JP2650943B2 - Bonding method and apparatus - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial Application Field] The present invention relates to a TAB multi-pin chip mounting method in which leads on a tape and bumps formed on an IC chip are aligned and pressure-bonded. The present invention relates to a method for manufacturing a semiconductor device for performing bonding of a TAB inner lead capable of positioning the inner lead and a bump on an IC chip with high precision and in a state where both are overlapped.

[Conventional technology]

The TAB method is, as shown in FIG. 2, an inner lead 2 formed on a tape 1 and a bump 3 formed on an IC chip 4.
Are aligned, and then both are collectively joined and crimped.

Conventionally, a wire bonding method has been widely used as an IC chip connection method. In the wire bonding method, the minimum pitch between connectable electrodes is set to about 160 μm due to the size limitation of a bonding tool that heats and compresses a wire.

On the other hand, IC chips with a large number of input / output pins, such as logic LSIs for computers and drive ICs for liquid crystal displays, require connections of 200 pins or more at a narrow pitch of 160 μm or less due to chip cost and high density mounting requirements. It is becoming impossible to cope with the method. In the TAB method, because of the collective lead connection, there is no dimensional restriction due to the above-mentioned bonding tool, and extremely narrow pitch and multi-pin connection becomes possible.

Alignment in the TAB method generally used in the past detects the alignment of the tape 1 and the IC chip 4 separately,
This has been done by moving each to a preset position. The prior art disclosed in Japanese Patent Publication No. 62-27735 and Japanese Patent Laid-Open No. 58-141 discloses an alignment mark provided at an arbitrary point on a tape 1 remote from a bonding position, as shown in FIG. Use 65 to adjust the position or store the specific shape of the inner lead pattern, detect the pattern each time a new tape is supplied, calculate the amount of deviation from the set position, and correct the tape position The way to do it was taken. Similarly, in the alignment of the IC chip 4, a pattern having a specific shape in the IC chip 4 is stored, a deviation from a position set for each IC chip 4 is obtained, and the position is corrected.

In order to position the tape 1 and the IC chip 4 in the rotational direction, a station for mechanically correcting the rotational direction of the IC chip 4 is provided between the tray on which the IC chip 4 before bonding is mounted and the bonding position. A method of correcting the shift amount has been adopted.

A conventional tape bonding apparatus is disclosed in
As described in Japanese Patent No. 105972, the impact load on the IC chip during bonding is suppressed by lowering the tool while applying a low air pressure to the tool and switching the air pressure to a high pressure at the start of bonding. At that time, the switching of the air pressure is synchronized with the bonding start position by a timing cam provided in the tool driving mechanism.

In a conventional bonding tool and stage, as described in, for example, JP-A-62-97341, leads on a tape and bumps on an IC chip are heated and pressed by a bonding tool having a heater. At that time, on the stage on which the IC chip was mounted, a heater was provided on the stage immediately below the IC chip in order to reduce the temperature at the time of thermocompression bonding.

[Problems to be solved by the invention]

In the above-described TAB method, the number of pins is increased, and the bumps 3 and the leads 2 become finer, so that more accurate alignment is required. However, the above-described prior art uses the alignment mark 65 located at a position distant from the bonding position.
Tape 1 and IC chip 4 depending on mechanical accuracy
Is corrected, and bonding is performed as it is. Therefore, the positioning accuracy includes mechanical accuracy in addition to the detection accuracy of the alignment mark and the like. For this reason, there has been a problem that sufficient alignment accuracy cannot be obtained for the miniaturized leads 2 and bumps 3 for a multi-pin TAB.

Further, the above prior art does not take into consideration that there is variation in the height of a bump as a bonding portion between individual IC chips. Therefore, in the case of an IC chip with a low bump height,
The air pressure was switched to a high pressure before the tool contacted the bump via the lead, and it was highly possible that the tool exerted an impact pressure on the lead and the bump. As a result, stress concentrates on the leads and bumps, leading to breakage of the leads, peeling of the bumps,
There is a problem that so-called bonding damage such as wallac in the lower layer portion of the bump is likely to occur.

Further, in the collective bonding, it is inevitable that a pressing force is applied to only a small number of bumps at the start of pressurization due to a variation in bump height due to bump formation accuracy. In the above prior art, by switching the air in two stages, it is intended that many bumps are brought into contact at a low pressure. Since the load resistance of each bump is reduced with respect to the required pressing force, the above-described bonding damage is also likely to occur.

Further, the above-mentioned conventional technology does not consider the variation in the temperature distribution at the connection between the corner and the center of the side, particularly on the IC chip. Furthermore, Sn or solder whose surface has been treated on the lead on the tape adheres to the bottom surface of the bonding tool, causing one-side contact of the tool. Therefore, there has been a problem that a sufficient connection state has not been obtained.

An object of the present invention is to provide a highly accurate and highly reliable TAB inner lead bonding method and apparatus corresponding to a multi-pin TAB and an alignment method therefor.

It is also an object of the present invention that the accuracy of the pump height is insufficient.
It is an object of the present invention to provide a bonding method and apparatus that achieves good bonding without applying an impact force and an excessive pressing force to an IC chip.

It is a further object of the present invention to provide a bonding tool and a stage having a uniform temperature distribution and a bonding apparatus that achieves good bonding using the same.

Still another object of the present invention is to provide an IC free from chip breakage and lead breakage.

[Means for solving the problem]

To this end, the present invention has formed bumps on the surface.
The IC chip and the inner lead formed on the carrier tape are opposed to each other at a bonding station, the position of the IC chip on the stage is detected by a camera at the bonding station, the position correction amount of the stage is calculated, and A method of manufacturing a semiconductor device in which the IC chip is aligned and bonded, wherein one or more fields of view to be detected at corners of the IC chip are set, and the position of the inner lead or the IC chip is detected within the field of view. And the position of the inner lead or the position of the IC chip in the field of view,
Equipped with one or more visual field positions and IC chip
The correction amount in the XYθ direction is calculated from the rotation center position of the XYθ stage, and the lead and the IC chip are aligned and bonded.

In addition, the position of the inner lead formed on the carrier tape is detected by a camera at the bonding station, and the bump is formed on the surface by the bonding station.
A method of manufacturing a semiconductor device in which a chip and the inner lead are opposed to each other, and the lead and the IC chip are aligned and bonded at the bonding station, wherein a field of view detected at a corner of the IC chip is one or more places. Setting, detecting the position of the inner lead or the IC chip in the visual field, and detecting the position of the inner lead or the IC chip in the visual field, one or more visual field positions, and XYθ on which the IC chip is mounted. The correction amount in the XYθ direction is calculated from the rotation center position of the stage, and the leads and the IC chip are aligned and bonded.

In the method of manufacturing a semiconductor device, after the alignment of the inner lead and the IC chip, a relative positional relationship between the inner lead and the IC chip is fixed, and the inner lead and the IC chip are bonded with a bonding tool. It is characterized by being pressed.

Further, in the method of manufacturing a semiconductor device described above, performing oblique illumination orthogonal to both surfaces of the IC chip and the carrier tape,
The position of the IC chip is detected from an image detected in this lighting state.

Further, in the method of manufacturing a semiconductor device described above, oblique illumination is performed obliquely on both surfaces of the IC chip and the carrier tape, and the position of the inner lead is obtained from an image detected in this illumination state.

Also, the IC chip with the bumps formed on the surface and the inner leads formed on the tape face each other at the bonding position, optically magnify the state where both parts overlap, and detect the enlarged image with the image sensor. The amount of misalignment is determined by processing the resulting image, the inner lead and the IC chip are aligned by finely moving the XYθ stage on which the IC chip is mounted, and after alignment, the relative positional relationship between the two in the XYθ direction is fixed. Both are pressed by a bonding tool.

Further, the IC chip having bumps formed on the surface thereof and the inner leads formed on the tape are opposed to each other at a bonding position, and the oblique illumination obliquely oblique to both surfaces makes the inner leads brighter and brighter. While determining the lead position, the inner leads and bumps were darkly detected by epi-illumination perpendicular to the both surfaces, and the IC chip position was determined using the inner lead position obtained from this pattern and the oblique illumination image. Things.

Further, the above object is achieved by providing means for detecting a pressing force acting between the tool and the IC chip, means for detecting the amount of movement of the tool, and means for changing the driving force of the tool. This is achieved by performing the bonding while changing the tool driving force based on the detection result obtained.

In addition, for the above purpose, the present invention relates to a method in which
The shape and heater arrangement are devised so that the temperature distribution on the bottom surface of the bonding tool becomes uniform when thermocompression bonding of bumps on an IC chip.

Further, the bottom surface of the bonding tool is coated with a chemically stable material having excellent thermal conductivity and abrasion resistance to perform thermocompression bonding.

Further, the stage is provided with a function of absorbing a shocking pressure from the bonding tool and a deviation of the parallelism between the chip and the bottom surface of the tool.

[Action]

With the above configuration, the tape, chip, tool, and stage can be aligned with the optical axis of the lens as the reference position, so that high-precision bonding can be performed and the number of pins in the IC can be increased. Can be.

Further, by finely moving the stage in the XYθ direction, the alignment between the lead and the chip can be performed with high accuracy.

Also, by using oblique illumination and epi-illumination, alignment can be performed with the lead and bump overlapping at the bonding position, so that the alignment accuracy is maintained when bonding, and the height of the lead and the IC chip is maintained. Accurate alignment becomes possible.

Further, the pressing force between the tool and the IC chip in the bonding can be constantly detected and controlled, and the pressing force can be set according to the contact state between the tool and the bump and the crushing state during the bonding.

In addition, since each connection portion at the time of bonding has a uniform temperature distribution, variation in the connection state can be reduced, and highly reliable bonding can be performed.

In addition, it is possible to prevent Sn and solder from adhering to the bottom surface of the bonding tool, to reduce connection failure due to one-side contact of the tool, and to improve bonding yield.

Furthermore, since the impact force from the tool can be absorbed during bonding, damage to the IC chip can be reduced,
Highly reliable bonding can be performed. In addition, since the difference in parallelism between the tip and the tool bottom can be absorbed, the adjustment time can be reduced, and the throughput can be improved.

〔Example〕

 Hereinafter, examples of the present invention will be described in detail.

First, the overall configuration of the apparatus to which the present invention is applied and the principle of operation will be described with reference to FIGS. 1 and 3 to 9.

As shown in FIG. 1 and FIG.
An optical system base 102 is mounted on the optical disk, and an optical system plate 104 is fixed on an upper surface 103 thereof. This optical system plate 104
The objective lens 14 is arranged at the tip 105 of the camera. All tables are arranged with the optical axis 106 of the objective lens 14 as a reference axis.

The light emitted from the epi-illumination light source 8 is mirror 9 and shutter 1
1. The light is projected onto the inner lead 2 on the tape 1 via the mirror 12, the half prism 13, and the objective lens 14. Similarly, the light from the light source 16 for oblique illumination drives a rotary solenoid 17 and opens a shutter 18 so that a glass fiber 19,
The light is projected on the IC chip 4 via the ring lighting device. Both reflected light is objective lens 14, half prism 13, mirror 23,
Through the field lens 24, the relay lens 25, and the mirror 26b, an image of one corner is captured by the TV camera 29b via the mirrors 27b and 28b. The image of the other corner is mirror 26
The video signal is captured by the TV camera 29a via a, 27a, and 28a.

As shown in FIG. 1 and FIG. 7, etc., 36 is attached to the upper surface 102 of the base 100, and comprises a pulse motor drive stage 36a, 36b that moves back and forth in a horizontal plane, and a turning table 36c that turns. Above the lower surface 4b of the IC chip 4, the vacuum pump (not shown) solenoid valve 107;
It is held by suction at the central portion 108 of 6b. With the IC chip 4 held, the drive stages 36a, 36 of the XYθ stage 36 are located at positions where the turning center 109 of the turning table 36c and the optical axis 106 match.
work b. Reference numeral 1 denotes a tape. The sprocket base 1 is rotated by rotating a pulse motor 110 via a pair of timing pulleys 111 and 112 and a timing belt 113.
When the drive sprocket 115 attached to and held in a rotatable state is rotated, the tape 1 is conveyed at a fixed pitch in the direction of arrow A via a sprocket hole 1b provided in the tape 1. On the other side of the sprocket base 114, an idle sprocket 117 is held in a rotatable state.

As shown in FIG. 1 and FIG.
4 is attached to the XYZ table 118, and comprises pulse motor drive stages 118a, 118b, 118c, and sprocket base 11
4 can be moved back and forth, left and right, and up and down. A tape guide 119 is attached to the center of the sprocket base 114, and the center 120 of the device hole 1a of the tape 1 and the center of the bonding hole 121 provided in the tape guide 119.
122 guides the tape 1 to a position where it coincides with the optical axis 106, and the lower surface 123 of the tape guide 119 is processed into an arc shape.
3 (see FIG. 6). As shown in FIG. 4, reference numeral 124 denotes a tool table base, and a front-rear sliding plate 1 is provided via a pair of slide guides 125 and 126.
27, the motor 128 is attached to the front-rear sliding plate 127 via a motor shaft 128a, a ball screw 129, and a nut 130. By rotating the motor 128, the front-rear sliding plate 127 is moved by arrows B and C. Can be slid in any direction.

The motor 131 is mounted on the front-rear sliding plate 127, and is connected to the vertical sliding plate 1 via links 136 and 137 connected by a motor shaft 132, a coupling 133, a speed reducer 134, and a pin 135.
Connected to 38. The vertical slide plate 138 is a pair of slide guides 139, 1 attached to the front and rear slide plate 127.
The vertical sliding plate 138 can be slid in the directions of arrows D and E by being guided by 40 and rotating the motor 131.

As shown in FIG.
The pitching plate 143 for holding the bonding tool 7 for thermocompression bonding and adjusting the inclination of the lower surface 7a of the bonding tool 7, the rolling plate 144, the groove 145 for holding the shaft 7b of the bonding tool 7, and the holding plate 146 , A pair of slide guides 148 and 149 for holding the head 147 slidably up and down, and a head base 150 for mounting the same. 151 is a load cell, which is attached to the gap between the upper part 152 of the head base 150 and the upper surface 153 of the head 147, and the head base 150 descends,
The lower surface 7a of the bonding tool 7 comes into contact with the IC chip 4 and the inner lead 2 on the chip stage 6, the head 147 stops, slides via the slide guides 148, 149, the load cell 151 is compressed, and the bonding load can be detected. it can.

The bonding tool 7 has a hole 1 for mounting the heater 154
55 and a hole 157 for attaching a thermocouple 156 are provided.
158 and are maintained at a predetermined temperature by a temperature control unit (not shown). Similarly, reference numeral 159 denotes a stage heater, which heats an upper portion 6c of the chip stage 6, and a thermocouple 160 attached to the upper portion 6c through a hole 6d is attached by a fixing screw 161.
A predetermined temperature is maintained by a temperature control unit (not shown). Insulated table in the center of chip stage 6
6e is attached, there is a hole 6f in the center to adsorb the IC chip 4, and there is a chip stage base 6g in the lower part.
It is attached to the θ stage 36. The chip stage base 6g has a vacuum suction hole 6h, fitting 162, tube 16
3. Connected to a vacuum pump (not shown) via a solenoid valve 107.

A vacuum suction pad 164 for transferring the IC chip 4 is connected to a vacuum pump (not shown) via a fitting 165, a tube 166, and a solenoid valve 167. Suction pad 16
4 is attached to the arm 168, and a pair of slide guides 169
Guided to a, 169b, attached to the hand base 170,
The rod 172 of the air cylinder 171 mounted on the hand base 170 is slidably held up and down,
It is connected to. The air cylinder 171 is connected to a high-pressure air pipe (not shown) via a fitting 173, a tube 174, and a solenoid valve 175. Solenoid valve 175
, The suction pad 164 slides up and down via the arm 168.

The hand base 170 moves in the horizontal plane back and forth, left and right X
The Y stage 176 is attached to the pulse motor drive stages 176a and 176b.

Reference numeral 178 denotes a tray base on which the tray 180 is positioned on the upper surface 179 by a pair of positioning pins 181 and 182.
A plurality of recesses 183 are provided on the upper surface of the IC chip 4.
Are positioned and mounted in advance.

Reference numeral 184 denotes a reel plate, on which a feed motor 185 is mounted, and a square shaft 189 for fitting a square hole 188 of the reel 187 to a motor shaft 186 is mounted. As shown in FIG.
A stopper lever 191 that bends around the pin 190 as a fulcrum is attached to the end of the square shaft to prevent the reel 187 from coming off. 192
Reference numeral denotes a fixed roller, to which a reel plate 184 fulcrum shaft 193 is fixed and held rotatably via a bearing (not shown). 194 is a tension roller, via a slide guide 195 attached to the reel plate 184,
The tension roller shaft 196 is fixed and held rotatably via a bearing (not shown).

Reference numeral 197 denotes a spacer, which is wound around the reel 187 together with the tape 1. Reference numeral 198 denotes a spacer fixing roller, and a fulcrum shaft 199 is fixed to the reel plate 184 and is held rotatably via a bearing (not shown). Reference numeral 200 denotes a tension roller. A tension roller shaft 202 is fixed via a slide guide 201 attached to a reel plate 184, and is held rotatably via a bearing (not shown).

Reference numeral 203 denotes a take-up motor mounted on a reel plate 184. The motor shaft 204 has a square shaft 18 similar to the feed motor 185.
9 is installed. A pair of limit switches 205 and 206 for detecting the position of the tension roller 194 are attached to the reel plate 184, and operate by contacting the slide portion 195a of the slide guide 195, and take up the feed motor 185 and the arrow G direction, respectively. The motor 203 is driven.

In the above configuration, the XY stage 176 is driven, and after the center 164a of the suction pad 164 is moved to the center 4a of the IC chip 4 at an appropriate position mounted on the tray 180, the solenoid valve 175 is moved. Drive air cylinder 171
After the suction pad 164 is lowered via the arm 168, the solenoid valve 167 is driven to drive the upper surface 4 of the IC chip 4.
After vacuum-adsorbing a and raising the arm again, the XY stage 176 is driven to move to a position where the IC chip 4 is mounted on the chip stage 6. At the same time, drive the XYθ stage 36,
The chip stage 6 has been moved to the IC chip 4 receiving position.

Next, the solenoid valve 175 is driven, and the air cylinder 171 is operated.
Is operated, the suction pad 164 is lowered, the solenoid valve 167 is driven, the IC chip is mounted on the chip stage 6, the solenoid valve 175 is driven, the air cylinder 171 is operated, and the suction pad 164 is raised. Let it.

Next, the XYθ stage 36 is driven to return the turning center 109 of the turning table 36c to a position where it coincides with the optical axis 106. At the same time, the pulse motor 110 is rotated, and the driving sprocket 115 is rotated via the timing pulley 111, the timing belt 113, and the timing pulley 112, and the tape 1 is conveyed at a constant pitch in the direction of arrow F as shown in FIG. And tape guide 119
Of the tape 1 to the center 122 of the bonding hole 121
Match 0. At this time, the tension roller 194b descends by its own weight according to the length of the tape 1 conveyed, the limit switch 206b operates, and the winding motor 203 rotates,
The tape 1 and the spacer 197 are wound up while the reel 187b rotates. As a result, the tape 1 is pulled, the tension roller 194b is pushed up, the limit switch 205b operates, and the winding motor 203 stops. At the same time, the tape 1 is pulled, the tension roller 194a is pushed up, the limit switch 205a operates, the feed motor 185 rotates, and the reel 1
The tape 87 and the spacer 197 are sent out while rotating 87a.
As a result, the tape 1 is loosened, and the tension roller 194a
Drops, the limit switch 206a operates, and the feed motor
185 stops. By repeating this series of operations each time the drive sprocket 115 rotates and conveys the tape 1, a constant tension can always be applied to the tape 1.

Next, the shutter 18 is closed, the shutter 11 is opened, and the image from the epi-illumination light source 8 is taken into the TV camera 29a29b.
The position of each inner lead 2 which is closest to the corner and forms a right angle is detected, and as shown in FIG. 9, a difference obtained by adding an offset amount to a predetermined reference position 207a is obtained. The XYZ table 118 is driven to adjust the inner lead 2 to the reference position 207a. Next, the shutter 11 is closed.
Open 18 and view the image from the oblique illumination light source 16 on the TV camera 29a,
At 29b, the positions of the chip corners 4c and 4d are detected, and the differences from the reference positions 207a and 207b are obtained.
And the difference between the front, rear, left and right directions after correcting the inclination is calculated, the XYθ stage 36 is driven, and the reference position 207 is calculated.
The corners 4c and 4d of the IC chip 4 are matched with a and 207b.

After the above operation is repeated and set within the predetermined position deviation, the average position 208a, 208b of the plurality of inner leads 2 is similarly obtained again using the epi-illumination light source 8, and the oblique illumination light source 16 is set. Using a plurality of average positions 209a, 209b of the bump 3
XYθ stage 36 is driven, and the average position of the bumps 3 is compared with the average positions 208a and 208b of the inner leads 2. Match 209a and 209b. This operation is repeated until the position is within the predetermined positional deviation amount, and the alignment is completed.

At this time, if a defect is found in the IC chip 4 by image processing, the XYθ stage 36 is returned to the position where the IC chip 4 before alignment is mounted, and the XY stage 176 and the solenoid valves 167 and 175 are driven. After returning the IC chip 4 to the empty concave portion 183 of the tray 180, another IC chip 4 is mounted on the chip stage 6 again, and alignment is performed by the same means.

If a defect is found in the inner lead 2 of the tape 1, the drive sprocket 115 is rotated, the tape 1 is conveyed at a constant pitch, a new inner lead 2 is positioned, and the positioning is performed again by the same means. Do.

After the alignment is completed, the motor 128 is driven, and the fourth
As shown in the figure, the front and rear sliding plate 127 is moved in the direction of arrow B, and after positioning the bonding tool 7 at the bonding position 39, the motor 131 is driven to drive the coupling 133, the speed reducer 134, and the pin. While lowering the head base 150 in the direction of arrow E via the 135, the links 136, 137, and the vertical sliding plate 138, the bonding tool 7, the head 14
The reaction force of the load cell 151 is detected via 7. The lower surface 7a of the bonding tool 7 comes into contact with the IC chip 4 on the chip stage 6 via the inner lead 2 and after detecting a predetermined load, presses and heats the bonding tool 7 for a predetermined time, so that the inner lead 2 is heated. After being joined to the bump 3, the motor 128 and the motor 131 are driven again to return to the original position,
The thermocompression bonding of one IC chip 4 and the inner lead 2 on the tape 1 is completed. By repeating all the above operations, the inner lead bonding can be performed continuously.

Next, an alignment method for TAB inner lead bonding according to an embodiment of the present invention will be described with reference to the drawings.

FIG. 10 shows a TA to which the alignment method according to the present invention is applied.
FIG. 11 is a perspective view showing a configuration around a bonding position of a B inner lead bonder and a configuration of a detection optical system for alignment, and FIG.
The figure is a functional block diagram showing the overall configuration of the alignment system.

As shown in FIG.
Are provided in order. The tape 1 is fed by a feed mechanism (not shown) at a set pitch of the inner leads 2 in the direction of the arrow (X direction), and the inner leads 2 are sequentially positioned at the bonding position 39. On the other hand, the IC chip 4 is mounted on the XYθ stage 36, mounted on the chip heating stage 6 coated with heat resistant black, and also positioned at the bonding position 39. After aligning all the inner leads 2 on the tape 1 and the bumps 3 formed on the IC chip 4,
After moving the bonding tool 7 to the position shown by the broken line in the Y direction, all the inner leads 2 and the bumps 3 are simultaneously thermocompression-bonded together.

The alignment detection optical system shown in FIG. 10 is constituted by an optical system having the bonding position 39 as the optical axis position, and detects the overlapping inner lead 2 and IC chip 4 simultaneously.
This detection optical system includes an oblique illumination system, an epi-illumination illumination system, a parameter detection system, and an observation system. The oblique illumination system includes a light source 16, a shutter 18 operated by a rotary solenoid 17, a glass fiber 19 for guiding light, and a ring illumination device 20 in which the openings of the fibers are arranged on a circumference. The epi-illumination system includes a light source 8, a mirror 9,
It comprises an aperture 10, a light-blocking shutter 11, a mirror 12, a half prism 13, and an objective lens 14. The pattern detection system consists of an objective lens 14 and an aperture 2 installed at the rear focal point of the objective lens 14.
1, mirror 23, field lens 24, relay lens 25,
It is composed of split mirrors 26a, 26b, mirrors 27a, 27b, 28a, 28b, and TV cameras 29a, 29b. The observation system branches off at the half prism 22 behind the diaphragm 21 and includes a mirror 33, a zoom lens 34, and a TV camera 35.

The alignment method according to the present invention, that is, the alignment of all the inner leads 2 and the bumps 3 is performed in a state where the inner leads 2 and the IC chip 4 are overlapped at the bonding position 39, as shown in the enlarged view 37. In order to perform alignment, two fields of view at the diagonal corners of the IC chip 4
The patterns in 5a and 5b are detected. Further, the positions of the inner lead 2 and the IC chip 4 are obtained in the detected images of both fields of view. Using these position data, the position correction amount of the XYθ stage on which the IC chip 4 is mounted is calculated.
With the fixed, the position of the IC chip 4 is corrected and aligned. After alignment is completed, the inner lead 2 and IC
The chip 4 is bonded by a tool 7 on the spot while being fixed. For this reason, bonding can be performed without deteriorating the alignment accuracy due to the accuracy of the mechanism.

In order to reliably determine the positions of the inner lead 2 and the IC chip 4 from the overlapping state, images in two illumination states, epi-illumination and oblique illumination, are used. Therefore, when performing alignment, place the oblique illumination ring illumination device 20 in the bonding position.
39 Installed above. An illumination method is selected by controlling opening and closing of the shutter 18 of the oblique illumination system and the shutter 11 of the epi-illumination system. Inner lead 2 with any lighting
After the image formation of the IC chip 4 near the field lens by the objective lens 14, it is further enlarged by the relay lens 25. In order to detect the diagonal corner in two fields of view, the detection light passing through the relay lens 25 is split right and left by mirrors 26a and 26b. The light reflected by the mirror 26a is further reflected by the mirrors 27a and 28a,
An image is formed on the imaging surface of the TV camera 29a. The TV camera 29a detects an enlarged image in the upper left half quadrant 38a with the bonding position 39 as the center. Similarly, the TV camera 29b detects an enlarged image in the lower right half quadrant 38b around the bonding position 39. The TV cameras 29a and 29b are respectively held on XY stages (not shown), and by moving these, the positions of the visual fields 5a and 5b are adjusted according to the dimensions of the IC chip 4 to detect the pattern of the diagonal corners. . 30a and 30b are images in which patterns in the visual fields 5a and 5b are detected by the TV cameras 29a and 29b, and these are input to the image processing unit 31 and processed.

At the time of alignment, the inner leads 2 are held with a slight gap between the bumps 3 to correct the position of the IC chip 4. Therefore, in order to simultaneously detect the overlapping inner lead 2 and IC chip 4 with one imaging system as shown in the enlarged view 37, the detection optical system needs to have a wide focusing range. In the present invention, a telecentric optical system is configured by installing the stop 21 at the rear focal position of the objective lens 14 to enable simultaneous detection. The telecentric optical system is used for measuring projectors and the like.Even if the sample moves slightly from the correct focus position in the direction of the optical axis, the image is slightly blurred, but the size is hardly changed. It has the feature of little influence. In addition, since the incident illumination light is guided from the half prism 13 between the objective lens 14 and the aperture 21, by changing the diameter of the aperture 21, the NA (aperture) of the detection system can be changed without changing the field of the illumination. number)
To adjust the depth of focus,
The required focusing range can be obtained.

In this embodiment, an enlarged image is obtained by two-stage imaging using the objective lens 14 and the relay lens 25. Therefore, the magnification of the objective lens 14 is reduced, and the working distance (the objective lens 14
Distance from the tip of the sample to the sample) can be lengthened. This realizes a configuration in which the detection optical system and the bonding tool 7 do not interfere with each other. Further, by using a photomechanical lens having a large pupil diameter as the objective lens 14, a field of view corresponding to various IC chip dimensions can be secured. With the above configuration, a fixed detection optical system that detects the diagonal corners of the IC chip 4 in two fields of view can be realized.

In addition, the ring illumination device 20 used in the oblique illumination system is fixedly mounted on the stage on which the tool 7 moves,
It can be retracted to the position shown by the broken line at the time of bonding without providing a dedicated mechanism.

The observation system splits the image forming light of the objective lens 14 by the half prism 22, forms an image once in the vicinity of the lens 32, reflects the light on the mirror 33, enlarges the image with an appropriate magnification by the zoom lens 34, and
To detect. The zoom lens 34 and the TV camera 35 have an integral structure, and are moved by an XYZ moving mechanism (not shown).
A pattern at an arbitrary position of the inner lead 2 and the IC chip 4 shown in the enlarged view 37 can be observed from directly above at an appropriate magnification.

FIG. 11 shows a functional block diagram of the entire configuration of the alignment system. The alignment system includes a detection optical system 47 for simultaneously detecting the inner lead 2 and the IC chip 4 on the tape 1 overlapping at the bonding position, and a pattern in the two fields of view at the diagonal corners of the IC chip 4. Image processing unit that calculates the inner lead position (hereinafter abbreviated as the lead position) and the IC chip position (hereinafter abbreviated as the chip position) by image processing
31, a position control amount of the XYθ stage 36 on which the IC chip 4 is mounted using the position data in the two fields of view, a mechanism control unit 46 for controlling the position correction amount, and a mechanism control unit 46 for moving the IC chip 4
The XYθ stage 36 includes an XYZ moving mechanism for the tape 1 (not shown).

First, referring to FIG. 12, the mechanism controller 46 uses the position data in the two fields of view 5a and 5b of the diagonal corner received from the image processor 31 to correct the position of the XYθ stage on which the IC chip 4 is mounted. The procedure for calculating the amount will be described. In the figure, XOY is the coordinate system of the XYθ stage 36 on which the IC chip is mounted, x
₁ o ₁ y ₁ is the coordinate system of the field of view 5 a detected by the TV camera 29 a, x ₂ o ₂ y ₂
Is a coordinate system of the visual field 5b detected by the TV camera 29b, and the θ direction is positive in the counterclockwise direction in FIG. R is XYθ
The rotation center position of the stage 36. The lead positions L ₁ , L _{2 and the} chip positions C ₁ , C ₂ are obtained by image processing from within each field of view. The L ₁ and C _1, and L ₂ and C ₂ is the position to be matched after the alignment. The position correction amount of the XYθ stage 36 is
It is obtained as the amount of movement in the XYθ direction for matching ▲ ▼ to ▲ ▼.

 First, the coordinates of each position are defined as follows.

_{R (X R, Y R)} O 1 (X 1, X 1) O 2 (X 2, Y 2) L 1
_{(X 1L, y 1L) C} 1 (x 1C, y 1C) L 2 (x 2L, y 2L) C
₂ (x _2C , y _C2 ) is the XOY coordinate system, is the x ₁ o ₁ y ₁ coordinate system, is x ₂ o
₂ y Coordinates in the ₂ coordinate system. ▲ ▼, ▲
The angles θ _L and θ _C that ▼ makes with the X axis are Becomes Fixed angle [Delta] [theta] in the theta direction of the chip position _{θ L = θ C + Δθ ∴Δθ} = θ L -θ C (= (1) - (2)) obtained as a ... (3). When the tip is rotated around the rotation center R by Δθ, the tip position C ₁ moves to C ₁ ′ and C ₂ moves to C ₂ ′. Since the θ direction has already been corrected, ▲ ▼ becomes parallel to ▲ ▼. Here, the coordinates of C ₁ ′ and C ₂ ′ are represented by an XOY coordinate system, and C ₁ ′ (X ′ _1C , Y ′ _1C ) and C ₂ ′
(X ′ _2C , Y ′ _1C ).

The coordinates of C ₁ ′ are obtained by the following equation.

▲ ▼ = ▲ ▼ + f (Δθ) ▲ ▼ ……
(4) Ie Becomes Further, the correction amounts ΔX ₁ and ΔY _{1 in} the XY direction for making C ₁ ′ coincide with L ₁ are ▲ ▼ = ▲ ▼-▲ ▼ (6) Is obtained as Similarly, the movement amounts ΔX ₂ and ΔY ₂ in the X and Y directions can be obtained from the following expression from the movement of C ₂ .

▲ ▼ and ▲ ▼ are parallel Then, the chip position may be corrected using the XY direction movement amount obtained by the equation (7) or (8).

As described above, the rotation center position R and each visual field position
O ₁ , O ₂ and the lead position in each field of view and IC from the chip position C
The movement correction amount in the XYθ direction of the XYθ stage 36 on which the chip 4 is mounted can be calculated. Note that each visual field position O ₁ , O ₂ is I
It is a value measured in advance each time the position of the TV camera 5a, 5b is moved according to the size of the C chip 4.

An embodiment of the above-described definition of the lead position L and the chip position C will be described with reference to the pattern in the visual field 5 shown in FIGS. 13, 14 and 15 as an example.

One embodiment will be described with reference to FIG. In this case, the chip position C is the intersection of the orthogonal linear portions on the outer periphery of the chip, that is, the chip corner position. The lead position L is most corner side of the first lead position l ₁ determined from the horizontal and vertical directions respectively one lead, the first Lee degree position correction amount [Delta] X, corrected by ΔY position. The first lead position correction amounts ΔX, ΔY
The chip corner position C in a state where the inner leads 2 and the bumps 3 are accurately aligned as shown in FIG.
When indicating the difference between the first lead position l _1. Therefore, the lead position L and the chip position C defined here are positions to be matched after the alignment. Note that the first lead position correction amounts ΔX and ΔY are values set before the execution of the alignment.

Next, a second embodiment is shown in FIG. in this case,
The inner leads 2 and the bumps 3 that are actually joined always have an overlapping portion, and the inner leads 2 and the bumps 3 that are not joined have no overlapping portion. As shown in the figure, the chip position C is an average bump position obtained from the positions of the plurality of bumps 3 included in the visual field 5.
This is a position determined from the X coordinate obtained as the average value of the center position of each bump in the Y direction for all the bumps arranged in the X direction, and the Y coordinate obtained in the same manner. Although FIG. 2 shows the case where two bumps are included in both the horizontal and vertical directions, for example, when six bumps are included in one direction, the average position of the six bumps is used. On the other hand, as the lead position L, an average lead position obtained from a plurality of lead positions included in the visual field 5 is used as in the case of the bump.

According to the definition shown in FIG. 13, the chip corner position C and the first
It is a necessary condition for the alignment lead position l ₁ is within the field of view 5. In the method using the average lead / bump position shown in FIG. 15, the alignment is performed to some extent by some method, and the above-described conditions must be satisfied.

Next, the image 30 detected by the TV cameras 29a and 29b shown in FIG.
A method of obtaining the lead position L and the chip position C defined as described above from a and 30b will be described with reference to FIGS.

A method of detecting the lead position will be described with reference to FIG.
In the case of detecting the lead position, the oblique illumination system shutter 18 shown in FIG. 10 is opened and the epi-illumination illumination system shutter 11 is closed to perform oblique illumination. In this illumination, light is emitted from an opening of a fiber arranged on the circumference of the ring illumination device 20. The diameter of the opening is larger than the outer diameter of the IC chip 4, so that the illumination light is applied to both surfaces of the inner lead 2 and the IC chip 4.
The light enters obliquely evenly from each direction. The inner lead 2 is formed by etching a copper foil, and its surface is rough, so that many irregularly reflected components are detected brightly.
On the other hand, the surface of the IC chip 4 is detected darker than the inner lead 2 because the surface thereof is smooth and has less diffuse reflection components. Therefore, if the state in which the inner lead 2 is superimposed on the IC chip 4 is detected by oblique illumination, an image in which the inner lead 2 is made bright and visible is obtained. When this is binarized, a binary image as shown in FIG. 16 (a) is obtained. Here, a white portion indicates “1” and a hatched portion indicates “0”. A procedure for detecting the lead positions arranged in the X direction from the binary image will be described below.

(1) “1” along the Y direction with the projection width shown in FIG.
A partial projection waveform lead-pr (X) (FIG. 9B), that is, a waveform indicating the number of “1” pixels at the same X coordinate is created. In this case, the projection width is set so that the upper end of the image is the starting point and the maximum value of lead-pr (X) is equal to the projection width. In particular,
First, projection processing is performed on the entire image, and the maximum value of the projection waveform is compared with the projection width. If they are not equal, the projection processing is repeated again with the maximum value as the projection width. With this processing, the projection waveform under the above conditions is automatically obtained.

(2) the intersection L _l of lead-pr (X) with a threshold Thl, obtains the read position L _C as the midpoint of the L _r. Thl uses a value obtained by multiplying the projection width by an appropriate ratio r (0 <r <1).

The lead positions arranged in the Y direction are detected in the same procedure. According to the above-described procedure, all the lead positions included in the visual field are detected. Can determine the average lead position shown in the first lead position l _l and 15 shows from these in Figure 13.

Next, a method of detecting the chip corner position as the chip position C will be described with reference to FIG. The chip corner position detection is performed after the lead position detection described above is performed. In this case, the epi-illumination system shutter 11 shown in FIG. 10 is opened, and the oblique illumination system shutter 18 is closed to perform epi-illumination. In this illumination, the illumination light is incident on the inner lead 2 and the surface of the IC chip 4 almost perpendicularly. Since the surface of the IC chip 4 is smooth, although the circuit pattern is bright and dark, it is detected bright because of the large number of specular reflection components. On the other hand, since the inner lead 2 has a rough surface, it has a large amount of irregularly reflected components, and is detected darker than the surface of the IC chip 4. Bump 3 on the surface of IC chip 4
Has a rough surface due to plating, and is detected as dark as the inner leads 2. Therefore, IC chip 4
When the state where the inner lead 2 is superimposed on the inner lead 2 is detected by epi-illumination, an image in which both the inner lead 2 and the bump 3 are dark and only the chip surface is bright is obtained. When this is binarized, a binary image as shown in FIG. 17 (a) is obtained. In the chip corner position detection method, an approximate straight line of the outer periphery of the chip in both the X and Y directions is obtained from the binary image shown in FIG. 3A, and the chip corner position C is detected as the intersection of these two straight lines. A procedure for linearly approximating the outer periphery of the chip in the X direction will be described below.

(1) “1” along the X direction with the projection width shown in FIG.
A partial projection waveform pelet-pr (X) (FIG. 9B) is created. This waveform is searched from the left side with the threshold Thp, and the first intersection is set as pe. A search point shown in FIG. 4B is set with a point advanced by dp from pe as a start point. The reason why dp is used is to set a range for correctly searching the chip outer periphery in the X direction even when the inclination θ of the IC chip 4 is large.

(2) The binary image shown in FIG. 17 (a) is searched along the Y direction from the upper end of the image within the search range, and the image is changed from “0” (shaded portion) to “1” (white portion). First extract the points that change. (FIG. (C)) (3) In FIG. (C), in the portion where the inner lead 2 overlaps the IC chip 4, a portion other than the chip outer periphery is extracted. Therefore, the extraction point near the lead portion is removed (masked) by using the lead position data already obtained from the above-described oblique illumination image. ((D) in the figure) (4) The result of extracting the outer periphery of the chip in the figure (d) is approximated by the least square method. (E in FIG. 13) The same processing is performed in the Y direction, and the chip corner position shown in FIG. 13 can be obtained as the intersection of the approximate straight lines in both the X and Y directions.

Next, a method for obtaining the average bump position as the chip position C will be described with reference to FIGS. The bump position detection is performed after the lead position detection in the same manner as the above-described chip corner position detection, and an image detected by incident illumination is used. FIG. 18 (a) shows the procedure for detecting the positions of bumps arranged in the X direction in this binary image.

(1) A “1” portion projection waveform pelet-pr (X) (FIG. 18B) along the X direction with a projection width p shown in FIG. 18A is created. This waveform is searched from the upper side with the threshold Thp to find the first intersection point pe.

(2) Set the projection width b based on pe obtained in (1),
A “0” portion projection waveform bump-pr (x) (FIG. 3C) along the Y direction is created. The projection width b is limited to a range in which bumps around the IC chip 4 are present, and is set so as not to be affected by a pattern inside the chip.

(3) The method of obtaining the bump position from the projected waveform bump-pr (x) in FIG. 18 (c) will be described with reference to FIG. For this purpose, the following two methods are selected depending on the alignment between the inner leads 2 and the bumps 3.

(I) Both left and right edges of bump 3 are inner leads 2
Optionally visible from the side (FIG. (D) -1) read position detection (FIG. (D) -2) to the start point of known lead position L _C, bump-pr (x) to the right and left on the threshold Thp I searched, each first intersection B _l, Request B _r. Obtaining a bump position B _C As these midpoints. (FIG. (D) -3) (ii) When one edge of the bump 3 is hidden under the lead and cannot be seen (FIG. (E) -1) Intersections B ′ _l , B ′ as in (i) _{Find r} . Then, from the intersection having the longer search distance (in this case, B ′ _l ), b _W / 2
The: (b _W bump width) by a position backward to the lead position L _C side and the bumps located B _'C.

The method (i) or (ii) to determine which method to use
The search is performed using the longer distance S. Ie In the case of (l _W : lead width), (i), In the case of (2), the bump position is obtained by the method (ii).

By searching for bump positions as described above from all the lead positions in the X and Y directions known by the lead position detection, all bump positions in the visual field can be detected. This makes it possible to detect the average bump position shown in FIG.

Next, an embodiment of the image processing unit 31 for detecting the above-described lead position L and chip position C will be described with reference to FIG.

The image processing unit 31 includes a switch 48, an A / D converter 40, a multi-valued memory 41, a binarization circuit 42, a binary memory 43, and a projection processing circuit 4.
4, consisting of a microcomputer 45. The image signals detected by the TV cameras 29a and 29b are switched by a switch 48 and processed alternately. A / at 40
After the D conversion, the data is temporarily stored in the multi-valued memory 41. Further, the image is binarized by 42 and the binary image is stored in the memory 43. This 2
A projection width is set for the value image by the microcomputer 45, and a projection waveform is obtained by 44. The microcomputer 45 inputs a projection waveform, that is, a one-dimensional waveform, and processes this to detect the positions of leads, bumps, and the like. In addition, the microcomputer 45 includes shutters 18 and 11 of the detection optical system.
The opening and closing of the camera is controlled, and the lighting method is selected according to the flow of image processing. The position data in each image of the lead position L and the chip position C detected from the images in the two fields of view are sent to the mechanism control unit 46.

The position detection method according to the present invention is performed only by repeating the projection waveform processing, and the device configuration of the image processing unit 31 is simple.
In addition, the scale can be reduced.

Next, an embodiment of an alignment operation flow of a TAB bonder to which the alignment method according to the present invention is applied will be described with reference to FIG.

Alignment of inner lead 2 and IC before bonding
After positioning the chip 4 at the bonding position 39, the chip 4 is started (48). First, oblique illumination is performed, and the lead position is detected 49 in any field of view to obtain the first lead position, and the lead position is corrected (50). This aims at positioning the inner lead 2 below the crimping surface of the bonding tool 7, and corrects the position in the XY direction. The correction target position is obtained as follows. As shown in FIG. 10, the tool 7 merely repeats the movement of a certain distance in the YZ direction, and the position of the crimping surface does not change. Therefore, before performing actual bonding, a thin plate on which a polyimide tape or the like is adhered is placed on the chip stage 6, and the tool 7 is pressed in the same manner as bonding. Burn marks are formed on the tape on the thin plate, which can be used for alignment.
By observing with the TV camera 29a (or 29b), the position of the crimping surface can be known. Since the size of the crimping surface is almost equal to the size of the IC chip 4, the IC chip 4 is adjusted to the position of the burn mark. Then, the first lead position in a state where the inner lead 2 is aligned with the chip is a target position, and the lead is aligned with this position.

After positioning the inner lead 2 on the crimping surface of the tool 7 as described above, the lead position detection (51) is performed again. Thereafter, the leads are fixed and the chips are moved to perform alignment. Therefore, the first lead position and the average lead position detected in 51 are the target positions for the subsequent alignment operation. Next, a chip corner position or an average bump position is detected by the chip position detection 52 according to the alignment state of the inner lead 2 and the bump 3. At 53, the amount of deviation between the lead position L and the chip position C is checked, and if the target alignment accuracy has been reached, bonding (56) is performed. If the accuracy is not achieved, correct the chip position (55)
Then, the chip position detection (52) is repeated again.

FIG. 21A shows the contents of the deviation amount check 53. Chip position C with respect to lead position L in two fields of view 5a, 5b
When the deviation amounts dx ₁ , dy ₁ , dx ₂ , and dy _{2 in} the XY directions are all less than or equal to the target accuracy D, the alignment is terminated. That is, as shown in the figure, if the chip position C is within the alignment accuracy range 58 of ± D in the XY direction with respect to the lead position L in each field of view, the alignment is terminated and bonding is performed.

However, if an actual product is aligned only by the above-described displacement amount checking method, the accuracy may not be reduced and bonding may not be performed. One of the causes is poor work accuracy. For example, the tape 1 on which the inner leads 2 are formed is made of a heat-resistant material selected and used.
Thermal deformation is inevitable. FIG. 22 shows an example of alignment when the tape is expanded. In this embodiment, the movement of the XYθ stage 36 in the XY direction such that the chip position C coincides with the lead position L in one of two fields of view in principle, as described in the expression (9) above. You should be able to align if you calculate the amount. However, when the tape expands and becomes larger than the IC chip 4 as shown in FIG. 22, if the movement amount in the XY direction is obtained only by the visual field 5a, it becomes as shown in FIG. In other words, although the inner leads 2 and the bumps 3 are well aligned in the field of view 5a,
In 5b, the displacement increases. As a result, if the expansion is large, the displacement vector ▲ in the field of view 5b becomes large, and the alignment is not completed. The same thing occurs when attention is paid to the field of view 5b as shown in FIG.

Therefore, in the alignment method of the present invention, as another embodiment of the calculation of the movement amount in the XY direction, the average value of the movement amounts calculated by focusing on each visual field obtained by the equations (7) and (8) is calculated by the movement amount in the XY direction. And By doing so, in the case of FIG. 22, it is possible to perform alignment with the same shift amount in both the visual fields 5a and 5b. As a result, even for a target having a low work accuracy, an optimum alignment corresponding to the work accuracy can be realized.

The above-described average alignment of the two fields of view can provide an optimum alignment even for a work with a poor accuracy to some extent. However, even in this method, if the work accuracy is further worse, the alignment cannot be completed because the accuracy has not been reached. FIG. 21 (b) shows a state in which the correction in the θ direction and the correction of the average position in the XY directions of the two visual fields have been completed. However, displacement vectors ▲ ▼, ▲
▼ are almost equal in size, and both are larger than the alignment accuracy range 58 shown in FIG. 9A, and therefore the alignment cannot be completed only by the above-described displacement amount checking method. In order to prevent this, in the present invention, the following method is added as another embodiment of the deviation amount checking method. (1) The correction amount in the θ direction is sufficiently small.

(2) In both fields of view, the amount of displacement is larger than the target accuracy.

(3) Displacement vectors ▲ ▼, ▲
The vector (Δdx, Δdy) of the sum of ▼ is smaller than the set Δd in both the XY directions.

If all of the above conditions (1) to (3) are satisfied, it is determined that further alignment is impossible, and the bonding is stopped (57) as shown in FIG.

In the alignment operation flow shown in FIG. 20, an alignment count check 54 is performed before the chip position is corrected after the shift amount check 53 is completed. Originally, if the calculation of the position correction amount is correct, the alignment should be completed by several chip position corrections. However, for some reason, the convergence of the accuracy is poor, and the bonding is stopped (57) even when the chip position is corrected more than a specified number of times.

After the bonding is stopped, an alarm is issued to inform the operator of the bonder that a problem such as a work defect or a device defect has occurred, and the operator takes necessary measures. By doing so, it is effective in early detection of a tape or the like with low precision, prevention of a potential defect by forcibly bonding a workpiece with low precision, prevention of a reduction in tact due to an infinite loop of the alignment operation, and the like.

Next, an alignment method from a state where the corner portion of the IC chip 4 is detected in only one visual field will be described with reference to FIG. In the present invention, the state in which the inner lead 2 and the corner portion of the IC chip 4 are detected in the two visual fields is premised on the calculation of the position correction amount of the XYθ stage 36. Inner lead 2
As shown in the operation flow of FIG. 20, each time the tape 1 is fed by one pitch, the lead position detection 49 and the lead position correction 59 are performed.
I do. For this reason, the corner portion can be positioned at a fixed position in both visual fields. On the other hand, the IC chip 4 is mounted on a chip stage 6 shown by a broken line in FIG. If a positioning device is provided in the middle of the transfer, it is easy to insert the corner of the IC chip into two fields of view. In this embodiment, an application example of the alignment method according to the present invention for reducing the supply accuracy of the IC chip 4 to a bonder having no such a positioning mechanism will be described.
FIG. 23 (a) shows a case where the corner portion of the IC chip 4 is detected in the visual field 5a but not detected in the visual field 5b. In such a case, the correction amount in the XY direction is obtained and aligned only from the position detected in the visual field 5a. That is, the first in the field of view 5a
The position correction amount (ΔX,
ΔY). After the correction, as shown in FIG.
The corner of the IC chip can be detected in any one field of view.
Thereafter, the position in the XYθ direction is corrected based on the position data in the two visual fields, and the alignment operation may be repeated until the state shown in FIG. However, if the tip tilt is large, XY
Even if the direction is corrected, the corner of the IC chip 4 may enter only one field of view. In this case, since the correction amount in the θ direction cannot be calculated and alignment cannot be performed any more, the automatic alignment is stopped, and the bonder operator is requested to assist. According to the present embodiment, in addition to the alignment device according to the present invention, a mechanism dedicated to positioning the IC chip 4,
No device or the like is required, and the configuration of the bonder can be simplified.

In the present invention, there is a value set in advance before the alignment operation. Hereinafter, an embodiment of the setting method will be described.

First, a method of setting the first lead position correction amounts ΔX and ΔY shown in FIG. 14 will be described. First, as shown in the drawing, the inner leads 2 and the bumps 3 are aligned together in two fields of view. Actually, the TV cameras 29a and 29 shown in FIG.
While observing the detected image of b on a monitor (not shown),
The stage can be moved manually for alignment. After the alignment, the lead position detection 51 and the chip position detection 52 shown in FIG. 20 are executed, and the first lead position and the chip corner position are automatically detected by the algorithm described in the present embodiment. As a difference between these two positions, a first lead position correction amount in both visual fields is obtained. According to this method, the correction amount can be obtained according to the appearance of the peripheral pattern of the IC chip 4. Further, the change depending on the type can be easily performed only by performing the manual alignment once before the bonding.
Further, since the design values are not used, it is possible to provide an optimum correction amount even for a workpiece with low accuracy.

Next, a method of setting the visual field positions O ₁ and O ₂ shown in FIG. 12 will be described. For this purpose, an alignment mark 60 is provided on the chip stage 6 shown in FIG. Since the black painted surface of the chip stage 6 is detected as dark,
The mark 60 is formed by forming a pattern so that a material that can be detected brightly becomes a surface. Chip stage 6 is XYθ
The stage 36 is fixed on the θ stage, and assembled so that the alignment mark 60 coincides with the rotation center of the θ stage. When assembling, rotate the θ stage to mark 60
Observe and know the eccentric direction from the movement. By finely moving the chip stage 6 using the fine adjustment mechanism 61,
Eliminates the eccentricity between the mark 60 and the rotation center of the θ stage.

Further, as shown in FIG. 3B, the alignment mark 60 is detected in each field of view 5 and, based on the position of the mark 60 in the field of view and the stage position at the time of detection, the origin position of both fields of view is determined.
O ₁ and O ₂ can be obtained in the stage coordinate system.

Next, an embodiment of the means for detecting the position of the alignment mark 60 in the field of view will be described with reference to FIG. First, the XYθ stage 36 is moved, and the positioning mark 60 is positioned in the visual field 5. This mark 60 is captured by a TV camera, binarized,
A "1" portion projection waveform is created in the same manner as the lead position detection. The projection width is set to the entire image, the intersection between the waveform projected in both the X and Y directions and an appropriate threshold is obtained, and the position of the mark 60 in the field of view is obtained from the middle point. In another embodiment, as shown in FIG.
The binary image of the registration mark 60 stored in the
The position of the mark 60 is determined by performing pattern matching, which is a known technique, with the image of the field of view 5 in which 60 has been detected. In another embodiment, the position can be obtained by circumscribing a cursor or the like to the outer periphery of the mark 60 detected in the image of the visual field 5.

Next, referring to FIG. 25, the bonding position 39 according to the present invention is shown.
5 shows an embodiment of an inner lead bonding method utilizing the features of the alignment method by detecting the inner lead 2 and the bump 3 simultaneously.

FIG. 25 (a) shows the configuration near the bonding target,
Reference numeral 62 denotes a tape guide for guiding the tape 1, and reference numeral 63 denotes a tape Z stage connected to the tape guide 62. In the alignment, the bump 3 is aligned with the fixed inner lead 2 by moving the IC chip 4. Therefore, a slight gap is provided between the inner lead 2 and the bump 3. After the alignment is completed, if thermocompression bonding is performed on the tool 7 in this state, cracks 64 are likely to be generated in the bent portion of the inner lead 2 as shown in FIG. In addition, as shown in FIG. 3C, the inner lead may be shifted laterally during pressing, and the alignment accuracy after crimping may be deteriorated. The lateral displacement is caused by a failure of the parallel contact of the crimping surface of the tool 7 to the surface of the IC chip 4 or the pressing direction of the tool 7
It is considered as a cause that it is not perpendicular to.

As a solution to the above problem, as shown in FIG. 25 (d), after the alignment is completed, the tape Z stage 63 is lowered, and the inner leads 2 are brought close to the bumps 3 and then pressed. By doing so, as shown in FIG.
Since the bonding can be performed without bending the inner lead 2, the occurrence of the clutch 64 can be prevented. On the other hand, the lateral displacement of the inner lead 2 hardly occurs when the tool 7 is pressed. However, since the tape Z stage 63 is driven after the alignment, if the moving direction of this stage is inclined with respect to the IC chip 4, a lateral displacement similar to that shown in FIG. This can be dealt with by an alignment method because of a phenomenon with reproducibility. That is, as shown in FIG. 21 (a), the lead position L of each field of view, which is the target position of the alignment,
The amount of deviation caused by the operation of the tape Z stage 63 is added as an offset amount, and alignment is performed using the offset amount as a target position. In this method, alignment is performed with the inner leads 2 and the bumps 3 separated. After the alignment, when the tape Z stage 63 is lowered, the inner leads are shifted laterally by a fixed amount, so that the alignment can be performed correctly in the state of FIG. 25 (d).

Further, the above-mentioned shift amount can be obtained by detecting the lead position before and after the tape Z stage 63 descends, and obtaining the difference between the two.

In the present embodiment, two diagonal corners of the IC chip 4 are detected. Further, two fields of view are detected simultaneously by two heads, but two fields of view may be detected by driving this with a detection system of one head. In addition, the position detection method described in the present embodiment can be applied to only one visual field, and it is needless to say that it can be applied to alignment in the XY direction. In addition, this embodiment can be applied as a part for finally increasing the accuracy of the alignment by using in combination with a chip and tape position detecting method using a known pattern matching technique. In this embodiment, a ring illumination device for oblique illumination is used.
Although 20 was used, this can be replaced by any lighting device or the like having a function of uniformly illuminating an object obliquely, such as a lighting device composed of a plurality of glass fibers. In addition, epi-illumination is not limited to guiding light through the objective lens 14, and light can be guided from between the objective lens 14 and an object.

The present invention realizes alignment in the XYθ direction completely without contact after the IC chip 4 is supplied, and can prevent defects such as cracking and chipping of the chip during alignment.

Next, an embodiment of a pressing mechanism used in the present invention will be described with reference to the drawings.

FIG. 26 shows a vertical movement mechanism of the bonding tool according to the present invention. (A) shows a front view and (b) shows a side view. The tool stage 502 slidably mounted on the base 501 in the Y direction is driven by a permanent magnet type DC servo motor 503. A permanent magnet DC servo motor 505 for driving the tool is provided on the tool stage 502, and the angle of the input link 508a of the link mechanism 508 can be changed via a harmonic drive 507 for deceleration.
The output link 508c of the link mechanism is a sliding guide mechanism 517.
Along with the tool support 509 and the tool 510 coupled therewith. Tool support
The notch 509 has a notch 509a, and when a force component 518 in the Z direction is applied to the tip of the tool, it is elastically deformed as shown in FIG. 27 and presses the load cell 511 against the upper surface of the tool support. The load cell 511 generates a distortion signal according to the contact force. The distance between the load cell 511 and the upper surface of the tool support 509 can be changed by finely adjusting the screw 512. For the initial adjustment of the above interval, the tool 510 is
In the state where no external force equivalent to 18 is received, the load cell
Screw 51 until a contact force is generated between 511 and screw 512
Perform by inserting 2.

Next, a block diagram of the system is shown in FIG. In this figure, reference numeral 519 denotes a main controller for managing the entire bonding apparatus, and 520 shown by a dashed line is a tool vertical movement controller. The bonding command 522 sent from the main controller 519 is processed by the bonding algorithm 523 to control the driving of the tool. When bonding is completed, the bonding algorithm analyzes the data being bonded and transmits the result to the main controller 519.
I do.

Details of the bonding algorithm will be described later.

The tool vertical movement control device 520 detects the position of the tool in the Z direction by inputting the output pulse signal 528 of the optical encoder 506 directly connected to the tool vertical movement drive motor 505 to the counter 529 and performing a counting process. ing. At the same time, the load cell force detection value 527 determined by the rotation angle displacement 525 of the motor 505 and the displacement of the tool 510 is input to the A / D converter 531 via the amplifier 530 and is captured as a digital amount. . The detection of the position in the tool Z direction and the detection of the force are performed at the same time interval. Based on this, the bonding algorithm 523 performs a predetermined calculation within each time interval, and outputs the current output to the motor 505. The value is set.
This output current command value is input to the current amplifier 524, and the current of the motor is controlled.

The operation of the tool vertical drive mechanism with the above configuration
This will be described with reference to the flowcharts of FIGS. 29 and 30.

Before the start of each bonding operation, the tool stage 502 is retracted backward, and is in a state of waiting for the completion of alignment between the chip and the lead (302). In this state, the communication from the main controller 519 is constantly monitored to determine the presence or absence of a bonding command (304). Main controller 5
19 is the Y coordinate of the chip position when alignment is completed
Each data of T _y , pressurization target value F _B , and pressurization time target value T _B is transmitted as a bonding command and a block. Upon receiving this, the tool vertical movement control device 520 stores each data as a variable of the control program,
The operation shifts to the drive control state of the tool vertical movement mechanism (306).

The tool first moves to the upper limit in the Z direction and stops (308). Then the tool stage is moved forward, and stops the tip position T _y (310). Next, the tool moves in the Z direction, approaches the point _AZ where the distance between the tool tip and the tip can be expected to be a predetermined value, and stops (312). Next, the predetermined speed command value Vref is substituted into the command value of the control algorithm (314), and the tool approaches the chip and the lead while maintaining the speed of the tool in the Z direction at a constant value using the speed control algorithm. (316). At this time, the signal value from the load cell is monitored at the same time to determine whether there is contact between the tool and the chip (318). Thereafter, the processes of (316) and (318) are repeated at predetermined time intervals,
The process is continued until the signal value detected from the load cell becomes a value corresponding to the target value of the pressing force (320).

Feedback control is performed on the detection signal value of the load cell corresponding to the pressing force target value (322), and when the target pressurizing time value and the actual feedback control time of the pressing force become equal (324). ), The tool is moved to the upper limit point in the Z direction and stopped (326). After moving the tool stage to the rear limit point in the Y direction and stopping it (32
8) The deviation distribution of the force detection values during pressurization is totaled (330), the result is transmitted to the main controller, and one cycle is completed (33).
2).

FIG. 31 schematically shows the movement of the tool according to the present embodiment. FIG. 31 (a) shows a time change of each coordinate. (B) shows the spatial movement of the tool. , Is a positioning mode, ~ is a mode in which force is continuously detected while lowering the tool at a low speed, and is a feedback control based on the detected force to maintain the pressing force at a target value. Mode.

FIG. 32 is a block diagram showing the control calculation method in each mode. In this figure, 531 is the 28th
It is a block in which the current amplifier 524, the motor 505, and the optical encoder 506 in the figure are integrated, and is hereinafter referred to as a motor system. Numeral 532 denotes a differential element for calculating a difference between two terms before and after the time series signal. In FIG. 532, a mechanism for approximately differentiating a motor output shaft rotation angle which is an output of the motor system to obtain a rotation angular velocity as a digital quantity. It is. 533 is a sample and hold element, 534 is a difference element, and an element opposite to 532, which is an integrator that performs approximate integration. Z is a Z conversion factor.

In each mode, time series m of output m to motor system
(K) (K = 1, 2,...) Is calculated as follows.

(A) m (K) = K _p (θ _r (K) −θ (K)) − K
_v {[theta] (K)-[theta] (K-1)} In this mode, [theta] _r is not a constant value, but a value obtained by internally dividing the given initial position and target position.

(B) i (K) = i (K−1) + K _vi ｛v _ref − (θ
(K) −θ (K−1))｝; i (θ) ≡0 m (K) = i (K) −K _vv {θ (K) −θ (K−1)} As described above, this mode _Vref is a constant value.

(C) i (K) = i (K−1) + K _fi ｛F _ref −K _F (θ
(K) −T _Z (K))｝ m (K) = i (K) −K _ff ｛F _ref −K _f (θ (K) −T
_Z (K))｝ In this mode, Fref is also used as a constant.

When bonding is performed by the tool vertical movement mechanism described above, the condition setting as shown in FIG. 33 can be realized. First, due to the repulsive force characteristic K _F of the object contacted by the tool, the rate of increase of the pressing force, that is, the inclination in FIG.
Approximately given by K _F · v _ref . Therefore, by arbitrarily changing the low-speed command value, which is a parameter that can be set, the inclination at the rise of the pressing force can be arbitrarily realized.
Also, the pressing force target value F _ref can be arbitrarily changed because it can be a variable in the program.

As an application example of this embodiment, a description will be given of a case where bumps on a chip to be bonded have uneven height. No.
FIG. 34 is a flowchart of the application example, and FIG. 35 shows the behavior of the tool and the bump in this application example.

Initially, the tool is separated from the pellet (IC chip) and leads (400). When there is a command to start bonding, the control device detects the amount of movement of the tool and holds this as a current value in the memory (402). Then the process moves the initial pressure f _o to repeat the execution of the tool drive control program after you assign to the pressure setting value f _r.

First pressing force is detected (406), the servo computation operation is performed for the pressure-force setting value f _r, the result is output to the means for changing the driving force of the tool (408). Subsequently the amount of movement of the tool is detected, it is compared with the value x _o when previously detected (410, 412). If x _o and x _n is not equal tool moving, pressure-force setting value f _r is returned to the beginning of the repeated portions while holding unchanged value. Is this a tool is not yet in contact with the pellets, corresponding to either the status of whether the crush bumps initial pressure f _o following pressure.

If x _o and x _n it had no movement equally become tools back to the top of the repeated portion by increasing the pressure set value f _r increment Δf (414,416). This corresponds to a state in which both are balanced while the tool presses the bump.

After continuing this repeated until pressure setpoint f _r becomes greater than the predetermined bonding startable pressure F _c, performs a bonding operation in which the bonding load F _b predetermined and pressure-force setting value (418), Thereafter, the tool is retracted to complete the bonding (420, 422). At this time, the value of F _c is sufficiently large number of bumps experimentally determined to become in contact with the tool as a condition.

FIG. 35 illustrates a change in the contact state between the tool 7 and the bumps 3a and 3b corresponding to the signal of the above algorithm. Here, for ease of understanding, it in contact with the slowest tool 7 in the previous bump 3a, and the pressure set value f _r is greater than F _c bump 3 is in contact with the tool 7 earliest illustrated Limited to the two bumps 3b.

FIG (a) is a state where the drop as a set value of the initial pressure f _o without contacting 3 both which bumps tool 7.
(B) the tool 7 is in contact with the bump 3a, and shows a state in which balanced in pressure f _o. (C) is a state in which the tool 7 is descending while crushing the bump 3a while gradually increasing the pressing force. At this time, the pressing force takes the minimum value necessary to keep crushing the bump 3a, No force or excessive pressure is applied to the bump 3a. (D) is a state immediately before the lowering of the tool 7 and the contact of the bump 3b with the tool 7. This time bump 3 a predetermined number including the bump 3a is crushed by applied pressure F _c -.DELTA.f. (E) the bump 3b is in contact with the tool 7, with the pressure set value is larger than the F _c, performs bonding of regular was predetermined value F _b the bonding load from this state.

According to this application example, even when there is an error in the height of the bump 3 for each IC chip (pellet) 4, the impact load change at the time of contact between the tool 7 and the bump 3 is within a predetermined initial pressing force. There is an effect that can be suppressed. Further, there is an effect that an excessive pressing force on a small number of bumps 3 in the initial stage of bonding is avoided.

According to the present embodiment, since the parameters of the pressing force can be changed during bonding, it is possible to set conditions and change algorithms according to the characteristics of the object, which is effective in reducing damage to the chip.

Next, a method of heating the bonding tool used in the present invention will be described.

Conventional bonding tool 700 is usually Inconel,
This is the structure shown in FIG. 36 (a) in which sintered diamond is bonded to the bonding portion.

To improve the bonding quality, the temperature of the bonding surface at the tip of the tool 700 needs to be uniform. No.
FIGS. 37 (a), (b) and (c) show the effects of simulating the temperature distribution of the tool assuming various tool shapes. The most uniform temperature distribution in the tool is the concave tool shape shown in FIG. 37 (c). A tool having the shape shown in FIG. 36 (b) was prototyped and tested. The distribution could be within ± 2 ° C. Therefore, by adopting the shape shown in FIG. 36 (b), the quality of bonding can be improved.

36 and 37, reference numeral 700 denotes a bonding tool, 701 denotes a heater, 702 denotes a thermocouple, and 703 denotes an isotherm.

In the conventional tool shape, the tool 700 is heated by a heater 701 embedded above the tool 700 as shown in FIG. 36 (a). Structurally, heat is supplied to the bonding surface by conduction through Inconel tool 700. For this reason, due to heat loss due to radiation from the tool surface, the temperature distribution in the tool must be as shown in FIG. 37 (b), and there is a difficulty in making the temperature of the bonding portion uniform. In the case of continuous bonding, the temperature of the tool decreases each time, and it takes several seconds to recover the temperature, resulting in an increase in bonding tact time and a decrease in throughput. Therefore, as shown in FIG. 38, by adopting a structure in which heat is supplied from a heater block 704 surrounding the periphery of the tool 700, the temperature of the bonding portion can be made more uniform.

On the other hand, TAB inner bonder has a bonding temperature of 450
One problem is that the thermal shock applied to the chip is extremely large because it is as high as 550550 ° C. In particular, due to the heating method involving pressurization, problems such as chip damage and tape deformation due to bonding are becoming more and more problematic due to the increase in chip size and the increase in the number of pins. Therefore, most of the inner bonding apparatuses are arranged such that a heater 601 is arranged in a bar or other shape in a stage 600 as shown in FIG.
It is designed so that preheating of about ℃ is possible. The problem with this structure is that one is that little consideration is given to the uniformity of the temperature distribution of the stage 600, and the other is that the pressure during bonding is usually about 3 to 10 kgf. Therefore, the heater 601 may be damaged in some cases.

Therefore, in the present embodiment, taking this point into consideration, the support stage 600 of the tool is circular, and the periphery thereof is a ring-shaped heater 6.
The point is that it is surrounded by 02. As a result, a substantially uniform temperature distribution is obtained on the stage, and a load due to tool pressurization is not transmitted to the heater 602, so that breakage of the heater can be prevented.

In the stage structure of the inner bonder, the entire stage is usually mounted on an XY-θ table for positioning the tape and the chip. Stage 600
In this method, since the heat generated by the bonding tool or the preheating is involved, a cooling and heat insulating structure is employed to prevent the accuracy of the XY-θ table from deteriorating due to the heat.

Particularly, when preheating is actively performed, the amount of heat generated is large, and therefore, as shown in FIG.
There is also an example in which heat is dissipated by attaching a radiating fin 603 or the like to the periphery.

With multi-pin inner bonding (200 pins or more)
Although preheating is effective, it is required that the positioning accuracy between the leads and the bumps be extremely high, so this heat dissipation structure is particularly important.

In view of the above points, in the embodiment, in order to more effectively release heat, as shown in FIG.
A water-cooled water channel 604 is provided with a hollow structure. By adopting this structure, it is possible to preheat up to 300 ° C. without increasing the temperature of the XY-θ table, as well as raising the preheating temperature more than before.

As shown in FIG. 41 (a), the conventional bonding tool 700 uses a sintered diamond on the bonding surface to uniformize the temperature within the bonding surface and minimize the adhesion of oxides such as Sn plating on the leads. It has a structure with 710 attached. However, when sintered 710 is sintered,
Since Co is used as an auxiliary agent, Co is concentrated and remains at the grain boundaries and the like on the sintered surface, and oxides and the like adhere to this, so that 10 to 50
Polishing is required for each IC.

As a result of the analysis, it was found that this Co became a nucleus and that SnO ₂ mainly adhered. Therefore, in this example, FIG. 41 (b)
Focus on the point where diamond thin film can be formed by plasma CVD etc.
A bonding experiment was performed by forming a film 711 of a degree. As a result, it was found that adhesion was rare even at about 100 IC, and polishing was extremely easy, confirming the effectiveness.

In FIG. 41, reference numeral 712 denotes a heater hole, and 713 denotes a thermocouple hole.

The conventional bonding stage uses a highly rigid material such as ceramic or stainless steel for the stage. Therefore, even if the parallelism between the bonding surface of the tool 700 and the chip surface is slightly
There is a problem that bonding failure due to contact with 700 pieces frequently occurs. This tendency is remarkable as the size of the chip is increased and the number of pins is increased, so that a large amount of time is required for adjustment before the start of bonding, and furthermore, there is a problem that the trial yield is required and the bonding yield is reduced. Further, in the conventional bonding method, since bonding is performed with a hard chip, stage, and tool, a very large impact force is generated at the moment when the tool is hit, which may cause damage to the chip.

In the embodiment shown in FIG. 42, an elastic body 72 such as Teflon, silicone rubber or polyimide is placed on the stage 600.
By attaching 0, a slight shift in parallelism is absorbed, and the impact force is effectively absorbed.

In the embodiment, for simplicity, on a ceramic stage,
The experiment was conducted with a structure in which a 125 μm polyimide tape was attached. With this, the adjustment that conventionally took about 3 hours,
The same bonding yield was achieved in only 30 minutes, confirming the effect of the present embodiment. 42A shows a state before bonding, and FIG. 42B shows a state at the time of bonding.

The conventional stage structure does not basically have a structure for absorbing a deviation in the parallelism between the bonding surface of the tool 700 and the chip 4, and is limited to a structure having a mechanical spring structure. This alone is not sufficient for the function of absorbing the deviation in parallelism, and also insufficient for absorbing the impact load during bonding.

Therefore, in the embodiment shown in FIG. 43, attention is paid to the viscous behavior of a fluid such as air or oil, and the above problem is solved. That is, as shown in FIG.
Below the 600, a damper 730 using a cylinder structure was attached.

As a result, a slight parallelism shift (-2 μm / chip surface), which poses a practical problem, is absorbed, and the same bonding yield can be obtained even if the conventional parallelism adjustment time is reduced from 3 hours to about 30 minutes. I understood. It was also found that the chip damage under the same conditions can be reduced to less than 1/10 of the conventional one.

〔The invention's effect〕

According to the present invention, highly accurate alignment can be performed,
Accuracy at the time of alignment is maintained during bonding, and the yield in the inner bonding step can be improved. In addition, even if there is an error in the height of the bump, an excessive pressure or a shocking pressure is not applied to the lead or the bump, and an effect that bonding failure such as chip breakage, lead breakage, and lead peeling does not occur. is there. further,
Efficient thermocompression bonding can be performed by, for example, making the temperature distribution of the tool uniform.

[Brief description of the drawings]

FIG. 1 is an overall configuration diagram of an embodiment of the present invention, and FIG.
FIG. 3 is an overall view of the apparatus of the present invention, FIG. 4 is a perspective view of an operating part, FIG. 5 is a perspective view of a tape and spacer winding and tensioner part, Sixth
FIG. 7 is an enlarged perspective view of a bonding portion, FIG. 7 is a cross-sectional view of a chip stage, FIG. 8 is a perspective view of a head portion, and FIG.
FIG. 10 is a diagram showing an alignment state, FIG. 10 is a perspective view showing a configuration around a bonding position and a configuration of an alignment detection optical system, FIG. 11 is a functional block diagram showing an entire configuration of the alignment system, and FIG. Is an explanatory diagram of a method of calculating a position correction amount in the XYθ direction, FIGS. 13 to 15 are explanatory diagrams of definitions of a lead position and a chip position in a visual field, FIG. 16 is an explanatory diagram of a lead position detecting method, 17 is an explanatory diagram of a chip corner position detecting method, FIGS. 18 and 19 are explanatory diagrams of a bump position detecting method, FIG. 20 is a block diagram of an alignment operation flow,
FIG. 21 is an explanatory diagram of a method for checking a shift amount, and FIG.
Illustration of field average alignment, FIG. 23 is an illustration of a method of returning from one field detection state, FIG. 24 is an illustration of an alignment mark on a chip stage, and FIG. 25 is bonding after lead / bump proximity 26 is a diagram showing a vertical movement mechanism of a bonding tool, FIG. 27 is a diagram showing a main part of FIG. 26, and FIG. 28 is a circuit block diagram for performing drive control of the tool. FIG. 29 and FIG. 30 are operation flowcharts of the tool vertical movement drive mechanism, FIG. 31 is a diagram schematically showing the movement of the tool, and FIG. 32 is a control operation method of each mode in FIG. The block diagram shown in FIG. 33 is a diagram for explaining the condition setting of the tool vertical movement mechanism, and FIG.
FIG. 35 is a flowchart showing an application example of the pressurization control, and FIG.
FIG. 34 shows the behavior of the tool and the bump in FIG. 34, FIG. 36 to FIG. 38 show the shape for heating the bonding tool, FIG. 39 and FIG. 40 show the heating and cooling of the chip stage FIG. 41 is a view showing a shape of a tool on which a diamond thin film is formed, FIG. 42 is a view showing a stage structure with an elastic body attached thereto, and FIG. 43 is a view showing a damper attached FIG. 4 is a diagram showing a stage structure in which the stage is placed. 1 ... tape, 2 ... inner lead, 3 ... bump, 4
…… IC chip, 5,5a, 5b …… Field of view, 6 …… Chip stage, 7 …… Bonding tool, 8 …… Light source for epi-illumination, 9 …… Mirror, 10 …… Aperture, 11 …… Shutter, 12 …
... mirror, 13 ... half prism, 14 ... objective lens,
16: Light source for oblique illumination, 17: Rotary solenoid, 18
...... Shutter, 19 ... Glass fiber, 20 ... Ring illumination device, 21 ... Variable aperture, 22 ... Half prism, 23
...... Mirror, 24 ... Field lens, 25 ... Relay lens, 26a, 26b, 27a, 27b, 28a, 28b ... Mirror, 29a, 29b ...
… TV camera, 30a, 30b …… Detected image of TV camera, 31…
Image processing unit, 32: Lens, 33: Mirror, 34: Zoom lens, 35: TV camera, 36: XYθ stage, 37:
... Enlarged view of bonding position, 38a, 38b ... Detection range of enlarged image, 39 ... Bonding position, 40 ... A / D converter, 4
1 ... multi-valued memory, 42 ... binarization circuit, 43 ... binary memory, 44 ... projection processing circuit, 45 ... microcomputer, 46 ... mechanism control unit, 47 ... detection optical system, 48 ... …switch.

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Toshimitsu Hamada 292 Yoshida-cho, Totsuka-ku, Yokohama-shi, Kanagawa Prefecture Inside the Hitachi, Ltd.Production Technology Laboratory (72) Inventor Naofumi Iwata 292 Yoshida-cho, Totsuka-ku, Yokohama-shi, Kanagawa Hitachi, Ltd. Production Technology Laboratory Co., Ltd. (72) Inventor Aizo Kaneda 292, Yoshida-cho, Totsuka-ku, Yokohama-shi, Kanagawa Prefecture Inside of Hitachi, Ltd. Production Technology Laboratory Co., Ltd. Address: Hitachi, Ltd., Production Technology Research Laboratories (72) Inventor Hiroyuki Tanaka 292, Yoshida-cho, Totsuka-ku, Yokohama-shi, Kanagawa Prefecture Hitachi, Ltd. Production Technology Research Laboratories (72) Koichi Sugimoto: Yoshida-cho, Totsuka-ku, Yokohama-shi, Kanagawa Prefecture 292 Hitachi Manufacturing Co., Ltd.Production Technology Laboratory (72) Inventor Toshihiko Sakai Kana 292 Yoshida-cho, Totsuka-ku, Yokohama-shi, Kawasaki Prefecture Inside Hitachi, Ltd.Production Technology Research Laboratories (72) Inventor Keizo Matsukawa 1450 Kamimizu Honcho, Kodaira-shi, Tokyo Inside Hitachi, Ltd.Musashi Factory (72) Inventor Riki Mimi 1450, Kamimizuhoncho, Kodaira-shi, Tokyo Inside Musashi Plant of Hitachi, Ltd. (56) References JP-A-57-160135 (JP, A) JP-A-61-39530 (JP, A) JP-A-58-139 (JP, A) JP-A-61-27645 (JP, A) JP-A-57-90955 (JP, A) JP-A-57-36840 (JP, A)

Claims (5)

(57) [Claims] An IC chip having bumps formed on a surface thereof and an inner lead formed on a carrier tape facing each other at a bonding station, wherein a position of the IC chip on a stage is detected by a camera at the bonding station; Calculate the position correction amount of the lead and the
A method of manufacturing a semiconductor device in which an IC chip is aligned and bonded, wherein one or more visual fields to be detected are set at corners of the IC chip, and the position of the inner lead or the IC chip is detected within the visual field. Calculating the correction amount in the XYθ direction from the position of the inner lead or the position of the IC chip in the field of view, the position of the visual field at one or more positions, and the rotation center position of the XYθ stage on which the IC chip is mounted, A method for manufacturing a semiconductor device, comprising: aligning and bonding the IC chip. 2. A position of an inner lead formed on a carrier tape is detected by a camera at a bonding station, and an IC chip having bumps formed on a surface thereof is opposed to the inner lead at the bonding station. A method of manufacturing a semiconductor device in which the lead and the IC chip are aligned and bonded, wherein one or more fields of view to be detected at corners of the IC chip are set, and the position of the inner lead or the IC chip is set to the field of view. And calculates the correction amount in the XYθ direction from the position of the inner lead or the position of the IC chip in the visual field, the visual field position at one or more locations, and the rotation center position of the XYθ stage on which the IC chip is mounted. Manufacturing the semiconductor device, wherein the leads and the IC chip are aligned and bonded. Construction method. 3. The method for manufacturing a semiconductor device according to claim 1, wherein after the alignment of the inner lead and the IC chip, a relative positional relationship between the inner lead and the IC chip is fixed. A method for manufacturing a semiconductor device, comprising: bonding the inner lead and the IC chip with a bonding tool. 4. The method for manufacturing a semiconductor device according to claim 1, wherein the oblique illumination orthogonal to both surfaces of the IC chip and the carrier tape is performed, and the IC chip is obtained from an image detected in this illumination state. A method for manufacturing a semiconductor device, comprising detecting a position. 5. A method for manufacturing a semiconductor device according to claim 2, wherein oblique illumination is performed obliquely on both surfaces of the IC chip and the carrier tape, and a position of the inner lead is obtained from an image detected in this illumination state. A method for manufacturing a semiconductor device, comprising: