substrate WT based on the captured image including the image of the alignment mark MC2b provided on the chip CP and the image of the alignment mark MC1b provided on the substrate WT, and calculates the misalignment amounts Δxb and Δyb between the alignment marks MC1b and MC2b by the vector correlation method. The control unit 90 then calculates the horizontal positional deviation amounts Δx, Δy, and Δθ of the chip CP relative to the substrate WT based on the positional deviation amounts Δxa, Δya, Δxb, and Δyb. The control unit 2009 then controls the Z-direction drive unit 2034 and θ-direction drive unit 2037 of the head drive unit 2036 and the X-direction drive unit 2321 and Y-direction drive unit 2323 of the stage 2031 to correct the position and orientation of the chip CP relative to the substrate WT, in accordance with the calculated positional deviation amounts. The control unit 2009 also controls the holder frame drive unit 2113 to correct the position and tilt of the sheet holder frame 2112 about the Z-axis, in accordance with the position and orientation of the chip CP cut out by the pickup mechanism 2111. Here, the control unit 2009 recognizes the position and orientation of the chip CP based on the image captured by the supply chip imaging unit 2015.

[0080] Next, the operation of the chip bonding system 2 according to this embodiment will be described with reference to FIGS. 23 to 25. This chip bonding system 2 continuously bonds, for example, multiple chips CP to one substrate WT, and multiple chips CP are sequentially supplied from the chip supply device 2010 to the bonding device 2030. First, in the chip bonding system 2, the chip transport device 2039 faces the plate 2391 toward the chip supply unit 2011. Next, the chip supply device 2010 moves the pickup mechanism 2111 vertically upward, thereby cutting out one chip CP from the side of the sheet TE opposite the side containing the multiple chips CP, and releasing the single chip CP from the sheet TE. In this state, the chip transport device 2039 protrudes the arm 2394 from the plate 2391. 23A , the needle 2111a of the pickup mechanism 2111 is placed between the two legs 2393a of the chip holding part 2393, and the cut chip CP is placed vertically above the chip holding part 2393. Then, the chip supply device 2010 moves the pickup mechanism 2111 vertically downward, thereby transferring the chip CP from the pickup mechanism 2111 to the chip holding part 2393.

[0081] Next, the chip transport device 2039 pivots the plate 2391 to position the chip holding portion 2393 at the tip of the arm 2394 of the plate 2391 at the transfer position Pos1 vertically above the head 2033H of the bonding unit 2033. That is, the chip transport device 2039 transports the chip CP received from the chip supply device 2010 to the transfer position Pos1 where the chip CP is transferred to the head 2033H. Then, the head drive unit 2036 of the bonding device 2030 moves the bonding unit 2033 vertically upward to bring the head 2033H closer to the chip holding portion 2393 of the chip transport device 2039. Next, the support drive unit 2432b moves the chip support portion 2432a vertically upward. As a result, the chip CP held by the chip holder 2393 is positioned vertically above the chip holder 2393, while being supported by the upper end of the chip support 2432a, as shown in FIG. 23B . Next, the chip transport device 2039 retracts the arm 2394 into the plate 2391. Thereafter, the support driver 2432b moves the chip support 2432a vertically downward. This results in the chip CP being held by the tip of the head 2033H. At this time, the distance between the alignment marks MC2a and MC2b of the chip CP and the alignment marks MC1a and MC1b of the substrate WT is a second distance longer than a preset first distance that falls within the depth of field of the imaging units 2035a and 2035b. Then, the chip supply device 2010 and the chip transport device 2039 repeat the above-mentioned operations, whereby a plurality of chips CP are successively supplied to the bonding device 2030 .

[0082] Next, the bonding device 2030 bonds the chip CP to the substrate WT. Here, the bonding method by which the bonding device 2030 bonds the chip CP to the substrate WT will be described with reference to FIGS. 24 and 25 . The substrate WT and the chip CP are each vibrating in a direction perpendicular to the vertical direction as indicated by arrows AR2011 and AR2012 in FIG. 21 . The imaging units 2035a and 2035b vibrate in both the vertical direction and a direction perpendicular to the vertical direction, but it is the vertical vibration indicated by arrow AR2013 that affects the measurement of the misalignment amount. It is also assumed that the substrate WT and the chip CP are each vibrating with a vibration waveform having a frequency of 10 Hz or less or a vibration amplitude of 1 μm or less. First, as shown in FIG. 24 , the bonding device 2030 measures the relative misalignment amount between the substrate WT and the chip CP (step S201). Here, the bonding device 2030 transfers the chip CP from the chip transport device 39 to the head 33H, and then, with the chip CP and the substrate WT spaced apart by the second distance, images of the alignment marks MC2a and MC2b on the chip CP and the alignment marks MC1a and MC1b on the substrate WT are captured by the imaging units 2035a and 2035b. The imaging units 2035a and 2035b capture images of the alignment marks MC2a and MC2b on the chip CP, and the imaging unit 2041 captures images of the alignment marks MC1a and MC1b on the substrate WT. That is, the imaging units 2035a, 2035b and the imaging unit 2041 capture images of the alignment marks MC2a and MC2b on the chip CP and the alignment marks MC1a and MC1b on the substrate WT separately. Then, the bonding device 2030 measures the relative positional deviation between the substrate WT and the chip CP from the captured images of the alignment marks MC1a and MC1b and the captured images of the alignment marks MC2a and MC2b captured separately by the imaging units 2035a, 2035b and 2041 (step S201).

[0083] Next, the bonding device 2030 aligns the chip CP with the substrate WT by moving the stage 2315 horizontally relative to the head 2033H so as to eliminate the measured misalignment amounts (step S202). Here, the bonding device 2030 moves the stage 2315 in the X direction, Y direction, and rotational direction around the Z axis while keeping the head 2033H fixed so as to eliminate the misalignment amounts Δx, Δy, and Δθ. That is, the bonding device 2030 moves the chip CP horizontally relative to the substrate WT, i.e., in a direction parallel to its bonding surface WTf, based on the calculated misalignment amounts.

[0084] Next, the bonding device 2030 moves the head 2033H holding the chip CP vertically upward to bring the chip CP even closer to the substrate WT (step S203). At this time, as shown in Figure 25A, the bonding device 2030 brings the chip CP closer to the substrate WT until the distance between the alignment marks MC2a and MC2b of the chip CP and the alignment marks MC1a and MC1b of the substrate WT is a predetermined first distance G2 that falls within the depth of field of the imaging units 2035a and 2035b. Here, the first distance G2 is set to a distance in the range of 10 μm to 100 μm, for example.

[0085] Thereafter, the bonding device 2030 executes a positional deviation measurement process in which, with the chip CP and the substrate WT spaced apart by a first distance G2, it repeatedly measures the positional deviation of the chip CP relative to the substrate WT during a preset waveform measurement period (step S204). The length of the waveform measurement period is set to be at least longer than the vibration period of the chip CP relative to the substrate WT, for example, 1 sec or more. Furthermore, the bonding device 2030 repeatedly measures the positional deviation at preset time intervals shorter than the vibration period of the chip CP relative to the substrate WT. In the bonding device 2030, the alignment marks MC1a and MC2a and the alignment marks MC1b and MC2b are spaced apart by a distance G2 that falls within the depth of field of the imaging units 2035a and 2035b, and the imaging units 2035a and 2035b are disposed at positions where they can capture images of the pair of alignment marks MC1a and MC2a and the pair of alignment marks MC1b and MC2b, respectively. The bonding device 1 controls the imaging units 2035a and 2035b so that the imaging units 2035a and 2035b simultaneously capture images of the corresponding pair of alignment marks MC1a and MC2a and the corresponding pair of alignment marks MC1b and MC2b in a single image capture at the same timing during each repeated measurement. The bonding apparatus 2030 then calculates the amount of misalignment of the substrate W2 with respect to the substrate W1 based on the captured images taken by the imaging units 2035a and 2035b. Here, the bonding apparatus 2030 calculates, for example, the amount of misalignment WD1 of the alignment mark MC1a with respect to the alignment mark MC2a or the amount of misalignment WD2 of the alignment mark MC1b with respect to the alignment mark MC2b, as shown in FIG.

[0086] 24 , the bonding apparatus 2030 then executes a vibration waveform specification process (step S205) to specify the vibration waveform of the vibration of the chip CP relative to the substrate WT from the time transition of the misalignment amount measured during the waveform measurement period. Here, the bonding apparatus 2030 specifies the vibration waveform of the vibration of the chip CP in the horizontal and rotational directions relative to the substrate WT from the time transition of the misalignment amount in the horizontal and rotational directions. Then, the bonding apparatus 2030 specifies the misalignment amounts Δx, Δy, Δθ, vibration amplitude, and vibration period corresponding to the vibration center of the vibration waveform.

[0087] Next, the bonding device 2030 determines whether all of the misalignments Δx, Δy, and Δθ corresponding to the vibration centers of the identified vibration waveforms are equal to or less than the preset misalignment thresholds Δxth, Δyth, and Δθth (step S206). Assume that the bonding device 2030 determines that any of the misalignments Δx, Δy, and Δθ corresponding to the vibration centers of the identified vibration waveforms is greater than the preset misalignment thresholds Δxth, Δyth, and Δθth (step S206: No). In this case, the bonding device 2030 calculates a correction amount for the chip P relative to the substrate WT to reduce all of the misalignments Δx, Δy, and Δθ corresponding to the vibration centers of the identified vibration waveforms to the misalignment thresholds Δxth, Δyth, and Δθth or less (step S207). The control unit 2009 then calculates a correction amount for moving the chip P relative to the substrate WT by the misalignments Δx, Δy, and Δθ corresponding to the vibration centers of the identified vibration waveforms in the direction opposite to the misalignment direction.

[0088] Next, the bonding device 2030 performs alignment to correct the relative positional deviation amounts Δx, Δy, and Δθ of the chip CP with respect to the substrate WT (step S208). Here, the bonding device 2030 moves the head 2033H in the X direction, Y direction, and rotational direction around the Z axis by the correction movement amount calculated in step S207, while the position of the substrate WT is fixed. In this way, the bonding device 2030 adjusts the relative position of the chip CP with respect to the substrate WT so that the positional deviation amounts Δx, Δy, and Δθ are reduced, while the chip CP and the substrate WT are spaced apart from each other. Then, the bonding device 2030 again performs the process of step S204.

[0089] On the other hand, suppose that the bonding device 2030 determines in step S206 that all of the calculated misalignment amounts Δx, Δy, and Δθ are equal to or less than the preset misalignment amount thresholds Δxth, Δyth, and Δθth (step S206: Yes). In this case, the bonding device 2030 executes a timing estimation process based on the identified vibration waveform to estimate a target timing at which the misalignment amount of the chip CP relative to the substrate WT will become a target amount, i.e., a misalignment amount corresponding to the vibration center of the vibration waveform (step S209). Here, the bonding device 2030 estimates the target timing at which the misalignment amount of the chip CP relative to the substrate WT will become a misalignment amount corresponding to the vibration center of the vibration waveform.

[0090] Thereafter, the bonding device 2030 executes a contacting step of bringing the chip CP into contact with the substrate WT based on the estimated target timing (step S210), and then executes a bonding step of pressing the chip CP directly against the substrate WT to bond the chip CP to the substrate WT (step S211).

[0091] Next, the chip bonding system 2 moves the head 33H vertically downward, and then uses the imaging unit 2041 to capture an image including at least the alignment marks MC2a and MC2b, thereby determining whether the chip CP has been bonded to the substrate WT (step S212). Here, the bonding device 2030 determines that the chip CP has been bonded to the substrate WT if the captured image captured by the imaging unit 2041 includes the alignment marks MC1a and MC1b as well as the alignment marks MC2a and MC2b. On the other hand, the bonding device 2030 determines that the chip CP has not been bonded to the substrate WT if the captured image captured by the imaging unit 2041 does not include the alignment marks MC1a and MC1b. When the bonding device 2030 determines that the chip CP has been bonded to the substrate WT (step S212: Yes), it proceeds to the process of bonding the next chip CP to the substrate WT.

[0092] On the other hand, it is assumed that the bonding device 2030 determines that the chip CP is not bonded to the substrate WT (step S212: No). In this case, the bonding device 2030 determines the holding state of the chip CP by the head 2033H based on whether or not the alignment marks MC2a and MC2b are included in the captured images captured by the imaging units 2035a and 2035b (step S213). Here, if the alignment marks MC2a and MC2b are included in the captured images captured by the imaging units 2035a and 2035b, the bonding device 2030 determines that the chip CP is being held by the head 2033H. On the other hand, if the alignment marks MC2a and MC2b are not included in the captured images captured by the imaging units 2035a and 2035b, the bonding device 2030 determines that the chip CP is not being held by the head 2033H. Then, the bonding device 2030 determines the holding state of the chip CP by the head 33H, and then proceeds to the process of bonding the next chip CP to the substrate WT.

[0093] Furthermore, when the chip bonding system 2 determines that the chip CP has not been bonded to the substrate WT by the bonding device 2030 and that the chip CP is being held by the head 2033H, the chip bonding system 2 transfers the chip CP held by the head 2033H back to the chip holding part 2393 of the chip transport device 2039. Thereafter, the chip bonding system 2 rotates the plate 2391 to position the chip holding part 2393 vertically above a chip recovery part (not shown). Then, the chip bonding system 2 recovers the chip CP held by the chip holding part 2393 to the chip recovery part. Thereafter, the series of operations from steps S201 to S213 are repeatedly executed each time a new chip CP is transferred from the chip transport device 2039 to the bonding device 2030.

[0094] Here, the bonding device 2030 causes particles generated when bonding the chip CP to the substrate WT to fall vertically downward, and moves the stage 2315 holding the substrate WT so that the portions of the substrate WT on which each of the multiple chips CP is mounted are sequentially positioned vertically above the chip CP held by the head 2033H, thereby continuously bonding the multiple chips CP.

[0095] As described above, the chip bonding system 2 according to this embodiment identifies the vibration waveform of the vibration of the chip CP relative to the substrate WT from the time progression of the misalignment amount of the chip CP relative to the substrate WT, and estimates the target timing at which the misalignment amount of the chip CP relative to the substrate WT will reach the target amount based on the identified vibration waveform.The chip bonding system 2 then brings the chip CP into contact with the substrate WT based on the estimated target timing.This allows the chip CP to be bonded to the substrate WT with high positional accuracy, even if the chip CP is vibrating at a relatively low frequency relative to the substrate WT.

[0096] Furthermore, according to the chip bonding system 2 of this embodiment, the alignment marks MC1a, MC1b, MC2a, and MC2b are simultaneously imaged by the imaging units 2035a and 2035b while the alignment marks MC1a, MC1b, MC2a, and MC2b are spaced apart by a first distance G1 that falls within the depth of field of the imaging units 2035a and 2035b. The chip bonding system 2 then calculates the relative positional misalignment between the substrate WT and the chip CP from the images of the alignment marks MC1a, MC1b, MC2a, and MC2b captured by the imaging units 2035a and 2035b. This reduces the influence on the calculated positional misalignment due to vibration and positional changes over time of the imaging units 2035a and 2035b, or changes over time in the positions of the alignment marks MC1a, MC1b, MC2a, and MC2b due to thermal expansion of the substrate WT and the chip CP. Therefore, the chip CP can be bonded to the substrate WT with high positional accuracy.

[0097] 26A and 26B , in a conventional chip bonding system according to Comparative Example 2, an imaging unit 9035a for imaging alignment marks MC2a and MC2b provided on a chip CP arranged vertically above and an imaging unit 9035b for imaging alignment marks MC1a and MC1b provided on a substrate WT arranged vertically below are separate, and it has been common to image the pair of alignment marks MC2a and MC1a and the pair of alignment marks MC2b and MC1b in sequence while moving a mirror 9337 horizontally between the chip CP and the substrate WT. First, the mirror 9337 is positioned as shown in FIG. 26A to image the pair of alignment marks MC2a and MC1a, and then the mirror 9337 is moved as shown by arrow AR900 in FIG. 26B , and then the pair of alignment marks MC2b and MC1b is imaged. However, in the chip bonding system according to this comparative example, the relative positions of the two imaging units 9035a and 9035b may shift due to thermal expansion or changes over time. In this case, the alignment accuracy between the chip CP and the substrate WT decreases, necessitating calibration of the positions of the imaging units 9035a and 9035b. Furthermore, there is a risk that the operating state of the chip bonding system may become unstable during daily operation, causing the positions of the imaging units 9035a and 9035b to change. Furthermore, in the chip bonding system according to the comparative example, the mirror 9337 is moved while sequentially capturing images of the pair of alignment marks MC2a and MC1a and the pair of alignment marks MC2b and MC1b, resulting in a time difference between the timing of capturing images of the pair of alignment marks MC2a and MC1a and the pair of alignment marks MC2b and MC1b. This causes vibrations of the imaging units 9035a and 9035b to affect the calculated misalignment amount.

[0098] In contrast, in the chip bonding system 2 according to the present embodiment, a set of alignment marks MC2a (MC2b) provided on the chip CP and a set of alignment marks MC1a (MC1b) provided on the substrate WT are simultaneously imaged by a single imaging unit 35a (35b), so that the system is not affected by misalignment of the imaging unit 35a (35b). Also, in the chip bonding system 2, the imaging units 35a and 35b are arranged corresponding to the set of alignment marks MC2a and MC1a and the set of alignment marks MC2b and MC1b, respectively, and the two sets are simultaneously imaged, so that the effects of vibrations of the imaging units 35a and 35b can be canceled out.

[0099] Furthermore, in the chip bonding system according to the comparative example, the pair of alignment marks MC2a and MC1a and the pair of alignment marks MC2b and MC1b are imaged at different times, which causes errors due to a difference in the timing of imaging when the chip CP and the substrate WT are vibrating. In contrast, in the chip bonding system 2, the pair of alignment marks MC2a and MC1a and the pair of alignment marks MC2b and MC1b are imaged simultaneously by the two imaging units 35a and 35b. This has the advantage of preventing errors due to a difference in the timing of imaging, as occurs in the chip bonding system according to the comparative example.

[0100] Furthermore, with the above-described configuration, it is not possible to correct misalignment of the chip CP with respect to the substrate WT when bonding the chip CP to the substrate WT. In contrast, with the chip mounting system according to this embodiment, the chip CP and the substrate WT are heated to a bonding temperature using infrared rays, and the amount of misalignment between the chip CP and the substrate WT can be calculated in a state in which the chip CP and the substrate WT are thermally expanded. Therefore, regardless of the states of the chip CP and the substrate WT, the amount of misalignment can always be calculated with high accuracy, and the misalignment of the chip CP with respect to the substrate WT can be corrected.

[0101] In a configuration in which the head 2033H holding the chip CP approaches the substrate WT from vertically above the substrate WT to bond the chip CP, particles generated from the chip CP during bonding fall around the chip CP on the substrate WT. As a result, when multiple chips CP are bonded consecutively around one another as described above, the next chip CP is bonded in the area where the particles fell, which can result in particles getting caught between the chip CP and the substrate WT and causing voids or bonding defects. In contrast, in the bonding device 2030 according to this embodiment, particles generated during bonding of the chip CP fall vertically downward from the chip CP and do not adhere to the bonding surface of the substrate WT. Therefore, when multiple chips CP are bonded consecutively around one another as described above, the next chip CP can be bonded to a clean, particle-free area on the bonding surface of the substrate WT around the chip CP bonded to the substrate WT. Therefore, it is possible to suppress the occurrence of voids between the chip CP and the substrate WT and to bond the chip CP to the substrate WT well.

[0102] Furthermore, if a chip CP deviates from its intended bonding position to the bonding position of the next chip CP to be bonded, it may interfere with the displaced chip CP while being moved at high speed to a position before the next chip CP is pressed. In this case, the substrate may crack. This is a major problem because all of the chips CP bonded to the substrate WT up to that point are wasted. In contrast, in the chip bonding system 2 according to this embodiment, the stage 2315 of the bonding device 2030 holds the substrate WT in an orientation in which the bonding surface WTf of the substrate WT faces vertically downward. As a result, for example, if an attempt is made to bond a chip CP to the substrate WT but the chip CP is not bonded to the substrate WT, the chip CP is released from the substrate WT as the head 2033H moves vertically downward. This prevents chips CP remaining on the bonding surface WTf of the substrate WT from being blown to the location where other chips CP are to be bonded and interfering with the other chips CP when the other chips CP are bonded to the substrate WT.

[0103] Furthermore, after the process of bonding the chip CP to the substrate is completed, the chip bonding system 2 according to the present embodiment determines whether the chip CP has been bonded to the substrate WT, depending on whether the alignment marks MC2a and MC2b are included together with the alignment marks MC1a and MC1b in the captured image taken by the imaging unit 2041. Furthermore, after the process of bonding the chip CP to the substrate WT is completed, the chip bonding system 2 determines whether the chip CP is held by the head 2033H, depending on whether the alignment marks MC2a and MC2b are included in the captured image taken by the imaging units 2035a and 2035b. As a result, if it is determined that the chip CP that could not be bonded to the substrate WT is held by the head 2033H, the chip CP can be recovered.

[0104] Although the embodiments of the present invention have been described above, the present invention is not limited to the configurations of the above-described embodiments. For example, the stage 401 and head 402 may not have the pressing mechanisms 431, 432, and the head 402 holding the substrate W2 may be brought closer to the stage 401 holding the substrate W1, and the entire bonding surfaces of the substrates W1 and W2 may be brought into surface contact with each other at the target timing described above.

[0105] In the first embodiment, after the first contacting step in which the central portions of the substrates W1 and W2 are brought into contact with each other, the bonding wave spreads toward the peripheries of the substrates W1 and W2 in the second contacting step, and the bonding surfaces of the substrates W1 and W2 are brought into full surface-to-surface contact with each other. In contrast, according to this modification, the bonding apparatus can stop the vibration of the substrate W2 relative to the substrate W1 when the bonding surfaces of the substrates W1 and W2 are brought into surface-to-surface contact at the target timing. Therefore, by appropriately adjusting the time from when the bonding surfaces of the substrates W1 and W2 are separated to when they are brought into surface-to-surface contact at which vibration stops, the effect of the vibration on the positional deviation of the substrate W2 relative to the substrate W1 can be further reduced. Note that when the bonding surfaces of the relatively large substrates W1 and W2 are brought into full surface-to-surface contact with each other, the substrates W1 and W2 are brought into contact and pressurized with each other in a relatively short time, which may cause distortion in the bonded portions of the bonded substrates W1 and W2. Therefore, as explained in the first embodiment, it is preferable to bring the centers of the substrates W1 and W2 into contact with each other, stop the vibration of the substrate W2 relative to the substrate W1, and then allow the bonding wave to naturally spread to the peripheries of the substrates W1 and W2 over time to make full-surface contact and bond them. In contrast, when a chip CP with a relatively small area is brought into surface-to-surface contact with the substrate WT, the entire bonding surface of the chip CP comes into contact with the substrate WT at approximately the same time, and distortion is small due to the relatively small area. Therefore, a configuration in which the chip CP is brought into surface-to-surface contact with the substrate WT may also be used.

[0106] In the first embodiment, the bonding apparatus 1 captures images of at least one pair of alignment marks MK1a and MK2a on the substrate W1 and alignment marks MK1b and MK2b on the substrate W2, identifies misalignment amounts Δx, Δy, and Δθ corresponding to the vibration centers of the vibration waveforms of the alignment marks MK1a and MK2a, and brings the substrates W1 and W2 into contact with each other when the misalignment amounts Δx, Δy, and Δθ reach target values. The bonding apparatus 1 also performs a vibration waveform specifying step to specify misalignment amounts Δx, Δy, and Δθ corresponding to the vibration centers of the vibration waveforms of the substrate W2 in the horizontal and rotational directions relative to the substrate W1, as well as the vibration amplitude and vibration period. However, the present invention is not limited to this. Instead of the vibration waveform specifying step, the bonding apparatus 1 may calculate misalignment intermediate values ​​corresponding to the intermediate values ​​of the multiple misalignment amounts Δx, Δy, and Δθ measured in the misalignment measurement step, and determine whether all of the calculated misalignment intermediate values ​​are equal to or less than the misalignment thresholds Δxth, Δyth, and Δθth. In this case, if the joining device 1 determines that all of the calculated intermediate positional deviation values ​​are less than the positional deviation threshold values ​​Δxth, Δyth, and Δθth, it executes the aforementioned vibration waveform identification process to identify the vibration waveform and then executes the aforementioned timing estimation process.

[0107] Also, in the second embodiment, an example has been described in which the bonding device 2030 specifies the misalignment amounts Δx, Δy, Δθ, vibration amplitude, and vibration period corresponding to the vibration center of the vibration waveform of the vibration of the chip CP in the horizontal and rotational directions relative to the substrate WT. However, without being limited to this, the bonding device 2030 may calculate misalignment amount intermediate values ​​corresponding to the intermediate values ​​of the multiple misalignment amounts Δx, Δy, Δθ measured in the misalignment amount measurement step, and determine whether all of the calculated misalignment amount intermediate values ​​are equal to or less than the misalignment amount thresholds Δxth, Δyth, Δθth.

[0108] In the first embodiment, the bonding apparatus 1 may be configured to, in the misalignment amount measurement step, repeatedly measure the position coordinates of the alignment marks MK1a, MK2a and the alignment marks MK1b, MK2b during a preset waveform measurement period while the substrates W1, W2 are spaced apart from each other, and calculate an intermediate value of the obtained position coordinates. In this case, the bonding apparatus 1 may calculate the misalignment amounts Δx, Δy, Δθ from the difference between the intermediate value of the position coordinates of the alignment marks MK1a, MK2a and the intermediate value of the position coordinates of the alignment marks MK1b, MK2b.

[0109] In the first embodiment, the bonding apparatus 1 may be configured to repeatedly measure the position coordinates of the alignment marks MK1a, MK2a and the alignment marks MK1b, MK2b during a predetermined waveform measurement period while the substrates W1, W2 are spaced apart from each other, and then identify the vibration waveforms of the alignment marks MK1a, MK2a and the vibration waveforms of the alignment marks MK1b, MK2b from the time transition of the obtained position coordinates. The bonding apparatus 1 may then identify the position coordinates corresponding to the vibration centers of the vibration waveforms of the alignment marks MK1a, MK2a and the position coordinates corresponding to the vibration centers of the vibration waveforms of the alignment marks MK1b, MK2b, and calculate the positional deviations Δx, Δy, and Δθ based on the identified position coordinates of the alignment marks MK1a, MK2a and the alignment marks MK1b, MK2b, respectively. In this case, the accuracy of calculating the positional deviation amounts Δx, Δy, Δθ can be improved compared to when calculating the positional deviation amounts Δx, Δy, Δθ from the intermediate values ​​of the position coordinates of the aforementioned alignment marks MK1a, MK2a and alignment marks MK1b, MK2b.

[0110] In the first embodiment, an example has been described in which the distance measurement unit 490 measures the distance between the stage 401 and the head 402. However, this is not limiting, and the distance measurement unit may measure the distance between the substrate W1 held on the stage 401 and the head 402, or the distance between the stage 401 and the substrate W2 held by the head 402. Alternatively, the distance measurement unit may measure the distance between the substrate W1 held on the stage 401 and the substrate W2 held by the head 402.

[0111] In the first embodiment, an example has been described in which the bonding apparatus 1 executes the above-described attitude adjusting step every time the alignment step is performed. However, the present invention is not limited to this, and the attitude adjusting step may be executed, for example, every time the bonding of the substrates W1 and W2 is performed a predetermined number of times, or once every time a predetermined period has elapsed.

[0112] In the first embodiment, an example in which the substrates W1 and W2 are bonded together under reduced pressure has been described, but the present invention is not limited to this, and the substrates W1 and W2 may be bonded together under atmospheric pressure.

[0113] In the second embodiment, the head 2033H may have a distance measurement unit that measures the distance between multiple locations on the bonding surface CPf of the chip CP and locations on the bonding surface WTf of the substrate WT that face the multiple locations. Here, the distance measurement unit is, for example, a laser rangefinder that measures the distance between the head 2033H and the underside of the substrate WT without contacting the head 2033H or the stage 2315. Furthermore, the distance measurement unit measures the distance between the head 2033H and the underside of the substrate WT from the difference between the light reflected from the underside of the substrate WT and the light reflected from the tip surface of the tip tool 2411 when laser light is irradiated from below the transparent tip tool 2411 toward the substrate WT held on the stage 2315. The control unit 2009 may then control the three piezo actuators 2333 mentioned above based on the distance measured by the distance measurement unit so that the bonding surface CPf of the chip CP held by the head 2033H is parallel to the bonding surface WTf of the substrate WT.

[0114] In the second embodiment, as shown in FIG. 27A , a bonding device may be provided with a head 3033H provided with a pressing mechanism 3431 that presses the center of the chip CP vertically upward while holding the peripheral portion of the chip CP. Note that in FIGS. 27A and 27B , components similar to those in the embodiment are denoted by the same reference numerals as in FIGS. 17A and 17B . The head 3033H includes a tip tool 3411, a head main body 3413, a tip support 3432a, and a support driver 3432b. The tip tool 3411, like the tip tool 2411 in the second embodiment, is formed from a material (e.g., silicon (Si)) that transmits imaging light (e.g., infrared light). The head main body 3413 includes a holding mechanism 2440 and a pressing mechanism 3431 that presses the center of the chip CP. The pressing mechanism 3431 has a pressing part 3431a that is vertically movable at the center of the tip surface of the head main body part 3413, and a pressing drive part 3431b that drives the pressing part 3431a. The head main body part 3413 also has a suction part (not shown) that fixes the tip tool 3411 to the head main body part 3413 by vacuum suction. The tip tool 3411 has a through hole 2411a formed at a position corresponding to the holding mechanism 2440 of the head main body part 3413, a through hole 3411b into which the pressing part 3431a is inserted, and a through hole 3411c into which the tip support part 3432a is inserted.

[0115] The chip support portion 3432a has, for example, a pin-like shape and is provided at the tip of the head 3033H and is movable in the vertical direction. The chip support portion 3432a supports the side of the chip CP opposite the bonding surface CPf. As shown in FIG. 27B, for example, four chip support portions 3432a are provided surrounding the pressing portion 3431a. The support portion driver 2432b drives the chip support portion 3432a in the vertical direction. The support portion driver 3432b positions the chip holding portion 2393 of the chip transport device 2039 at the transfer position Pos1 while holding the chip CP, and moves the chip support portion 3432a vertically upward relative to the chip holding portion 2393 while supporting the chip CP. This transfers the chip CP from the chip holding portion 2393 of the chip transport device 2039 to the head 3033H.

[0116] As shown in Figure 28, when the pressing unit 3431a is moved vertically upward while the peripheral portion of the chip CP is held by the chip tool 3411, the pressing drive unit 3431b presses the central portion of the chip CP vertically upward (in the +Z direction), causing the central portion of the chip CP to bend vertically upward compared to its peripheral portion.

[0117] In the bonding apparatus 2 according to this modification, the peripheral portion of the chip CP is held by suction on the chip tool 3411 (see arrow AR301 in FIG. 28 ), and the pressing drive unit 3431b drives the pressing unit 3431a in the vertical direction (see arrow AR302 in FIG. 28 ). This causes the chip CP to bend so that its central portion protrudes toward the substrate WT relative to its peripheral portion. Then, as shown by arrow AR303, the head 3033H approaches the substrate WT with the chip CP bent, causing the central portion of the chip CP to contact the bonding surface WTf of the substrate WT. Note that the bonding apparatus may also move the head 3033H vertically to approach the substrate WT to a predetermined distance, and then bend the chip CP to bring the central portion of the chip CP into contact with the bonding surface WTf of the substrate WT. Thereafter, the bonding device causes the pressing portion 3431a to sink vertically downward, and moves the head 3033H even closer to the substrate WT, thereby bonding the chip CP to the substrate WT.

[0118] According to this configuration, the entrapment of air between the substrate WT and the chip CP when bonding the chip CP to the substrate WT is suppressed, making it possible to bond the chip CP to the substrate WT well without voids.

[0119] In the second embodiment, an example of the chip bonding system 2 has been described in which the chip CP is bonded to the substrate WT by bringing the bonding surface CPf of the activated chip CP into contact with the bonding surface WTf of the substrate WT. However, the present invention is not limited to this, and the chip CP may be bonded to a composite substrate having a resin layer formed on one surface of the substrate WT, for example.

[0120] 29 , the chip joining system 2 according to the second embodiment may include a joining device having a head 4033H in which a recess 4411c is provided in a portion of the chip tool 4411 that holds the chip CP. Here, the recess 4411c is provided in a portion of the chip CP that corresponds to an area that comes into contact with the tip of the needle 2111a of the pickup mechanism 2111 of the chip supply unit 2011.

[0121] According to this configuration, it is possible to prevent the attitude of the chip CP from being tilted due to particles adhering to the contact portion of the needle 2111a of the chip CP, and therefore the chip CP can be bonded to the substrate WT in a good manner.

[0122] The chip bonding system 2 according to the second embodiment has been described as an example in which the alignment marks MC1a and MC1b on the substrate WT are captured by the imaging unit 2041 with the chip CP and the substrate WT spaced apart by the second distance before the chip CP is brought closer to the head 2033H. However, this is not limiting, and for example, the chip bonding system may capture images of the alignment marks MC1a and MC1b on the substrate WT by the imaging units 2035a and 2035b before the chip CP is held by the head 2033H. In this case, the control unit 2009 stores the captured images of the alignment marks MC1a and MC1b on the substrate WT in the main memory unit or the auxiliary memory unit. Next, the control unit 2009 images the alignment marks MC1a and MC1b on the substrate WT using the imaging units 2035a and 2035b, and then images the alignment marks MC2a and MC2b on the chip CP using the imaging units 2035a and 2035b while the chip CP is held by the head 2033H.The control unit 2009 then calculates the amount of relative positional deviation using the images of the alignment marks MC1a and MC1b on the substrate WT and the images of the alignment marks MC2a and MC2b on the chip CP obtained by imaging using the imaging units 2035a and 2035b. Furthermore, when bonding a plurality of chips CP to the substrate WT, the substrate WT is maintained in a state of being held by the stage 2315, so that the alignment marks MC1a and MC1b of the substrate WT are imaged only once when the substrate WT is transferred to the stage 2315 and stored in the memory of the control unit 2009, and when the stage 2315 moves, the positions of the alignment marks MC1a and MC1b can be predicted from the information on the alignment marks MC1a and MC1b of the substrate WT stored in the memory. In this case, when bonding the chips CP to the substrate WT, it is not necessary to image the alignment marks MC1a and MC1b of the substrate WT, and it is sufficient to image only the alignment marks MC2a and MC2b of the chips CP.

[0123] According to this configuration, even if the substrate WT is opaque, it is possible to calculate the amount of relative positional deviation between the substrate WT and the chip CP.

[0124] In the second embodiment, an example has been described in which the imaging units 2035a and 2035b each acquire an image including the alignment marks MC1a and MC1b of the chip CP and the alignment marks MC2a and MC2b of the substrate WT by using reflected light of illumination light (e.g., infrared light) emitted from a light source of a coaxial illumination system. However, this is not limiting. For example, an image including the alignment marks MC1a and MC1b of the chip CP and the alignment marks MC2a and MC2b of the substrate WT may be acquired by using transmitted light that passes through the chip CP from a light source provided on the opposite side from the imaging units 2035a and 2035b. For example, an imaging unit 2041 disposed vertically above the substrate WT may acquire an image including the alignment marks MC1a, MC1b, MC2a, and MC2b by using coaxial light from the imaging units 2035a and 2035b that enters the lower side of the chip CP. Alternatively, the imaging units 2035a and 2035b may be configured to acquire images including the alignment marks MC1a, MC1b, MC2a, and MC2b by using coaxial light emitted from an imaging unit 2041 arranged vertically above the substrate WT. Furthermore, if the substrate WT is transparent to visible light, the coaxial light emitted from the imaging units 2035a and 2035b or the imaging unit 2041 may be visible light.

[0125] Alternatively, the imaging unit 2041 may be configured to acquire an image including the alignment marks MC1a and MC1b of the chip CP and the alignment marks MC2a and MC2b of the substrate WT. In this manner, with the chip CP in contact with the bonding surface WTf of the substrate WT, the alignment marks MC1a and MC1b of the chip CP and the alignment marks MC2a and MC2b of the substrate WT are simultaneously recognized by the same imaging unit 2041 using infrared light, and the set of alignment marks MC1a, MC2a, and alignment marks MC1b, MC2b is simultaneously captured in a single capture without moving the focus axis, thereby enabling highly accurate recognition of the misalignment between the chip CP and the substrate WT. The same is true for a configuration in which the imaging units 2035a and 2035b on the chip CP side are used to simultaneously recognize the alignment marks MC1a and MC1b of the chip CP and the alignment marks MC2a and MC2b of the substrate WT.

[0126] In the second embodiment, the bonding apparatus 2030 may include a distance measurement unit (not shown) that measures the distance between the bonding surface WTf of the substrate WT and the bonding surface CPf of the chip CP at three or more locations on the bonding surface (flat surface) CPf of the chip CP. The distance measurement unit may include, for example, laser light sources (not shown) arranged at multiple locations on the side of the head 33H, and a light receiving unit (not shown) that receives laser light emitted from each of the multiple laser light sources and reflected by the substrate WT. The head driving unit 2036 may then move the head 2033H holding the chip CP closer to the stage unit 2031 holding the substrate WT based on the distance measured by the distance measurement unit. In addition, the three piezoelectric actuators 2333 may adjust at least one of the distance between the bonding surface WTf of the substrate WT and the chip CP and the inclination of the chip CP relative to the bonding surface WTf of the substrate WT based on the distance between the bonding surface WTf of the substrate WT and the bonding surface CPf of the chip CP measured by the distance measurement unit.

[0127] In embodiment 2, an example was described in which the alignment marks MC1a and MC1b of the chip CP are provided on the bonding surface CPf side, but this is not limited to this, and for example, the alignment marks MC1a and MC1b may be provided on the surface of the chip CP opposite the bonding surface CPf side.

[0128] In the second embodiment, an example has been described in which the stage 2315 is moved in a direction perpendicular to the vertical direction without moving the head 2033H in the direction perpendicular to the vertical direction, but this is not limiting, and a configuration may be used in which the head 2033H is moved in a direction perpendicular to the vertical direction without moving the stage 2315 in the direction perpendicular to the vertical direction. Alternatively, a configuration may be used in which both the head 2033H and the stage 2315 move in a direction perpendicular to the vertical direction.

[0129] In the second embodiment, the case where the alignment marks MC2a, MC2b on the substrate WT have a circular shape has been described, but the shape of the alignment marks MC2a, MC2b on the substrate WT is not limited to a circular shape and may be other shapes, such as a rectangular or triangular shape. If the alignment marks MC2a, MC2b on the substrate WT have a non-circular shape, the components in the XY directions and the rotational direction can be recognized using one alignment mark MC2a, MC2b. In this case, the amount of misalignment may be calculated by simultaneously capturing images of, for example, one alignment mark on the substrate WT and one alignment mark on the chip CP.

[0130] The present invention allows various embodiments and modifications without departing from the broad spirit and scope of the present invention. Furthermore, the above-described embodiments are intended to explain the present invention and do not limit the scope of the present invention. In other words, the scope of the present invention is defined by the claims, not the embodiments. Various modifications made within the scope of the claims and within the scope of the meaning of the invention equivalent thereto are considered to be within the scope of the present invention.

[0131] This application is based on Japanese Patent Application No. 2022-116044, filed on July 21, 2022. The entire specification, claims and drawings of Japanese Patent Application No. 2022-116044 are incorporated herein by reference.

[0132] The present invention is suitable for manufacturing, for example, CMOS image sensors, memories, computing elements, and MEMS.

[0133] 1, 2030: bonding device, 2: chip bonding system, 9, 2009: control unit, 41, 2041, 2042: stand, 120: chamber, 120a: window, 121a: vacuum pump, 121b: exhaust pipe, 121c: exhaust valve, 401, 2315: stage, 401a: upper surface, 401b, 402b, 2411a, 2411b, 3411b, 3411c, 2114a: through holes, 402, 2033H, 3033H, 4033H: head, 402a: lower surface, 403, 2320: stage driving unit, 404, 2036: head driving unit, 405: XY direction driving unit, 406: lifting and lowering Drive unit, 407: rotation drive unit, 408, 412: pressure sensor, 411, 2333: piezoelectric actuator, 431, 432, 3431: pressing mechanism, 431a, 432a, 3431a: pressing unit, 431b, 432b, 3431b: pressing drive unit, 431c, 432c: stopper, 441, 442: electrostatic chuck, 481, 482: substrate heating unit, 490: distance measurement unit, 500: imaging unit, 501, 502, 2035a, 2035b: imaging unit, 504, 505, 2337: mirror, 2010: chip supply device, 2011: chip supply unit, 2015: supplied chip Imaging unit, 2031: stage unit, 2033: bonding unit, 2034: Z-direction driving unit, 2037: θ-direction driving unit, 2038: linear guide, 2039: chip transport device, 2041: imaging unit, 2111: pickup mechanism, 2111a: needle, 2112: sheet holding frame, 2113: holding frame driving unit, 2114: cover, 2119: frame holding unit, 2301: fixing member, 2302: base member, 2311: X-direction moving unit, 2312, 2314, 2316: opening, 2313: Y-direction moving unit, 2321: X-direction driving unit, 2323: Y-direction driving unit, 233 1: Z-axis direction moving member, 2332: first disc member, 2334: second disc member, 2334a, 2334b: holes, 2336: mirror fixing member, 2337a, 2337b: inclined surface, 2351a, 2351b, 2418: image sensor, 2352a, 2352b, 2419: optical system, 2361: rotating member, 2363: camera Z direction driving unit, 2365: camera F direction driving unit, 2391: plate, 2392: plate driving unit, 2393: tip holder, 2394: arm, 2395: arm driving unit, 2411, 3411, 4411: tip tool, 2413,3413: head main body, 2415, 2416: hollow portion, 2432a, 3432a: chip support portion, 2432b, 2512, 3432b: support portion drive portion, CP: chip, CPf: bonding surface, Ga, Gb: photographed image, TE: sheet, MC1a, MC1b, MC2a, MC2b, MK1a, MK1b, MK2a, MK2b: alignment marks, W1, W2, WT: substrate, W1c, W2c: central portion, W1s, W2s: peripheral portion, WTf: bonding surface,

Claims

1. A bonding method for bonding a first object to a second object that is vibrating relatively to the first object, comprising: a vibration waveform specifying step of specifying a vibration waveform of the vibration of the second workpiece relative to the first workpiece from a time transition of a positional deviation amount of the second workpiece relative to the first workpiece; a timing estimation step of estimating a target timing at which an amount of misalignment of the second workpiece with respect to the first workpiece becomes a target amount, based on the vibration waveform; a contacting step of contacting the second workpiece with the first workpiece based on the estimated target timing; a positional deviation measuring step of repeatedly measuring a positional deviation of the second object relative to the first object, the first object to be bonded has at least one first alignment mark; the second object to be bonded has at least one second alignment mark; In the positional deviation amount measuring step, a positional deviation amount of the first alignment mark with respect to the second alignment mark is repeatedly measured based on an image of the first alignment mark and an image of the second alignment mark simultaneously captured within one field of view by an imaging unit. Joining method.

2. A method for joining a first workpiece to a second workpiece vibrating with a vibration waveform including a vibration component having a frequency of 10 Hz or less or an amplitude of 10 μm or less relative to the first workpiece, comprising: a vibration waveform specifying step of specifying a vibration waveform of the vibration of the second workpiece relative to the first workpiece from a time transition of a positional deviation amount of the second workpiece relative to the first workpiece; a timing estimation step of estimating a target timing at which an amount of misalignment of the second workpiece with respect to the first workpiece becomes a target amount, based on the vibration waveform; a contacting step of contacting the second workpiece with the first workpiece based on the estimated target timing. Joining method.

3. a positional deviation measuring step of repeatedly measuring a positional deviation of the second object relative to the first object; a correction movement step of moving the second workpiece relatively to the first workpiece to correct the position of the second workpiece with respect to the first workpiece based on the amount of positional deviation, The joining method according to claim 2 .

4. In the positional deviation amount measuring step of repeatedly measuring the positional deviation amount of the second workpiece relative to the first workpiece and / or the vibration waveform specifying step, the positional deviation amount is repeatedly measured at a time interval shorter than a vibration period of the second workpiece relative to the first workpiece. The bonding method according to any one of claims 1 to 3.

5. in the contacting step, an operation for bringing the first object into contact with the second object is started at a time point a necessary time required for bringing the first object into contact with the second object from a state in which the first object is separated from the second object, prior to the timing at which the positional deviation amount reaches the target amount. The bonding method according to any one of claims 1 to 3.

6. The first object to be bonded is a substrate, In the contacting step, the first object is bent so that a central portion of a joining surface of the first object to be joined with the second object to be joined protrudes toward the second object to be joined compared with a peripheral portion of the first object to be joined, and the central portion of the joining surface of the first object to be joined with the second object to be joined is brought into contact with the joining surface of the second object to be joined with the first object to be joined. The bonding method according to any one of claims 1 to 3.

7. In the position deviation amount measuring step, the vibration waveform specifying step, the timing estimating step, and the contacting step, a top plate on which a first object holding part for holding the first object and a second object holding part for holding the second object are disposed vertically above, and a plate support part having a vibration isolation mechanism and supporting the top plate vertically above so as to be freely movable is used to detect vibration transmitted to the top plate, and the top plate is moved relatively to the plate support part based on the detected vibration; The joining method according to claim 1 .

8. a posture adjusting step of measuring distances between the first object to be welded or a plurality of portions of a first object to be welded holding part that holds the first object to be welded and portions corresponding to the plurality of portions of the second object to be welded or a second object to be welded holding part that holds the second object to be welded, and adjusting a posture of the second object to the first object to be welded based on the measured distances, The bonding method according to any one of claims 1 to 3.

9. A bonding method for bonding a first object provided with a plurality of first alignment marks to a second object provided with a plurality of second alignment marks corresponding to the plurality of first alignment marks, comprising: At least one of the second object and an imaging unit that images the plurality of first alignment marks and the plurality of second alignment marks vibrates with respect to the first object with a vibration waveform including a vibration component of 10 Hz or less or an amplitude of 10 μm or less, a positional deviation amount measuring step of simultaneously capturing an image of the first alignment marks and the second alignment marks from at least one of an opposite side of the first object to the first object and an opposite side of the first object to the second object by an imaging unit in a state in which the first object is relatively close to the second object and the first alignment marks are spaced apart by a preset first distance within a range of a depth of field of the imaging unit, and measuring a relative positional deviation amount between the first object and the second object from the captured images of the first alignment marks and the second alignment marks; a vibration waveform specifying step of specifying a vibration waveform of the vibration of the second workpiece relative to the first workpiece from a time transition of a positional deviation amount of the second workpiece relative to the first workpiece; a timing estimation step of estimating a target timing at which an amount of misalignment of the second workpiece with respect to the first workpiece becomes a target amount, based on the vibration waveform; and a joining process of correcting a relative position of the second workpiece to the first workpiece based on the amount of positional deviation by relatively moving the second workpiece with respect to the first workpiece in a direction in which the amount of positional deviation becomes a target value, and then bringing the second workpiece into contact with the first workpiece and joining them based on the estimated target timing. Joining method.

10. A bonding method for bonding a first object provided with a plurality of first alignment marks to a second object provided with a plurality of second alignment marks corresponding to the plurality of first alignment marks, comprising: At least one of the second object and an imaging unit that images the plurality of first alignment marks and the plurality of second alignment marks vibrates with respect to the first object with a vibration waveform including a vibration component of 10 Hz or less or an amplitude of 10 μm or less, a positional deviation amount measuring process for imaging the plurality of first alignment marks and the plurality of second alignment marks with the imaging unit, the imaging unit being disposed in a position in which an optical axis is perpendicular to the vertical direction, and for imaging the plurality of first alignment marks and the plurality of second alignment marks by receiving light traveling in the vertical direction from the plurality of first alignment marks and the plurality of second alignment marks via an optical path conversion member that converts the light to a direction perpendicular to the vertical direction, and for measuring a relative positional deviation amount between the first object and the second object from the captured images of the plurality of first alignment marks and the plurality of second alignment marks; a vibration waveform specifying step of specifying a vibration waveform of the vibration of the second workpiece relative to the first workpiece from a time transition of a positional deviation amount of the second workpiece relative to the first workpiece; a timing estimation step of estimating a target timing at which an amount of misalignment of the second workpiece with respect to the first workpiece becomes a target amount, based on the vibration waveform; and a joining process of correcting the positional deviation amount of the second workpiece relative to the first workpiece by moving the second workpiece relative to the first workpiece in a direction in which the positional deviation amount becomes smaller based on the positional deviation amount, and then contacting and joining the second workpiece to the first workpiece based on the estimated target timing. Joining method.

11. A joining apparatus for joining a first workpiece provided with a plurality of first alignment marks and a second workpiece provided with a plurality of second alignment marks corresponding to the plurality of first alignment marks, comprising: A first object to be bonded holding section that holds the first object to be bonded; a second workpiece holding part that vibrates relative to the first workpiece holding part and holds the second workpiece; a holder driving section that moves at least one of the first object holding section and the second object holding section in a first direction in which the first object holding section and the second object holding section approach each other or in a second direction in which the first object holding section and the second object holding section move away from each other; a plurality of imaging units, each provided for each of a plurality of pairs of one first alignment mark and one second alignment mark, each imaging unit capturing an image of the pair of one first alignment mark and one second alignment mark; In a state where the first object to be bonded and the second object to be bonded are spaced apart from each other, the plurality of first alignment marks and the plurality of second alignment marks are spaced apart by a predetermined first distance within a range of a depth of field of the imaging units at predetermined time intervals, and the plurality of imaging units are disposed at positions where they can respectively image a set consisting of one first alignment mark and one second alignment mark, the plurality of imaging units simultaneously image the corresponding set consisting of one first alignment mark and one second alignment mark at the same timing by one image capture, and the captured images of the plurality of first alignment marks and the plurality of second alignment marks imaged by the plurality of imaging units are recorded. a control unit which repeatedly measures an amount of misalignment of the first object with respect to the second object in the first direction and a direction perpendicular to the second direction by measuring an amount of relative misalignment between the first object and the second object from an image of the first object, specifies a vibration waveform of vibration of the second object with respect to the first object from a time transition of the amount of misalignment, estimates a target timing at which the amount of misalignment of the second object with respect to the first object will reach a target amount based on the vibration waveform, and controls the holding unit driving unit to move at least one of the first object holding unit and the second object holding unit in the first direction to bring the second object into contact with the first object based on the estimated target timing. Bonding equipment.

12. a top plate above which a first object holding part for holding the first object and a second object holding part for holding the second object are disposed vertically; a plate support portion having a vibration isolation mechanism and movably supporting the top plate vertically above; a vibration detection unit that detects vibrations transmitted to the top plate; a plate driving unit that moves the top plate relatively to the plate support unit; and a vibration isolation control unit that controls the plate driving unit to move the top plate so as to cancel out the vibration based on the vibration detected by the vibration detection unit. The joining device according to claim 11.

13. a distance measuring unit that measures distances between a plurality of portions of a first object holding unit that holds the first object and portions corresponding to the plurality of portions of a second object holding unit that holds the second object; and a posture adjustment unit that adjusts a posture of the second object relative to the first object based on the distance measured by the distance measurement unit. The joining device according to claim 11 or 12.

14. The second workpiece is vibrated with respect to the first workpiece with a vibration waveform including a vibration component having a frequency of 10 Hz or less or an amplitude of 10 μm or less. The joining device according to claim 11 or 12.