Calibration method for a pickup device, method for manufacturing a semiconductor device, and pickup device

Abstract

This reduces defects that can occur during the process of peeling semiconductor chips from the sheet. [Solution] The calibration method includes: step S22, which involves moving the collet 61 relative to the block push-up surface 531a to position the collet 61 at a predetermined distance from the reference height H531; step S24, which involves the block motor 532 receiving a collet motor signal C64 to bring the block push-up surface 531a closer to the collet holding surface 61a; step S25, which involves the contact detection unit 65 detecting that the block push-up surface 531a has come into contact with the collet holding surface 61a; and step S3, which involves obtaining a block motor correction value R532 using the height from the reference height H531 to the collet holding surface 61a set in step S22 for positioning the collet 61, and the height of the push-up surface when it is detected that the block push-up surface 531a has come into contact with the collet holding surface 61a.

Description

Technical Field

[0001] The present invention relates to a calibration method for a pickup device, a method for manufacturing a semiconductor device, and a pickup device.

Background Art

[0002] Semiconductor chips obtained by dicing a semiconductor wafer are components of electronic devices. In the manufacturing steps of electronic devices, an operation is performed to pick up semiconductor chips attached to a wafer sheet from the wafer sheet and place them on a circuit board. For example, Patent Document 1 discloses a device for picking up semiconductor chips held by an adhesive sheet. This device peels the semiconductor chips from the adhesive sheet by gradually pushing up the semiconductor chips held by the adhesive sheet with a plurality of push-up bodies.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] When peeling a semiconductor chip from an adhesive sheet with a plurality of push-up bodies, it is necessary to precisely move each of the plurality of push-up bodies by a desired height. However, even when a signal for moving the first push-up body by a predetermined height is output from a controller and a signal for moving the second push-up body by a predetermined height is also output, the height by which the first push-up body actually moves and the height by which the second push-up body actually moves may be different. When such a difference occurs, there may be a defect in the operation of peeling the semiconductor chip from the sheet.

[0005] The present invention provides a calibration method for a pickup device that reduces defects that may occur during the operation of peeling a semiconductor chip from a sheet, a method for manufacturing a semiconductor device using the calibration method for the pickup device, and a pickup device capable of reducing defects that may occur during the operation of peeling a semiconductor chip from a sheet. [Means for solving the problem]

[0006] One embodiment of the present invention is a calibration method for a pickup device that peels semiconductor chips attached to a sheet from the sheet, comprising: a first module including a collet having a contact surface and a contact detection unit that detects when an object comes into contact with the contact surface; and a second module including a push-up block provided opposite the contact surface and a drive unit that moves the push-up surface of the push-up block closer to or further away from the contact surface. The calibration method for the pickup device includes the steps of: moving a collet relative to the thrust surface to position the contact surface at a predetermined distance from the reference height of the thrust surface; a drive unit receiving a drive signal generated from information on the thrust height of the thrust surface to bring the thrust surface closer to the contact surface positioned in the step of positioning the contact surface; a contact detection unit detecting that the thrust surface has come into contact with the contact surface; and obtaining a correction value for correcting the drive signal generated from information on the thrust height, using the height from the reference height to the contact surface set in the step of positioning the contact surface and the height of the thrust surface when it is detected that the thrust surface has come into contact with the contact surface.

[0007] This method involves bringing the push-up surface closer to the contact surface of the collet, whose position is being held, thereby bringing the push-up surface into contact with the contact surface. A correction value is then obtained to correct the control signal generated from the push-up height information for the drive unit, using the height from the push-up surface to the contact surface set in the placement step and the height of the push-up surface indicated by the control signal received by the drive unit when contact with the contact surface was detected. By using this correction value, the difference between the target push-up height and the actual push-up height caused by various factors can be eliminated, and the actual push-up height can be brought closer to the target push-up height. Therefore, defects that may occur during the operation of peeling semiconductor chips from the sheet can be reduced.

[0008] In the above method, the contact detection unit includes a sensor capable of acquiring sensor data corresponding to the contact load caused by the contact of the push-up block with the collet, and the step of detecting contact may be to detect that the push-up surface has come into contact with the contact surface, conditional on a change in the sensor data. According to this step, contact between the push-up surface and the contact surface can be detected by load. Detection by load can improve the accuracy of detection compared to detection by displacement.

[0009] In the above method, the step of detecting contact may be to detect that the thrust surface has come into contact with the contact surface, provided that the load value indicated by the sensor data decreases due to the application of the contact load.

[0010] In the above method, the step of detecting contact may be to detect that the thrust surface has come into contact with the contact surface, provided that the load value indicated by the sensor data increases due to the application of a contact load.

[0011] In the above method, the second module includes a drive mechanism including a drive unit and a push-up mechanism that includes a plurality of push-up blocks, each constituting a plurality of push-up surfaces, and is detachable from the drive mechanism, and may further include a step of replacing the first push-up mechanism with the second push-up mechanism before the step of arranging the contact surfaces. A correction value can be obtained each time the push-up mechanism is replaced. Therefore, even when the push-up mechanism is replaced, defects that may occur in the operation of peeling the semiconductor chip from the sheet can be reduced.

[0012] Another embodiment of the present invention is a method for manufacturing a semiconductor device. The method for manufacturing a semiconductor device is a pickup device for peeling a semiconductor chip attached to a sheet from the sheet, comprising: a first module including a collet having a contact surface and a contact detection unit for detecting when an object has come into contact with the contact surface; and a second module including a push-up block provided opposite the contact surface and a drive unit for moving the push-up surface of the push-up block closer to or away from the contact surface; a step of obtaining a correction value for calibrating the pickup device; a step of obtaining a corrected drive signal generated by correcting a drive signal generated from information on the push-up height of the push-up surface using the correction value; and a step of peeling the semiconductor chip attached to the sheet from the sheet by supplying the corrected drive signal to the drive unit. The steps for obtaining a correction value for calibrating the pickup device include: moving a collet relative to the push-up surface to position the contact surface at a predetermined distance from the reference height of the push-up surface; a drive unit receiving a drive signal generated from the push-up height information of the push-up surface to bring the push-up surface closer to the contact surface positioned in the step of positioning the contact surface; a contact detection unit detecting that the push-up surface has come into contact with the contact surface; and obtaining a correction value for correcting the drive signal generated from the push-up height information, using the height from the reference height to the contact surface set in the step of positioning the contact surface and the height of the push-up surface when it is detected that the push-up surface has come into contact with the contact surface.

[0013] This semiconductor device manufacturing method allows for correction of control signals for driving multiple blocks. As a result, defects that may occur during the process of peeling semiconductor chips from a sheet can be reduced.

[0014] A further embodiment of the present invention is a pickup device. The pickup device comprises a pickup unit for peeling a semiconductor chip attached to a sheet from the sheet, and a controller unit for controlling the pickup unit. The pickup unit comprises a first module including a collet having a contact surface and a contact detection unit for detecting when an object comes into contact with the contact surface, and a second module including a push-up block provided opposite the contact surface and a drive unit for moving the push-up surface of the push-up block closer to or further away from the contact surface. The controller unit performs the following operations: moving the collet relative to the thrust surface to position the contact surface at a predetermined distance from the reference height of the thrust surface; the drive unit receiving a drive signal generated from the information of the thrust height of the thrust surface and outputting a drive signal to bring the thrust surface closer to the contact surface of the first module whose position is being maintained; the contact detection unit detecting that the thrust surface has come into contact with the contact surface; and obtaining a correction value for correcting the drive signal generated from the information of the thrust height, using the height from the reference height to the contact surface set in the step of positioning the contact surface and the height of the thrust surface when it is detected that the thrust surface has come into contact with the contact surface.

[0015] This pickup device allows for correction of control signals for driving multiple blocks. As a result, defects that may occur during the process of peeling semiconductor chips from the sheet are reduced, thereby increasing the yield of manufactured products containing semiconductor chips. [Effects of the Invention]

[0016] According to the present invention, there are provided a calibration method for a pickup device that reduces defects that may occur during an operation of peeling a semiconductor chip from a sheet, a method for manufacturing a semiconductor device using the calibration method of the pickup device, and a pickup device capable of reducing defects that occur during an operation of peeling a semiconductor chip from a sheet.

Brief Description of the Drawings

[0017] [Figure 1] FIG. 1 is a diagram showing a bonding device according to an embodiment. [Figure 2] FIG. 2 is a diagram showing a pickup unit. [Figure 3] FIG. 3(a) is a diagram showing a bonding head provided with a contact detection unit as a first example. FIGS. 3(b) and 3(c) are diagrams visualizing loads acting on a coil body and a head body in the contact detection unit as the first example. [Figure 4] FIG. 4(a) is a diagram showing a bonding head provided with a contact detection unit as a second example. FIG. 4(b) is a diagram visualizing loads acting on a coil body and a head body in the contact detection unit as the second example. [Figure 5] FIG. 5(a) is a diagram showing a bonding head provided with a contact detection unit as a third example. FIG. 5(b) is a diagram visualizing loads acting on a coil body and a head body in the contact detection unit as the third example. [Figure 6] FIGS. 6(a), 6(b) and 6(c) are diagrams for explaining a pickup operation. [Figure 7] FIGS. 7(a) and 7(b) are diagrams for further explaining the pickup operation following FIG. 6. [Figure 8] FIG. 8 is a flowchart showing a method for manufacturing a semiconductor device according to an embodiment. [Figure 9] FIG. 9 is a flowchart showing a calibration method for a pickup unit according to an embodiment. [Figure 10]FIG. 10 is a functional block diagram of a controller unit included in a bonding apparatus. [Figure 11] FIGS. 11(a) and 11(b) are diagrams for explaining an operation of generating an initial load in a calibration method of a pickup unit. [Figure 12] FIGS. 12(a) and 12(b) are diagrams for explaining an operation of detecting contact of a block in a calibration method of a pickup unit. [Figure 13] FIGS. 13(a) and 13(b) are diagrams for explaining a case where the height of pushing up an instruction block matches the actual pushing-up height. FIGS. 13(c) and 13(d) are diagrams for explaining a case where the height of pushing up an instruction block does not match the actual pushing-up height.

Embodiments for Carrying Out the Invention

[0018] Hereinafter, embodiments for carrying out the present invention will be described in detail with reference to the accompanying drawings. In the description of the drawings, the same reference numerals are assigned to the same elements, and redundant descriptions are omitted.

[0019] In each figure, an XYZ orthogonal coordinate system is shown as necessary. In the following description, the normal direction of the stage mounting surface 52a of the wafer stage 52 described later is defined as the Z-axis direction, and the surface directions along the stage mounting surface 52a are defined as the X-axis direction and the Y-axis direction. Also, hereinafter, the positive direction of the Z-axis direction may be referred to as "up", and the negative direction of the Z-axis direction may be referred to as "down".

[0020] The bonding apparatus 1 shown in Figure 1 mounts a semiconductor chip 91, which is an example of a chip component, onto a semiconductor substrate 92. The semiconductor chip 91 and the semiconductor substrate 92 together are referred to as a semiconductor device 93. Multiple semiconductor chips 91 are attached to a dicing sheet 94 via a viscoelastic film 95 (see Figure 2). The thickness of the semiconductor chip 91 is, for example, about 20 μm. The thickness of the dicing sheet 94 is, for example, about 100 μm. The bonding apparatus 1 peels the semiconductor chips 91 attached to the dicing sheet 94 together with the viscoelastic film 95. Then, the bonding apparatus 1 fixes the peeled semiconductor chips 91 to the semiconductor substrate 92. The operation of fixing the semiconductor chips 91 to the semiconductor substrate 92 is called die bonding.

[0021] The bonding apparatus 1 comprises a controller unit 2, a pickup unit 3 (pickup device), and a bonding unit 4.

[0022] The controller unit 2 controls the operation of the bonding unit 4 and the pickup unit 3. The controller unit 2 outputs control signals to the bonding unit 4 and the pickup unit 3. The controller unit 2 is a computer that includes, for example, a CPU (processor 21 shown in Figure 10), a storage unit such as ROM and RAM (memory 22 shown in Figure 10), an input / output unit (input / output interface 23 shown in Figure 10), and a driver. The controller unit 2 operates the input / output unit according to the control of the CPU. Furthermore, the controller unit 2 reads and writes data to the storage unit. These operations generate control signals that are provided to the bonding unit 4 and the pickup unit 3. Based on these control signals, the bonding unit 4 and the pickup unit 3 perform their respective operations, such as bonding and pickup operations.

[0023] The pickup unit 3 picks up the semiconductor chip 91. Picking up the semiconductor chip 91 includes peeling the semiconductor chip 91 from the dicing sheet 94 and holding the peeled semiconductor chip 91 in the pickup unit 3. The semiconductor chip 91 is attracted to the pickup unit 3. The pickup unit 3 places the picked-up semiconductor chip 91 on the intermediate stage 96. The pickup unit 3 may also be configured to transfer the semiconductor chip 91 to the bonding unit 4 by a so-called flip-chip operation. The pickup unit 3 will be described in detail later. Bonding may also be performed by the pickup unit without the intermediate stage 96.

[0024] The bonding unit 4 picks up the semiconductor chip 91 placed on the intermediate stage 96 and bonds the picked-up semiconductor chip 91 to the semiconductor substrate 92. The bonding unit 4 includes, for example, a bonding stage 41 and a bonding head 42.

[0025] The bonding stage 41 is a stage on which the semiconductor substrate 92 is placed. The bonding stage 41 is for mounting the semiconductor chip 91 onto the semiconductor substrate 92. The bonding stage 41 may also have a function for adsorbing the semiconductor substrate 92.

[0026] The bonding head 42 includes a bonding head body 421 and a bonding tool 422. The bonding head body 421 is connected to a guide rail. The bonding head body 421 reciprocates along the guide rail. This guide rail extends, for example, from the pickup unit 3 to the area where bonding is performed. The bonding head body 421 includes a motor for driving the bonding tool 422. The bonding tool 422 detachably holds a semiconductor chip 91 on an intermediate stage 96. For example, the bonding tool 422 has a vacuum suction mechanism for the detachment function. The bonding tool 422 bonds the semiconductor chip 91 to a semiconductor substrate 92. For example, the bonding tool 422 may have a heater or the like to provide heat to the semiconductor chip 91.

[0027] <Pickup Unit> Figure 2 shows the pickup unit 3. The pickup unit 3 includes a push-up module 5 (second module) and a pickup head module 6 (first module).

[0028] <Push-up module> The push-up module 5 assists in the pickup operation. The push-up module 5 includes, for example, a wafer holder 51, a wafer stage 52, and a push-up mechanism 53. The wafer holder 51 holds the dicing sheet 94. With the wafer holder 51 holding the dicing sheet 94, tension acts on the dicing sheet 94 from the center outwards. This tension stretches the dicing sheet 94.

[0029] <Wafer Stage> The wafer stage 52 adjusts the relative position between the wafer holder 51 and the push-up mechanism 53. The wafer stage 52 may, for example, translate the wafer holder 51 in the X-axis direction. The wafer stage 52 may also rotate the wafer holder 51 around the Z-axis. This movement causes the wafer stage 52 to move the semiconductor chip 91 to be picked up onto the push-up mechanism 53.

[0030] More specifically, the wafer stage 52 shown in Figure 2 has, for example, a cylindrical shape extending in the Z-axis direction. The upper end of the wafer stage 52 in the Z-axis direction is the stage mounting surface 52a on which the dicing sheet 94 is placed. The normal direction of the stage mounting surface 52a is along the Z-axis direction. A stage opening 52h communicating with the interior of the wafer stage 52 is formed in the center of the stage mounting surface 52a. The stage opening 52h has a rectangular shape that is slightly larger than the semiconductor chip 91. Inside the stage opening 52h, a block set 530 is arranged to protrude upward from the stage mounting surface 52a and push up the dicing sheet 94.

[0031] Furthermore, the stage mounting surface 52a is provided with a stage adsorption groove 52g for adsorbing and holding the dicing sheet 94. The stage adsorption groove 52g is provided, for example, so as to surround the stage opening 52h. Adsorption holes connected to a vacuum pump are in communication with the stage adsorption groove 52g. When air is sucked out of the stage adsorption groove 52g through the adsorption holes, the dicing sheet 94 around the semiconductor chip 91 is adsorbed onto the stage adsorption groove 52g.

[0032] <Push mechanism> The push-up mechanism 53 includes a block set 530. When the block set 530 is in the reference position (reference height), it does not contact the dicing sheet 94. When the block set 530 is in the push-up position, it contacts the dicing sheet 94. In the push-up position, the block set 530 protrudes above the stage mounting surface 52a. As a result, the block set 530 presses against the dicing sheet 94, causing the semiconductor chip 91 to be pushed upward (in the positive direction of the Z-axis).

[0033] The block set 530 includes a plurality of blocks 531 (upper blocks). Each of the plurality of blocks 531 has, for example, a rectangular cylindrical shape extending along the Z-axis. Each of the plurality of blocks 531 has a block up-pull surface 531a.

[0034] Each of the multiple blocks 531 is capable of independently reciprocating along the Z-axis. The multiple blocks 531 are arranged concentrically with respect to a central axis along the Z-axis. The central axis may be, for example, an axis passing through the center of the rectangular stage opening 52h when viewed in the Z-axis direction. A gap 52p is formed between the inner surface of the stage opening 52h and the outer surface of the outermost block 531.

[0035] The block set 530 is housed in the block housing 530H. The block set 530 and the block housing 530H constitute a variety component 53S. The variety component 53S is used to pick up a specific type of semiconductor chip 91. For example, if the type of semiconductor chip 91 to be picked up changes, the variety component 53S is also replaced accordingly.

[0036] The push-up mechanism 53 has a block drive mechanism 53U that independently moves each block set 530 back and forth in the Z-axis direction. The block drive mechanism 53U further includes a block motor 532 (drive unit) that provides driving force to the block set 530. The block motor 532 receives a block motor signal C532 provided by the controller unit 2 and moves the block 531 along the Z-axis direction. As described above, the variety parts 53S are replaced depending on the type of semiconductor chip 91 to be picked up, so the variety parts 53S are detachable from the block motor 532.

[0037] <Pickup head module> The pickup head module 6 includes, for example, a collet 61, a pickup head body 62, a collet shaft 63, a collet motor 64, a contact detection unit 65, and a head drive unit 66.

[0038] The collet 61 is positioned at the tip of the pickup head module 6. The collet 61 reciprocates in the Z-axis direction relative to the pickup head module 6. The collet 61 includes a collet holding surface 61a (contact surface) that detachably holds the semiconductor chip 91.

[0039] The collet 61 detachably holds the semiconductor chip 91 that has been pushed up by the push-up mechanism 53. In other words, the collet 61 picks up the semiconductor chip 91. The collet 61 holds the semiconductor chip 91, for example, by vacuum suction. The collet 61 places the picked-up semiconductor chip 91 onto the intermediate stage 96.

[0040] The pickup head body 62 includes a lower head body surface 62b, an upper head body surface 62t, a lower head body groove 62c, and an upper head body groove 62d. The pickup head body 62 is provided with a collet motor 64 (load generating section), a collet shaft 63, and a contact detection section 65.

[0041] An example of a collet motor 64 is a voice coil motor (VCM). The collet motor 64 is connected to the collet 61 via the collet shaft 63. In response to the collet motor signal C64 received from the controller unit 2, the collet motor 64 applies a load along the Z-axis direction to the collet 61 via the collet shaft 63.

[0042] The contact detection unit 65 detects when an object comes into contact with the collet holding surface 61a. The contact detection unit 65 includes a piezoelectric element, such as a piezo element, as a component. The contact detection unit 65 outputs a voltage corresponding to the force as load sensor data D651. The load sensor data D651 is provided to the controller unit 2.

[0043] The head drive unit 66 reciprocates a module, which integrates a collet 61, a pickup head body 62, a collet shaft 63, a collet motor 64, and a contact detection unit 65, in the Z-axis direction. An example of the head drive unit 66 is a ball screw mechanism. The head drive unit 66 raises or lowers the collet 61 and the pickup head body 62, etc., as a single unit, in response to a head drive signal C66 provided by the controller unit 2.

[0044] Below are three specific examples of the contact detection unit 65.

[0045] <Example 1: Contact detection unit 65A> Figure 3(a) shows a contact detection unit 65A, which is a first example. The contact detection unit 65A is located below the head housing space 62s. More specifically, the contact detection unit 65A includes two load sensors 651 and a crossbeam section 652. The crossbeam section 652 is the head bottom 62f, which defines the bottom of the head housing space 62s. The upper surface of the head bottom 62f is the bottom upper surface 62f1, which faces the head housing space 62s. The lower surface of the head bottom 62f is the bottom lower surface 62f2, which faces the collet 61 side. The thickness from the bottom upper surface 62f1 to the bottom lower surface 62f2 is relatively thin, so that the effect of the load received when an object contacts the collet holding surface 61a becomes apparent through changes in internal load and shape. In the example shown in Figure 3(a), a bottom through-hole 62g is provided in the bottom surface 62f2 through which the collet shaft 63 passes. The axis of the bottom through-hole 62g coincides with the axis A63 of the collet shaft 63. Furthermore, a pair of load sensors 651 are attached to the bottom surface 62f2. The load sensors 651 are mounted at a predetermined distance from the axis 622A of the collet shaft 63.

[0046] From a mechanical standpoint, the crossbeam section 652 has a support end that is integrated with the pickup head body 62 in the region close to the outer surface 62h of the pickup head body 62, and a free end that can be deformed vertically around the bottom through-hole 62g through which the collet shaft 63 is inserted. In other words, when the crossbeam section 652 is viewed in cross-section, it can be said to be a so-called cantilever beam from a mechanical standpoint. In this case, the applied load F64 (see Figure 3(b)) that the motor body 641 applies to the pickup head body 62 can be considered to have the axis A63 of the collet shaft 63 as its line of action. That is, the line of action of the applied load F64 that the motor body 641 applies to the pickup head body 62 coincides with the line of action of the load that the collet shaft 63 receives due to contact with an object.

[0047] Figure 3(b) visualizes the forces acting on the motor body 641. The area around the bottom through-hole 62g, which is the free end through which the collet shaft 63 passes, receives an applied load F64 from the motor body 641. The motor body 641 is subjected to the applied load F64 and a reaction force F62 to the applied load F64. These applied loads F64 and F62 have the same magnitude but opposite directions.

[0048] Here, the applied load F64 is defined as the resultant force of the generated load F64a generated by the collet motor 64 and the contact load F63a acting on the collet shaft 63. For example, as shown in Figure 3(b), when the collet 61 is not in contact with an object, the contact load F63a acting on the collet shaft 63 is zero. Therefore, the applied load F64 is equal to the generated load F64a generated by the collet motor 64. On the other hand, as shown in Figure 3(c), when the collet 61 is in contact with an object, the contact load F63a acting on the collet shaft 63 has a predetermined value. Therefore, the applied load F64 is the resultant force of the generated load F64a generated by the collet motor 64 and the contact load F63a acting on the collet shaft 63. More specifically, since the direction of the generated load F64a and the direction of the contact load F63a are opposite to each other, the applied load F64 is the generated load F64a minus the contact load F63a.

[0049] The applied load F64 is detected by a load sensor 651 located at a predetermined distance from the axis A63 of the collet shaft 63. The line of action of the applied load F64 may be considered to coincide with the axis A63 of the collet shaft 63. The position where the applied load F64 is detected is shifted in a direction perpendicular to the axis A63 of the collet shaft 63. In other words, the position where the applied load F64 is detected does not lie on the line of action of the applied load F64. For example, if the load sensor 651 can detect a state value corresponding to the bending moment, the bending moment increases with distance, so a more accurate change in load due to contact can be detected.

[0050] The operation of detecting contact will be explained in detail using the first example. As shown in Figure 3(b), the collet motor 64 presses the motor body 641 against the crossbeam section 652. At this time, we assume that no object is in contact with the collet 61. The motor body 641 is then pressed against the bottom upper surface 62f1 with an applied load F64. As mentioned above, since no object is in contact with the collet 61 at this point, the applied load F64 is equal to the generated load F64a generated by the collet motor 64. Consequently, the crossbeam section 652 generates a reaction force F62 that is the same magnitude as the applied load F64 but in the opposite direction. The applied load F64 and the reaction force F62 balance each other, and the motor body 641 comes to rest. The reaction force F62 generated in the crossbeam section 652 is then detected as an internal load by the load sensor 651.

[0051] Then, as shown in Figure 3(c), when an object comes into contact with the collet 61, the collet shaft 63 receives an upward contact load F63a. At this time, the collet motor 64 continues to generate the generated load F64a, but receives an upward contact load F63a from the collet shaft 63. Consequently, the applied load F64 received by the crossbeam section 652 is the resultant of the generated load F64a and the contact load F63a. Since the generated load F64a and the contact load F63a are in opposite directions, the applied load F64 received by the crossbeam section 652 appears to have decreased from the generated load F64a. Consequently, the reaction force F62 generated by the crossbeam section 652 also decreases. The decrease in the reaction force F62 can be detected by the load sensor 651. Therefore, the change in the value of the load sensor 651 can be used to detect when an object comes into contact with the collet 61.

[0052] In other words, according to the first example, contact of an object with the collet 61 is detected based on a change in the internal load (reaction force F62) generated in the pickup head body 62. Other contact detection methods may involve detecting the movement of a component caused by the contact of an object. However, the movement of a component begins when the load caused by the contact becomes greater than the initial load. In other words, from the time an object makes contact until the load caused by the contact becomes greater than the initial load, the contact is not detected even though the object is in contact.

[0053] On the other hand, changes in internal load do not involve the movement of such parts. When the state of the internal load, which is caused by the initial load, changes due to another load resulting from contact with an object, it can be detected immediately. Therefore, contact detection based on internal load can capture the timing of contact with greater accuracy than contact detection based on the movement of parts.

[0054] <Second example: Contact detection unit 65B> Figure 4(a) shows a contact detection unit 65B, which is a second example. In the second example, the point where a load is applied to the pickup head body 62 is located in a different place from the motor body 641. In other words, the second example differs from the first example in that the motor body 641 is not in contact with the pickup head body 62.

[0055] As shown in Figure 4(a), the contact detection unit 65B includes a load sensor 651, a beam cross section 653, and a pressing section 654. The pressing section 654 is provided on the collet shaft 63. The pressing section 654 is a member having a width greater than the outer shape of the collet shaft 63. The pressing section 654 may be a rod-shaped member extending radially from the collet shaft 63, or it may be a disc-shaped member in plan view. The pressing section 654 is fixed to the collet shaft 63. The beam cross section 653 exhibits a so-called cantilever beam, similar to the cross beam cross section 652 in the first example. The beam cross section 653 has a beam upper surface 653a and a beam lower surface 653b. Neither the beam upper surface 653a nor the beam lower surface 653b is in contact with any other part that constitutes the pickup head body 62. In other words, the beam cross section 653 can be deformed vertically along the axis A63 of the collet shaft 63. The beam cross section 653 is provided with a through hole 653h through which the collet shaft 63 is inserted.

[0056] As shown in Figure 4(b), with such a pressing section 654, the generated load F64a from the collet motor 64 is transmitted via the collet shaft 63 from the pressing section 654 to the beam cross section 653 as an applied load F64. As a result, the area around the through hole 653h in the beam cross section 653 is subjected to an applied load F64 that pushes downward, causing a change in the internal load in the beam cross section 653. This change in internal load can be detected by two load sensors 651. The load sensors 651 are generally the same as those in the first example, so a detailed explanation is omitted.

[0057] The operation for detecting contact using the second example is, in principle, the same as in the first example. That is, the generated load F64a produced by the collet motor 64 is applied as a load F64 from the pressing part 654 to the pickup head body 62. When an upward contact load F63 caused by contact with an object acts on the collet shaft 63, the applied load F64 received by the beam cross section 653 appears to decrease. This phenomenon can be detected by the load sensor 651. Therefore, contact with the collet 61 can be detected by the change in the value of the load sensor 651.

[0058] <Third example: Contact detection unit 65C> Figure 5(a) shows a third example, the contact detection unit 65C. The pickup head body 62 is provided with a groove 62e on the top surface of the head body. The groove 62e has an opening formed on the top surface 62t of the head body, and a load sensor 651 is inserted into the groove 62e. In other words, the contact detection unit 65C includes a longitudinal beam cross section 655 and a load sensor 651.

[0059] As shown in Figure 5(b), when detecting contact with the collet 61 according to the third example, first, the motor body 641, which is integrated with the collet shaft 63, is placed at the upper end of the head housing space 62s. At this time, the upper surface 641a of the motor body 641 must be in contact with the head ceiling portion 62j. Note that it is sufficient for them to be in contact, so the motor body 641 may be pressed against the head ceiling portion 62j with a predetermined applied load F64. Alternatively, the motor body 641 may not be pressed against the head ceiling portion 62j with a predetermined applied load F64, and the motor body 641 may not be receiving a reaction force F62 from the head ceiling portion 62j. When an object comes into contact with the collet holding surface 61a, the contact load F63 received from the object is transmitted to the motor body 641 through the collet shaft 63.

[0060] The applied load F64 causes a slight deformation near the groove 62e on the top surface of the head body. Specifically, the longitudinal beam section 655 extending in the direction of the axis A63 of the collet shaft 63 deforms, narrowing the groove width of the groove 62e on the top surface of the head body. The load sensor 651 detects this internal load F655 that narrows the groove width.

[0061] Then, the motor body 641 presses against the head ceiling portion 62j. At this time, the applied load F64 on the head ceiling portion 62j is the resultant force of the generated load F64a generated by the motor body 641 and the contact load F63 transmitted to the head ceiling portion 62j. When the motor body 641 is merely in contact with the head ceiling portion 62j, the generated load F64a is zero. Therefore, when the motor body 641 is merely in contact with the head ceiling portion 62j, the applied load F64 on the head ceiling portion 62j is equal to the contact load F63. In contrast, when the motor body 641 is pressing against the head ceiling portion 62j, the applied load F64 on the head ceiling portion 62j is the resultant force of the generated load F64a and the contact load F63a. In the third example, the direction of the generated load F64a and the direction of the contact load F63a coincide. Therefore, the applied load F64 is greater than the generated load F64a. In other words, when the motor body 641 is pressing against the head ceiling portion 62j, the value output by the load sensor 651 increases from the initial applied load F64 by the contact load F63 due to the contact of the object.

[0062] <Example of operation of pickup unit 3: inverted multi-stage method> Next, we will briefly explain the inverse multi-stage method, which is an example of the pickup operation performed by the bonding apparatus 1. Figure 6(a) shows the initial state of the pickup unit 3. In the initial state, the wafer stage 52 places the semiconductor chip 91 to be picked up on the block set 530. At this time, the block push-up surface 531a is below the stage mounting surface 52a. In other words, there is a gap 52q between the block push-up surface 531a and the dicing sheet 94. Furthermore, the dicing sheet 94 is held in place by the wafer stage 52. Therefore, the dicing sheet 94 does not move relative to the wafer stage 52.

[0063] Next, the block set 530 is pushed up (see Figure 6(b)). At this time, the block pushing surface 531a is above the stage mounting surface 52a. In other words, the block set 530 comes into contact with the dicing sheet 94. Also, the dicing sheet 94 on the wafer stage 52 is held in place by suction and does not move even when pushed up by the block pushing surface 531a. The semiconductor chip 91 is then held in place by the collet holding surface 61a. The dicing sheet 94 in the gap 52p between the wafer stage 52 and the block set 530 is then further held in place (see Figure 6(c)). At this point, the semiconductor chip 91 has not yet been detached from the dicing sheet 94.

[0064] Next, the blocks are moved downwards in order, starting with the outermost block 531 (see Figure 7(a)). At this time, each of the multiple blocks 531 performs an action to attract the dicing sheet 94. As each of the multiple blocks 531 that have attracted the dicing sheet 94 moves downwards in sequence, the dicing sheet 94 is peeled off the semiconductor chip 91 from the outside in. When all the blocks 531 have been moved downwards, the semiconductor chip 91 is peeled off from the dicing sheet 94 (see Figure 7(b)).

[0065] <Problems with pickup unit 3> For example, in the case of an upward thrusting motion of block set 530, Figure 6(b) illustrates that the thrusting height of each block set 530 is the same. For instance, if the target thrusting height of block set 530 is "10", the controller unit 2 instructs a thrusting height of "10". The result is that each block set 530 moves by "10".

[0066] However, this is an ideal state. The target of the thrusting motion and the result of the thrusting motion can differ due to various factors. When each of the block set 530 is in its initial state, the heights of the block thrusting surfaces 531a may not match. Also, if the driving force of the block motor 532 does not match the instruction of the controller unit 2, the target of the thrusting motion and the result of the thrusting motion may differ. As these examples show, there are mechanical and electrical error factors between the target of the thrusting motion and the result of the thrusting motion. The discrepancy between the target of the thrusting motion and the result of the thrusting motion can cause defects in the operation of picking up the semiconductor chip 91.

[0067] The calibration method for the pickup unit 3 in this embodiment involves determining the deviation (block motor correction value) between the instruction from the controller unit 2 to move by the target upward height and the result of the upward movement of the block set 530 before performing the pickup operation of the semiconductor chip 91. Then, based on the deviation, the voltage and / or voltage supplied to the block motor 532, which is a signal generated based on the instruction from the controller unit 2, is corrected. This makes it possible to correct the voltage and current supplied to the block motor 532, which are generated from the block motor signal C532 generated by the controller unit 2 based on the target value of the upward movement. As a result, the result of the upward movement can be made to match the target of the upward movement. The calibration method for the pickup unit 3 and the manufacturing method of the semiconductor device 93 will be described in detail below.

[0068] <Manufacturing method for semiconductor device 93> Figure 8 is a flowchart of the manufacturing method for the semiconductor device 93 using the calibration method for the pickup unit 3. The manufacturing method for the semiconductor device 93 involves manufacturing a semiconductor device 93 in which a semiconductor chip 91 is mounted on a semiconductor substrate 92.

[0069] First, the type of component 53S is attached (S1). First, the type of semiconductor chip 91 to be mounted is identified. Next, the type of component 53S corresponding to the type of semiconductor chip 91 is identified. Then, the identified type of component 53S is attached to the block motor 532.

[0070] Next, the block motor correction value R532 is obtained (S2). This block motor correction value R532 is used to match the result of the thrusting motion to the target of the thrusting motion. The block motor correction value R532 may also be used to correct the voltage or current supplied to the block motor 532. In this case, the block motor correction value R532 is a corrected voltage value or a corrected current value. Obtaining the block motor correction value R532 (S2) is the calibration method for the pickup unit 3 in this embodiment. Therefore, the details of this step S2 will be described later.

[0071] Next, the corrected block motor signal B532 (corrected drive signal) is obtained (S3). The controller unit 2 corrects the block motor signal C532 using the block motor correction value R532. As a result, the corrected block motor signal B532 can be obtained.

[0072] Next, bonding is performed (S4). The controller unit 2 bonds the semiconductor chip 91 to the semiconductor substrate 92 by providing control signals to the pickup unit 3 and the bonding unit 4, respectively.

[0073] Then, it is determined whether the bonding is complete or not (S5). The controller unit 2 determines whether the bonding of all semiconductor chips 91 identified in step S1 is complete or not. If the bonding of all semiconductor chips 91 is not complete (S5: NO), the controller unit 2 performs bonding again (S4). If the bonding of all semiconductor chips 91 is complete (S5: YES), the controller unit 2 terminates the bonding. Then, it identifies the type of the next semiconductor chip 91 to be bonded (S1).

[0074] By performing the above steps S1 to S5, the semiconductor device 93 can be obtained. As described above, each time the type of component 53S is replaced due to a change in the type of semiconductor chip 91 to be bonded, the block motor correction value R532 is obtained (S2) and the corrected block motor signal B532 is obtained (S3). Therefore, even if the type of semiconductor chip 91 to be bonded changes, good pickup operation can be performed.

[0075] <Calibration method for pickup unit 3> Next, the calibration method for the pickup unit 3 will be described. The calibration method for the pickup unit 3 corresponds to step S2 above, which is used to obtain the block motor correction value R532. In this embodiment, calibration may be defined as obtaining the block motor correction value R532. Alternatively, calibration may be defined as including obtaining the block motor correction value R532 and obtaining a corrected block motor signal B532 using the block motor correction value R532.

[0076] Figure 9 is a flowchart of the calibration method for the pickup unit 3. Figure 10 is a functional block diagram of the controller unit 2. The calibration method for the pickup unit 3 is performed by the controller unit 2.

[0077] The controller unit 2 has a processor 21, a memory 22, and an input / output interface 23 as its physical components.

[0078] The processor 21 executes the calibration program for the pickup unit 3 stored in memory 22. As a result, the processor 21 functions as several mechanical components for the calibration method of the pickup unit 3. The processor 21 functions as a load data determination unit 211, a correction value calculation unit 212, a block motor signal unit 213, a collet motor signal unit 214, and a head drive signal unit 215.

[0079] Memory 22 stores the calibration program described above. Memory 22 also stores some data for executing the calibration method of the pickup unit 3. For example, memory 22 stores the load sensor data D651 passed from the input / output interface 23. Memory 22 passes data and signals requested by the processor 21 to the processor 21. For example, memory 22 passes the load sensor data D651, the block motor signal C532, and the target push-up height TH to the processor 21. Furthermore, memory 22 stores data and signals passed from the processor 21. For example, memory 22 stores the block motor correction value R532.

[0080] The input / output interface 23 accepts data and signals from the outside. The input / output interface 23 passes the data and signals received from the outside to the memory 22. For example, the input / output interface 23 accepts load sensor data D651 and passes it to the memory 22. The input / output interface 23 outputs signals received from the memory 22 and the processor 21 to the outside. For example, the input / output interface 23 passes the collet motor signal C64 to the collet motor 64. The input / output interface 23 passes the block motor signal C532 or the correction block motor signal B532 to the block motor 532.

[0081] The functions of the controller unit 2 are not limited to those exemplified. For example, the controller unit 2 has the function of controlling the position of the pickup unit 3, the function of causing the pickup unit 3 to hold the semiconductor chip 91, and the function of releasing the semiconductor chip 91 from the pickup unit 3. The controller unit 2 also has the function of controlling the position of the bonding unit 4 and the function of controlling the bonding operation by the bonding unit 4.

[0082] The calibration method for the pickup unit 3 will be described in detail below. In the following description, we will assume the pickup unit 3 equipped with the contact detection unit 65A, which is the first example shown in Figure 3.

[0083] First, the acquisition of load sensor data D651 is initiated (S20). The controller unit 2 receives the load sensor data D651 output by the load sensor 651. Specifically, the input / output interface 23 receives the load sensor data D651. The input / output interface 23 then passes the received load sensor data D651 to the memory 22.

[0084] Next, the collet motor 64 generates a first applied load F64a (see S21, Figure 11(a)). The controller unit 2 presses the motor body 641 against the crossbeam section 652 by supplying a collet motor signal C64 to the collet motor 64. As a result, the first applied load F64a is generated (see Figure 3(b)). This first applied load F64a is detected by the load sensor 651.

[0085] Next, the head drive unit 66 positions the collet 61 in a predetermined position (S22). The controller unit 2 positions the collet 61 directly above the block set 530 and the wafer stage 52. Specifically, the controller unit 2 provides the head drive unit 66 with a head drive signal C66 to position the collet 61 such that the collet holding surface 61a and the stage mounting surface 52a are within the movable range along the Z-axis direction of the collet 61. With this arrangement, when the collet 61 is moved downward in the Z-axis direction, the collet holding surface 61a can abut against the stage mounting surface 52a. At this time, the collet holding surface 61a is separated from the stage mounting surface 52a.

[0086] Next, the head drive unit 66 generates a predetermined initial load (second applied load F642, see Figure 11(b)) (S23). The controller unit 2 provides a head drive signal C66 to the head drive unit 66. Specifically, the head drive signal unit 215 provides the head drive signal C66 to the head drive unit 66 via the input / output interface 23. As a result, the head drive unit 66 moves the collet 61 and the pickup head body 62 downward in the Z-axis direction. Initially, the collet 61 is not in contact with the wafer stage 52, so the load sensor data D651 shows the first applied load F64a (see Figure 11(a)). Next, the collet 61 comes into contact with the wafer stage 52. Furthermore, the head drive unit 66 presses the collet 61 toward the wafer stage 52. As a result, the collet 61 receives a predetermined stage contact load F2 (e.g., 10N) from the wafer stage 52 (see Figure 11(b)). Therefore, the load sensor 651 detects the value obtained by subtracting the stage contact load F2 from the first applied load F64a as the second applied load F641. In this way, it is possible to detect that the collet holding surface 61a has come into contact with the stage mounting surface 52a, provided that the first applied load F64a has been reduced by the stage contact load F2.

[0087] At this time, each of the block sets 530 is separated downward along the Z-axis direction from the stage mounting surface 52a. Therefore, a predetermined gap is created between the collet 61 and the block 531.

[0088] Next, the block motor signal C532 is output (S24). The controller unit 2 provides the block motor signal C532 to the block motor 532. The block motor signal C532 is the voltage or current supplied to the block motor 532. In other words, the block motor signal unit 213 receives data indicating the target upward thrust height of the block 531 and obtains the voltage or current required to operate the block motor 532 by that target upward thrust height based on that data. The block motor signal unit 213 provides the block motor signal C532 to the block motor 532 via the input / output interface 23 to thrust the block 531 upward along the Z-axis. As a result, the block 531 gradually begins to move upward along the Z-axis.

[0089] Here, the position of block 531 immediately before step S24 begins is defined as the reference position P531. Similarly, the position of the block thrust surface 531a immediately before step S24 begins is defined as the reference height H531. As shown in Figure 11(b), etc., the reference height H531 is at least below the stage mounting surface 52a. For example, the reference height can also be defined as the distance from the stage mounting surface 52a to the block thrust surface 531a.

[0090] Furthermore, regarding the block motor signal unit 213, in addition to data indicating the target thrust height, the block motor signal unit 213 may receive a block motor correction value R532, correct the data indicating the target thrust height using the block motor correction value R532, and obtain a voltage or current to operate the block motor 532 by the corrected value. The function of converting the data indicating the target thrust height and the corrected data into voltage or current may be a function of the block motor signal unit 213 as described above. Alternatively, the function of converting the data indicating the target thrust height and the corrected data into voltage or current may be a function of the input / output interface 23.

[0091] At this time, controller unit 2 continues to output the collet motor signal C64. Similarly, controller unit 2 continues to acquire load sensor data D651.

[0092] Next, it is determined whether or not a change has occurred in the load sensor data D651 (S25). The controller unit 2 determines whether or not a change has occurred in the load sensor data D651. Specifically, the load data determination unit 211 receives the load sensor data D651 via the memory 22. The load data determination unit 211 determines whether or not a significant change has occurred in the load sensor data D651, which indicated the second applied load F641. A significant change may be defined as a change in the load sensor data D651 that exceeds the load threshold LT for the second applied load F641 (see Figure 12(b)).

[0093] When block 531 is not in contact with collet 61 (see Figure 12(a)), collet 61 is pressing the wafer stage 52 with a second applied load F641, so the load sensor data D651 indicates the second applied load F641. In other words, as long as the load sensor data D651 indicates the second applied load F641, it can be seen that block 531 is not in contact with collet 61.

[0094] As block 531 continues its upward thrusting motion, block 531 comes into contact with collet 61. In this paragraph, "come into contact" refers to a state where block 531 is merely touching collet 61. Therefore, when block 531 is in contact with collet 61, no load acts on block 531 that would cause it to thrust upward. Even when block 531 is in contact with collet 61, the load sensor data D651 shows the second applied load F641.

[0095] Furthermore, block 531 continues its upward thrusting motion. If the upward thrusting motion continues from a state where block 531 is already in contact with collet 61, block 531 will not move upward in the Z-axis direction. On the other hand, block 531 presses collet 61 with a predetermined block load F3 (see Figure 12(b)). In other words, block 531 applies a block load F3 to collet 61 that pushes it upward. From the perspective of collet 61, while it was pressing the stage mounting surface 52a with a second applied load F641, it receives a block load F3 from block 531 that is in the opposite direction to the second applied load F641. As a result, the second applied load F641 that collet 61 presses against the wafer stage 52 becomes a third applied load F642, which is the second applied load F641 minus the magnitude of the block load F3 received from block 531. Therefore, as the block load F532, which causes block 531 to push up collet 61, increases, the load value indicated by the load sensor data D651 gradually decreases. When the load sensor data D651 falls below the load threshold LT, it can be determined that block 531 is in contact with collet 61.

[0096] When it is determined that no change has occurred in the load sensor data D651 (S25: NO), the controller unit 2 continues to output the block motor signal C532 (S24). Specifically, the block motor signal unit 213 continues to output the block motor signal C532 so that the upward height of the block 531 increases over time. Then, it determines again whether or not a change has occurred in the load sensor data D651 (S25). In other words, the controller unit 2 repeats the process of outputting the block motor signal C532 and determining whether or not a change has occurred in the load sensor data D651 until it is determined that a change has occurred in the load sensor data D651.

[0097] When the controller unit 2 determines that a change has occurred in the load sensor data D651 (S25: YES), it stops outputting the block motor signal C532 (S26). As a result, the block motor 532 stops pushing the block 531 upward in the Z-axis direction. Then, the block motor signal section 213 of the controller unit 2 provides the block motor 532 with a block motor signal C532 to return the block 531 to its initial position. As a result, the block 531 moves downward along the Z-axis direction and returns to its initial position.

[0098] Furthermore, the controller unit 2 stores the block motor signal C532, which was output when it was determined that a change had occurred in the load sensor data D651, in the memory 22.

[0099] Next, the difference EH between the indicative upward height CH indicated by the block motor signal C532 and the target upward height TH is obtained (S27). The difference EH between the indicative upward height CH indicated by the block motor signal C532 and the target upward height TH is the block motor correction value R532.

[0100] Here, the target push-up height TH is the distance along the Z-axis from the block push-up surface 531a to the stage mounting surface 52a when the block set 530 is in its initial position. In other words, the target push-up height TH is the height from the block push-up surface 531a to the collet holding surface 61a before the push-up operation of the block 531 begins.

[0101] Specifically, the correction value calculation unit 212 receives the block motor signal C532, which was output when it was determined that a change had occurred in the load sensor data D651, and the target thrust height TH from the memory 22. The correction value calculation unit 212 obtains the instructed thrust height CH from the received block motor signal C532. As mentioned above, the timing at which it is determined that a change has occurred in the load sensor data D651 does not strictly coincide with the timing at which the block 531 contacts the collet 61. The timing at which it is determined that a change has occurred in the load sensor data D651 is delayed from the timing at which the block 531 contacts the collet 61. This delay is due to the load threshold LT. Therefore, in the calculation to obtain the instructed thrust height CH from the received block motor signal C532, the instructed thrust height CH may be corrected using the load threshold LT or the speed of the thrusting motion. Then, the correction value calculation unit 212 subtracts the instructed thrust height CH from the target thrust height TH.

[0102] If the upward thrusting motion of block 531 does not contain any error elements, the indicated upward thrusting height CH (see Figure 13(a)) indicated by the block motor signal C532 will match the target upward thrusting height TH (see Figure 13(b)). In other words, the result of subtracting the indicated upward thrusting height CH from the target upward thrusting height TH is zero.

[0103] If the upward movement of block 531 includes error elements, the indicated upward height CH will not match the target upward height TH (see Figure 13(c)). For example, the actual upward height RH may reach the target upward height TH before the indicated upward height CH reaches the target upward height TH. Conversely, when the indicated upward height CH reaches the target upward height TH, the actual upward height RH may not reach the target upward height TH.

[0104] As illustrated in Figures 13(c) and 13(d), when the actual thrust height RH reaches the target thrust height TH (see point P2 in Figure 13(d)), the indicated thrust height CH shown by the block motor signal C532 output at that time has not reached the target thrust height TH (see point P1 in Figure 13(c)). The difference EH between the indicated thrust height CH shown by the block motor signal C532 and the target thrust height TH is treated as the block motor correction value R532. In the example in Figure 13(c), the difference EH is a negative value. Depending on the error factors, the difference EH may also be a positive value.

[0105] By performing the above steps S20 to S27, the block motor correction value R532 for the first block 531 can be obtained. Then, the controller unit 2 obtains the block motor correction value R532 for all blocks 531 by performing steps S20 to S27 again.

[0106] <Effects and Effects> The pickup unit 3 peels the semiconductor chip 91, which is attached to the dicing sheet 94, from the dicing sheet 94. The pickup unit 3 includes a pickup head module 6 which includes a collet 61 having a collet holding surface 61a and a contact detection unit 65 which detects when an object comes into contact with the collet holding surface 61a, and a push-up module 5 which includes a block 531 provided opposite the collet holding surface 61a and a block push-up surface 532 which moves the block push-up surface 531a of the block 531 closer to the collet holding surface 61a or away from the contact surface. The calibration method for the pickup unit 3 involves moving the collet 61 relative to the block push-up surface 531a, thereby positioning the collet 61 at a predetermined distance from the reference height H531 of the block push-up surface 531a, and the block motor 532 receiving a collet motor signal C64 generated from information on the push-up height of the block push-up surface 531a, thereby bringing the block push-up surface 531a closer to the collet holding surface 61a positioned in step S22. The process includes step S24, step S25 in which the contact detection unit 65 detects that the block push-up surface 531a has come into contact with the collet holding surface 61a, and step S3 in which a block motor correction value R532 is obtained for correcting the collet motor signal C64 generated from the push-up height information, using the height from the reference height H531 to the collet holding surface 61a set in step S22 for positioning the collet 61, and the height of the push-up surface when it is detected that the block push-up surface 531a has come into contact with the collet holding surface 61a.

[0107] In other words, the calibration method for the pickup unit 3 of this embodiment is performed to determine the level of the push-up tool for multi-stage push-up and / or inverse multi-stage push-up. A sensor such as a piezoelectric element that detects force is used as a unit to detect when the collet 61, which is the pickup tool, has come into contact with the object to be picked up. The contact detection function of the force sensor then detects when the multiple blocks 531 for pushing up the semiconductor chip 91 have risen to the same height as the wafer suction surface. By performing this for each push-up block 531, it becomes possible to calibrate the push-up height of each block 531.

[0108] According to this method, the block push-up surface 531a is brought into contact with the collet holding surface 61a of the pickup head module 6, which is in a fixed position, by bringing the block push-up surface 531a closer to the collet holding surface 61a. Then, a block motor correction value R532 for correcting the block motor signal C532 given to the block motor 532 is obtained using the height from the block push-up surface 531a to the collet holding surface 61a and the height of the block push-up surface 531a indicated by the block motor signal C532 that the block motor 532 was receiving when it was detected that the block push-up surface 531a had come into contact with the collet holding surface 61a. By using this block motor correction value R532, the difference between the target push-up height and the actual push-up height caused by various reasons can be eliminated, and the actual push-up height can be brought closer to the target push-up height. Therefore, defects that may occur during the operation of peeling the semiconductor chip 91 from the dicing sheet 94 can be reduced.

[0109] The calibration method for the pickup unit 3 includes a contact detection unit 65 which includes a load sensor 651 capable of acquiring load sensor data D651 corresponding to the contact load caused by the contact of the block 531 with the collet 61. Step S25, which detects contact, detects that the block's thrusting surface 531a has come into contact with the collet holding surface 61a, provided that a change has occurred in the load sensor data D651. According to step S25, contact between the collet holding surface 61a and the block's thrusting surface 531a can be detected by load. Detection by load can improve the accuracy of detection compared to detection by displacement. According to step S25, contact between the block thrust surface 531a and the collet holding surface 61a can be detected by the load. Compared to position sensors, force sensors can capture the moment of contact, resulting in less time delay and smaller detection errors. Compared to displacement detection, load-based detection reduces the variability of contact detection, thus improving the accuracy and stability of contact detection.

[0110] In the calibration method for the pickup unit 3, step S25, which detects contact, detects that the block push-up surface 531a has come into contact with the collet holding surface 61a, provided that the load value indicated by the load sensor data D651 decreases due to the application of a contact load.

[0111] The push-up module 5 includes a block drive mechanism 53U including a block motor 532, and a set of block parts 53S that are detachable from the block drive mechanism 53U and each constitutes a set of block push-up surfaces 531a. The calibration method for the pickup unit 3 includes a step S1 in which the first part 53S is replaced with a second part 53S before the step S21 in which the collet holding surface 61a is positioned. Each time the part 53S is replaced, a block motor correction value R532 can be obtained. Therefore, even when the part 53S is replaced, defects that may occur during the operation of peeling the semiconductor chip 91 from the dicing sheet 94 can be reduced.

[0112] Furthermore, the calibration required when replacing the type of component 53S can be performed using the basic components of the pickup unit 3 without the need for special jigs or measuring devices. As a result, bonding work for multiple different types of semiconductor chips 91 can be streamlined and automated, and productivity can be improved.

[0113] The method for manufacturing the semiconductor device 93 includes the steps of: obtaining a block motor correction value R532 for calibrating the pickup unit 3 (step S2); obtaining a corrected block motor signal B532 by correcting the collet motor signal C64 generated from information on the push-up height of the block push-up surface 531a using the block motor correction value R532 (step S3); and peeling the semiconductor chip 91 attached to the dicing sheet 94 off the dicing sheet 94 by supplying the corrected block motor signal B532 to the block motor 532 (step S4). Step S2 for obtaining a block motor correction value R532 for calibrating the pickup unit 3 involves moving the collet 61 relative to the block push-up surface 531a, thereby positioning the collet 61 at a predetermined distance from the reference height of the block push-up surface 531a, and the block motor 532 receiving the collet motor signal C64 generated from the push-up height information of the block push-up surface 531a, thereby causing the collet holding surface 61a positioned by step S22 to push up the block. The process includes: step S24 of bringing the buckling surface 531a closer; step S25 of detecting that the block buckling surface 531a has come into contact with the collet holding surface 61a using the contact detection unit 65; and step S3 of obtaining a block motor correction value R532 for correcting the collet motor signal C64 generated from the buckling height information, using the height from the reference height H531 to the collet holding surface 61a set in step S22 of positioning the collet 61, and the height of the buckling surface when it is detected that the block buckling surface 531a has come into contact with the collet holding surface 61a.

[0114] According to this manufacturing method for the semiconductor device 93, the block motor signal C532 for driving multiple blocks can be corrected. As a result, defects that may occur during the operation of peeling the semiconductor chip 91 from the dicing sheet 94 can be reduced.

[0115] The controller unit 2 performs an operation (S22) to move the collet 61 relative to the block push-up surface 531a, thereby positioning the collet 61 at a predetermined distance from the reference height of the block push-up surface 531a, and the block motor 532 receives a collet motor signal C64 generated from information on the push-up height of the block push-up surface 531a, thereby generating a collet motor signal C6 to bring the block push-up surface 531a closer to the collet holding surface 61a of the pickup head module 6 whose position is being held. The following operations are performed: outputting 4 (S24), detecting by the contact detection unit 65 that the block push-up surface 531a has come into contact with the collet holding surface 61a (S25), and obtaining a block motor correction value R532 for correcting the collet motor signal C64 generated from the push-up height information, using the height from the reference height to the collet holding surface 61a set in step S22 for positioning the collet 61, and the height of the push-up surface when it is detected that the block push-up surface 531a has come into contact with the collet holding surface 61a (S27).

[0116] The bonding apparatus 1 can correct the block motor signal C532 for driving the block set 530. As a result, defects that may occur during the operation of peeling the semiconductor chip 91 from the dicing sheet 94 are reduced, thereby increasing the yield of products containing the semiconductor chip 91.

[0117] <Variation> The present invention may be implemented in various forms, including the embodiments described above, with various modifications and improvements based on the knowledge of those skilled in the art. Furthermore, modified versions may be constructed by utilizing the technical matters described in the embodiments described above.

[0118] In this embodiment, a piezoelectric element that measures force was used as an example of a contact detection unit. For example, a proximity sensor that measures distance may be used as a contact detection unit. [Explanation of symbols]

[0119] 1...Bonding device, 2...Controller unit, 3...Pickup unit (pickup device), 4...Bonding unit, 5...Push-up module (second module), 6...Pickup head module (first module), 53...Push-up mechanism, 61a...Collet holding surface (contact surface), 91...Semiconductor chip, 92...Semiconductor substrate, 93...Semiconductor device, 94...Dicing sheet (sheet), 531...Block, 532...Block motor (drive unit), 64...Collet motor (load generating unit), 651...Load sensor.

Claims

1. A calibration method for a pickup device for picking up semiconductor chips attached to a sheet, comprising: a first module including a collet having a contact surface and a contact detection unit for detecting when an object comes into contact with the contact surface; and a second module including a push-up block provided opposite the contact surface and a drive unit for moving the push-up surface of the push-up block closer to or further away from the contact surface, The steps include moving the collet relative to the thrust surface to position the contact surface at a predetermined distance from the reference height of the thrust surface, The drive unit receives a drive signal generated from the information of the upward height of the upward-pushing surface, and the step of bringing the upward-pushing surface closer to the contact surface that was positioned by the step of positioning the contact surface, The contact detection unit detects that the pushing surface has come into contact with the contact surface, A calibration method for a pickup device, comprising the step of obtaining a correction value for correcting the drive signal generated from the information of the upward thrust height, using the height from the reference height to the contact surface set in the step of positioning the contact surface, and the height of the upward thrust surface when it is detected that the upward thrust surface has come into contact with the contact surface.

2. The contact detection unit includes a sensor capable of acquiring sensor data corresponding to the contact load generated due to the contact of the push-up block with the collet. The calibration method for a pickup device according to claim 1, wherein the step of detecting contact is to detect that the pushing surface has come into contact with the contact surface, provided that a change has occurred in the sensor data.

3. The calibration method for a pickup device according to claim 2, wherein the step of detecting contact is to detect that the pushing surface has come into contact with the contact surface, provided that the value of the load indicated by the sensor data decreases due to the application of the contact load.

4. The calibration method for a pickup device according to claim 2, wherein the step of detecting contact is to detect that the pushing surface has come into contact with the contact surface, provided that the value of the load indicated by the sensor data increases due to the application of the contact load.

5. The second module includes a drive mechanism including the drive unit, and a push-up mechanism that includes a plurality of push-up blocks, each constituting a plurality of push-up surfaces, and is detachable from the drive mechanism. The calibration method for a pickup device according to claim 1, further comprising the step of replacing the first push-up mechanism with the second push-up mechanism before the step of arranging the contact surfaces.

6. A pickup device for peeling a semiconductor chip attached to a sheet from the sheet, comprising: a first module including a collet having a contact surface and a contact detection unit for detecting when an object comes into contact with the contact surface; and a second module including a push-up block provided opposite the contact surface and a drive unit for moving the push-up surface of the push-up block closer to or away from the contact surface; and a step of obtaining a correction value for calibrating the pickup device, The steps include obtaining a corrected drive signal by correcting the drive signal generated from the information of the push-up height of the push-up surface using the correction value, The process includes the step of peeling off the semiconductor chip attached to the sheet from the sheet by supplying the correction drive signal to the drive unit, The step of obtaining the correction value for calibrating the pickup device is: The steps include moving the collet relative to the thrust surface to position the contact surface at a predetermined distance from the reference height of the thrust surface, The drive unit receives the drive signal generated from the information of the upward height of the upward-pushing surface, and the step of bringing the upward-pushing surface closer to the contact surface that was positioned by the step of positioning the contact surface, The contact detection unit detects that the pushing surface has come into contact with the contact surface, A method for manufacturing a semiconductor device, comprising the steps of obtaining a correction value for correcting the drive signal generated from the information of the push-up height, using the height from the reference height to the contact surface set in the step of positioning the contact surface, and the height of the push-up surface when it is detected that the push-up surface has come into contact with the contact surface.

7. A pickup unit for peeling a semiconductor chip attached to a sheet from the sheet, The system includes a controller unit for controlling the pickup unit, The aforementioned pickup unit is A first module including a collet having a contact surface and a contact detection unit that detects when an object comes into contact with the contact surface, The module includes a push-up block provided opposite the contact surface and a second module including a drive unit that moves the push-up surface of the push-up block closer to or further away from the contact surface, The aforementioned controller unit is The operation involves moving the collet relative to the thrust surface to position the contact surface at a predetermined distance from the reference height of the thrust surface, The drive unit receives a drive signal generated from information on the upward height of the upward-pushing surface, and outputs the drive signal to bring the upward-pushing surface closer to the contact surface of the first module whose position is being held. The contact detection unit performs the operation of detecting that the pushing surface has come into contact with the contact surface, A pickup device that performs the following actions: obtaining a correction value for correcting the drive signal generated from the information of the upward thrust height, using the height from the reference height to the contact surface set in the step of positioning the contact surface, and the height of the upward thrust surface when it is detected that the upward thrust surface has come into contact with the contact surface.

Images