JP2820526B2 - Positioning method and apparatus for flip chip bonding - Google Patents

Description

The present invention relates to a method and an apparatus for aligning flip-chip bonding. SUMMARY OF THE INVENTION Means for Solving the Problems and Actions to Solve the Problems Embodiments Effects of the Invention With the object of providing a method and apparatus for performing flip chip bonding of an optical semiconductor chip with high precision, a first camera for observing a pattern of a substrate and a second camera for observing a bump of a chip are provided, and a reference scale is used. Calculating the absolute magnification of the first and second cameras, obtaining the offset distance between the center points of the first camera and the second camera on the XY plane, displaying an image captured by the first camera on a monitor, From this image, the absolute position coordinates of the pattern on the substrate are read based on the absolute magnification of the first camera, and the monitor is switched to shoot with the second camera The displayed image is displayed on the monitor, the absolute position coordinates of the bump of the chip are read from the image based on the absolute magnification of the second camera, and the bonding head that has absorbed the chip is moved to the offset distance and the pattern and bump. It is configured to move by the sum of the shift amounts between the absolute position coordinates.

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates generally to an alignment method and apparatus for flip-chip bonding, and more particularly, to a method for manufacturing an ultra-high-speed optical / electronic combined module in an optical communication system, the method includes the steps of: The present invention relates to a method and an apparatus for individually monitoring a pattern of a circuit board and performing alignment of flip chip bonding using absolute position coordinates of a chip side and a substrate side.

In recent years, the information transmission speed required for a communication system tends to increase according to the demand for long-distance communication. Especially,
An optical communication system is required to have a multi-gigabit-class transmission speed, and there is a demand for the development of an optical communication device that realizes this transmission speed.

In order to construct such a light-speed optical transmission system,
A mounting method that does not impair the high speed of the used optical semiconductor chip is indispensable. One such mounting method is flip-chip bonding between the optical semiconductor chip and the circuit board, eliminating the parasitic reactance that would otherwise occur when wires were connected, and mounting without impairing the high-speed characteristics of the optical semiconductor chip. can do. In the case of such flip-chip bonding of an optical semiconductor chip, the bump diameter is required to be as small as about several tens of μm as compared with the case of flip-chip bonding of an electric circuit chip in order to reduce the parasitic reactance. ing.

Since a conventional flip chip bonder for an electric circuit chip cannot achieve an alignment accuracy of 10 μm or less, a flip chip bonding method and apparatus having high precision for performing flip chip bonding of an optical semiconductor chip is demanded. ing.

2. Description of the Related Art As described in Japanese Patent Publication No. 1-31296, for example, a conventional flip chip bonder device is configured to combine an image from a substrate pattern observation camera and an image from a chip bump observation camera into one unit. They are synthesized on a monitor and the amount of displacement between the substrate pattern and the chip bump is detected. However, in this flip chip bonder, since the image of the substrate pattern and the image of the chip bump are combined on one monitor, the image of the plurality of small-diameter bumps and the pattern cannot be clearly identified. There is a problem,
Further, in order to facilitate the identification, it is necessary to make improvements such as adding a color processing step for each image. This has a disadvantage that the entire apparatus becomes large and the cost increases.

Further, in the conventional positional deviation correction method described in the above-mentioned publication, first, the rotational deviation between the chip and the substrate is corrected, and then the amount of deviation in the X and Y directions that has occurred thereafter is detected, and the positional deviation is detected. The deviation is corrected. However, this displacement correction method requires two steps of correction calculation and control, and has a problem that errors in the calculation routine are accumulated and accuracy cannot be improved.

Further, as another method of correcting the displacement, Japanese Patent Application Laid-Open No. 57-3684
As described in No. 0, the misregistration correction amount is detected by moving and adjusting the points to be aligned. However, in this method, it is necessary to perform drive control in order to move and align points to be aligned. Also, JP-A-1-31296 and JP-A-57-3
In the displacement correction method described in Japanese Patent No. 6840, bumps having a finite area are treated as points, and there is a problem that high accuracy within the bump diameter cannot be obtained.

When performing flip chip bonding, it is necessary to pick up the chip with a suction nozzle. However, a conventional chip pick-up method uses a second method for observing the chip surface as described in Japanese Patent Publication No. 2-12024. 3
The chip mounting table is moved so that the corner of the chip surface observed by the camera is aligned with the reference line, and the chip is picked up. However, with this method,
When picking up a light receiving element in which a lens is formed on the chip surface, the lens is not always located at the center of the chip surface, so that the lens may be damaged by the tip of the nozzle for chip pickup.

Problems to be Solved by the Invention The conventional method and apparatus for positioning flip-chip bonding described in the above-mentioned publications and publications are applied to flip-chip bonding of electric circuit chips, and have very small bump diameters. Tens of μ
There is a problem that it is difficult to apply the method to flip chip bonding of an optical semiconductor chip requiring about m.

The present invention has been made in view of such a point,
An object of the present invention is to provide a method and apparatus for performing flip chip bonding of an optical semiconductor chip with high accuracy.

Means and Action for Solving the Problems First, the outline of the flip chip bonding alignment method of the present invention will be described.

(1) First, a first camera for observing a pattern on a substrate and a second camera for observing bumps on a chip are provided, and absolute magnifications of the first and second cameras are obtained using a reference scale.

(2) An x, y coordinate system with the center point of the monitor screen as the origin is set for the images from the first camera and the second camera, respectively. At this time, one monitor is used to switch the images from the first camera and the second camera so that the images are not shifted in the rotation direction between the coordinate systems of the first camera and the second camera. Monitor.

(3) Three bumps are selected from the chip bump image from the second camera. First, for the first bump, the bump is specified by a rectangle circumscribing the bump. The circumscribing rectangle can be obtained, for example, by designating two points using a cross cursor and drawing the rectangle. Next, the center point of this quadrangle is determined, and this is set as the center point coordinates of the bump.

(4) The center coordinates of the second and third bumps are obtained in the same manner.

(5) From the coordinates of the center points of the three bumps thus obtained, the coordinates of the center of gravity are obtained.

(6) An angle θ1 between a y-axis or an x-axis formed by a straight line connecting the coordinates of the center point of the first bump and the coordinates of the center of gravity obtained in step 5 is obtained.

(7) Next, from the pattern image of the substrate by the first camera, three patterns on the substrate corresponding to the three bumps selected in step 3 are selected.

(8) Similar to the chip side, the coordinates of the center point of each pattern are obtained, and the coordinates of the center of gravity are obtained.

(9) Similarly to the chip side, an angle θ2 between a straight line connecting the center point coordinates of the first point pattern and the barycentric point coordinates obtained in step 8 with the same axis as the axis specified on the chip side is calculated.

(10) Calculate the rotation correction amount of the substrate mounting stage and the moving amount of the bonding head from the second camera to the first camera from the coordinates of the center point, the coordinates of the center of gravity, and the angle obtained above.

(11) The substrate mounting stage is rotated and the bonding head is moved in accordance with the rotation correction amount of the substrate mounting stage and the moving amount of the bonding head obtained in this manner, whereby the bumps on the chip and the substrate are moved. Positioning with the above pattern can be performed accurately.

 Next, details of the above-described calculation process will be described.

"Setting of x, y coordinate system" As described above, x, y with the center point of the monitor screen as the origin
The y-coordinate system is set for each of the images from the first and second cameras. At this time, one unit is set so that the rotation direction between the coordinate systems of the first and second cameras does not shift. Using a monitor, the images from the first camera and the second camera are switched and each image is monitored on one monitor. The displacement that occurs in this case is only the offset amount between the first camera and the second camera on the xy plane.

In order to obtain the offset amount, first, the center of the suction nozzle is set to the origin of the coordinates at the time of image processing by the second camera, and then the stage of the bonding head is driven.
The tip shape of the nozzle is transferred by a suction nozzle of the head onto a monitor plate on which a trace of a pressure point, such as a masking tape, which is previously disposed on a substrate mounting table, remains. Next, the central point of the transferred nozzle selecting unit is used to determine the amount of driving movement of the bonding head required to reach the origin of the coordinates when monitoring the image by the first camera, thereby obtaining the offset amount in the x and y directions. Can be requested.

“About Bump Designation and Rotation Angle” First, in order to simplify the calculation process, the bump having the largest coordinate point on the y-axis is set as the first bump.
In this case, as shown in FIG. 1, the center point (x2, y2) of the first bump is connected to the center of gravity (x1, y1) obtained by averaging the center points of the three bumps. The angle θ1 (−90 ° to + 90 °) formed by the straight line and the y-axis can be obtained by the following equation.

y1 ≠ y2: θ1 = tan ⁻¹ ((x1−x2) / (y2−y1)) (1) y1 = y2: θ1 = 90 × ((x1−x2) / abs (x2−x1)) (( 2) Abs is an absolute value, and θ1 is positive in the counterclockwise direction and negative in the clockwise direction. Similarly, three patterns are sequentially specified on the substrate side so as to correspond to the bumps of the chip. A straight line connecting the center point (x4, y4) of the first pattern on the substrate side and the center of gravity (x3, y3) obtained by averaging the center points of the three patterns is
The angle θ2 (−90 ° to + 90 °) formed with the y-axis can be obtained by the following equation.

y3 ≠ y4: θ2 = tan ⁻¹ ((x3−x4) / (y4−y3)) (3) y3 = y4: θ2 = 90 × ((x3−x4) / abs (x3−x4)) (( 4) From θ1 and θ2 obtained above, the required amount of rotation of the rotary stage of the substrate mounting table is as follows: Δθ = θ1−θ2 (5)

“About the correction amount in the xy direction” Next, the correction amount in the x and y directions will be described with reference to FIG. C1 and C2 are the centers of gravity on the chip side and the substrate side, respectively. Or is the rotation center point of the rotary stage of the substrate mounting table, and O is the center point of the monitor screen. As described above, the coordinates have the origin at the center point of the monitor screen. C2 ′
Shows the C2 point after being rotated around Or by Δθ obtained by the equation (5). Distance T between Or and C2, and Or to C2
The angle θ _0, which saw, T = ((x3-xr ) 2 + (y3-yr) 2) 1/2 ... (6) x3 ≠ xr: θ 0 = tan -1 ((y3-yr) / (x3 −xr)) (7) x3 = xy, y3> yr: θ ₀ = 90 (8) x3 = xy, y3 <yr: θ ₀ = −90 (9) Accordingly, the coordinate of the point C2 'is x3' = T × cos (Δθ + θ ₀ ) + xr (10) y3 ′ = T × sin (Δθ + θ ₀ ) + yr (11), and the correction amount in the x and y directions is Δx = X3'-x1 (12) Δy = y3'-y1 (13) From the above, assuming that the offset amounts are xf and yf, the moving amount of the bonding head is xm = Δx + xf (14) ym = Δy + yf (15)

Actually, after performing the above calculation, the rotation of the rotation and the movement of the bonding head are performed simultaneously or continuously, so that the bumps of the chip and the patterns of the substrate can be accurately bonded.

"The origin of the image from the first camera is θ
In the above method, it is difficult to accurately grasp the amount of displacement between the rotation center point Or of the rotation stage of the substrate mounting table and the center point O of the monitor screen. Yes, even if the value is obtained, the total calculation error may be accumulated by this value. Therefore, it is possible to eliminate the accumulated error by previously adjusting the displacement between the rotation center point Or of the rotation stage and the center point O of the monitor screen to zero. At this time, the moving amount of the bonding head (xm, y
m) is given by xm = x3′−x1 + xf (16) ym = y3′−y1 + yf (17) x3 ′ = T × cos (Δθ + θ ₀ ) (18) y3 ′ = _{T × sin (Δθ + θ 0} ) ... (19) T = (x3 2 + y3 2) 1/2 ... (20) θ 0 = tan -1 (y3 / x3) [3 ≠ xr] ... (21) θ ₀ = 90 [X3 = xr, y3> yr] (22) θ ₀ = −90 [x3 = xr, y3> yr] (23)

In addition, in order to make the displacement between the rotation center point Or of the rotating stage and the center point O of the monitor screen zero in advance, a mark is placed on the rotation center point of the rotating stage, and this mark matches the center point of the monitor screen. Adjust the X and Y stages provided in advance at the bottom of the rotary stage, or
This is possible by adjusting the X and Y stages provided on the fixed stage of the camera.

The flip chip bonding apparatus of the present invention that achieves the above-described alignment method is configured as follows.

That is, a substrate mounting table configured to be movable in the X, Y, θ directions and capable of being heated, a chip mounting table configured to be movable in the X, Y, θ directions, the chip mounting table and the substrate mounting A bonding head having chip suction means movable between the table and a first camera for observing a pattern of a substrate mounted on the substrate mounting table, and observing a bump of a chip sucked by the chip suction means; And a single monitor for displaying images of the first and second cameras.

Switching means for switching between images of the first and second cameras and displaying the images on the monitor;
Means for calculating the offset distance between the center points on the XY plane of the camera, reading the absolute position coordinates of the pattern on the substrate and the absolute position coordinates of the bumps on the chip, and calculating the deviation between the absolute position coordinates Means.

Embodiments Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

FIG. 3 shows a schematic configuration diagram of an embodiment of the present invention. Ten
Denotes a substrate mounting table, which includes an X stage and a Y stage that are manually operated, and a θ stage that is automatically operated by a pulse, whereby the substrate mounting table 10 is configured to be movable in the X, Y, and θ directions. I have. A heat column 18 for heating the substrate at the time of bonding to melt the solder bumps is provided on the X stage 12, and a substrate 20 having a predetermined pattern is mounted on the heat column 18. The substrate 20 mounted on the substrate mounting table 10 is above the substrate mounting table 10.
A first camera 22 for observing the pattern (1) is provided.
The first camera 22 is mounted so as to be adjustable in the X, Y, and Z axis directions.

Reference numeral 24 denotes a chip mounting table, which comprises an X stage 26, a Y stage 28, and a θ stage 30, each of which is automatically operated by a pulse, and on which an optical semiconductor chip 32 is mounted. Reference numeral 34 denotes a bonding head, and a suction nozzle 36 for suctioning the chip 32 is provided at the tip of the bonding head. 38 is the weight of the bonding head 34. The bonding head 34 is attached to the tip of an arm 40, and the base end of the arm 40 is mounted on an X stage 42, a Y stage 44, and a Z stage 46, which are automatically operated by pulses.

Reference numeral 48 denotes a second camera for observing the bumps of the chip 32 sucked by the suction nozzle 36, and a bonding head
It is provided below 34. 50 is the chip mounting table 24
This is a third camera for observing the surface of the chip 32 mounted thereon, and is provided above the chip mounting base 24. The second camera 48 and the third camera 50 are mounted so as to be adjustable in the X, Y, and Z axis directions.

Each stage for moving the substrate mounting table 10, the chip mounting table 24, and the bonding head 34 described above is controlled by a stage controller 52. And the first camera
22, the images of the second camera 48 and the third camera 50 are displayed on one CRT 54 by switching the switching means. The process of recognizing images captured by the first camera 22, the second camera 48, and the third camera 50 and performing correction calculation and the like is as follows.
It is executed by the CPU 56.

Hereinafter, a method of operating the flip chip bonding apparatus thus configured will be schematically described.

First, the absolute magnifications of the first camera 22 and the second camera 48 are obtained using the reference scale, and the offset distance between the center points of the first camera 22 and the second camera 48 on the XY plane is determined by the method described above. Ask.

The optical semiconductor chip 32 mounted on the chip mounting table 24 is driven by driving the stages 42, 44, 46 of the bonding head 34 so that the bonding head 34 is mounted on the chip mounting table.
Then, the bonding head 34 is moved to the position 24 and sucked by the suction head 36 as shown in the drawing, and the bonding head 34 is retracted to a position immediately above the second camera 48. The optical semiconductor chip 32 is observed by the second camera 48, and three bumps are selected. First, for the first bump, the bump is specified by a rectangle circumscribing the bump.
Next, the center point of this quadrangle is determined and used as the center point coordinates of the bump. Similarly, the center point coordinates are obtained for the second and third bumps. From the center point coordinates of the three bumps thus obtained, the barycentric point coordinates are obtained. Then, an angle θ1 formed by a straight line connecting the coordinates of the center point of the first bump and the coordinates of the center of gravity with the y-axis or the x-axis is obtained.

Next, the pattern of the substrate 20 is photographed by the first camera 22, and this image is displayed on the CRT. From this image, three patterns corresponding to the bumps of the optical semiconductor chip 32 are selected. Similar to the chip side, the coordinates of the center point of each pattern are obtained, and the coordinates of the center of gravity are obtained. Further, an angle θ2 between a straight line connecting the coordinates of the center point of the first point and the coordinates of the center of gravity and the same axis as the axis designated on the chip side is determined.

From the coordinates of the center point, the coordinates of the center of gravity, and the angles θ1 and θ2 obtained in this manner, the rotation correction amount of the substrate
The amount of movement of the bonding head 36 from the position directly above the camera 48 to the position directly below the first camera 22 is calculated. According to the rotation correction amount and the moving amount of the bonding head obtained in this way, the substrate mounting table 10 is rotated and simultaneously the bonding head 34 is driven to drive the bumps of the optical semiconductor chip 32 and the substrate.
The flip-chip bonding can be performed while the substrate 20 is heated by the heat column 18 by accurately matching the pattern of 20.

By the way, since the bumps of the optical semiconductor chip 32 are formed by a plating process, there are many surface sagging and chippings. Therefore, determining the shape only from an image from a camera and obtaining the center of the bump has an error. As a result, there has been a problem that the alignment accuracy at the time of flip chip bonding is reduced. In order to solve such a problem, according to the present invention, as shown in FIG. 4, a drawing 60 of a bump is input in advance by an image reader or other means, and collation with an image 62 from the second camera 48 is performed. By calculating the center of the bump by correcting the surface sagging and chipping,
The error can be reduced. Also, by performing image correction on the substrate pattern side by the same means, the center of the substrate pattern can be obtained without error.

Next, a chip pickup method will be described with reference to FIG. The conventional chip pickup method is, for example, as described in Japanese Patent Publication No. 2-12024, so as to match a line with a corner of the chip surface observed by a third camera for observing the chip surface as a reference. The chip is picked up by moving the chip mount. However, in this method, when picking up a light receiving element having a lens formed on the chip surface, the lens is not always located at the center of the chip surface, and the lens is damaged by the tip of the suction nozzle for chip pickup. There was fear.

In order to solve such problems, optical semiconductor chips
A third camera 50 for observing the surface of the chip 32 is disposed above the chip mounting table 24, the position of the micro lens 66 formed on the chip surface is observed, and an image from the third camera 50 is displayed on the CRT 54.
Project on top. Then, the nozzle bore 68a of the suction nozzle 68
Chip mounting base 24 so that
Are moved in the X, Y, and θ directions, or the bonding head 34 is moved in the X, Y directions as shown in FIG.
Is moved in the θ direction, the suction nozzle 68 can be positioned directly above the microlens 66 as shown in FIG. 5B, and the microlens 66 is damaged by the tip of the suction nozzle 68. The chip 32 can be picked up without giving.

In connection with this, a preferred embodiment of the tip structure of the chip suction nozzle will be described. FIG. 6 shows a suction nozzle having a conventional shape equivalent to the suction nozzle 36 shown in FIG. 5, and a tip portion 36b is formed in a substantially planar shape.
In the suction nozzle 36 having this structure, when a very thin chip having a small area such as the optical semiconductor chip 32 is suctioned,
In order to reliably suction the chip 32 without creating a gap between the chip surface and the flat end surface 36b of the suction nozzle 36, the flat surface 36b requires high surface accuracy and levelness.

In order to solve such a problem, as shown in FIG.
By forming a spherical shape having a center point on the center axis of 0a, the contact portion between the chip surface and the nozzle 70 can be reduced from the surface to the circumference, without requiring high surface accuracy and levelness. Thus, the chip 32 can be reliably sucked. When the light receiving element chip 32 having the microlenses 66 formed on the surface is to be sucked, the nozzle bore of the suction nozzle 70 is used.
By making the inner diameter of 70a slightly larger than the diameter of the microlens 66, the microlens 66 can be adsorbed without being damaged at the time of adsorption. Such a tip shape 70b of the suction nozzle 70 can be created by mechanical polishing.

Next, a scrubbing mechanism of the substrate mounting table 10 during bonding will be described with reference to FIG. In the related art, a means such as a honeycomb mechanism is provided in a bonding head to enable a scrubbing operation. However, in this method, the weight of the bonding head becomes heavy,
The impact load on the chip cannot be reduced. Furthermore, in order to improve the positioning accuracy, it is necessary to surely return the suction nozzle to the center point of the camera after scrubbing.

In order to solve such a problem, a vibration mechanism based on the piezoelectric element 72 is provided on the substrate mounting table 10 as shown in FIG. 8, thereby preventing the weight of the bonding head 34 from increasing.
The scrubbing operation with the bumps 33 made of the amorphous solder agent becomes possible. Further, when the piezoelectric element 72 is used, the driving voltage to the piezoelectric element is reduced to zero, so that the suction nozzle 68 after scrubbing can be reliably returned to the center point of the first camera 22, and before and after scrubbing. Can be prevented from being displaced.

Next, the automatic focusing mechanism of the camera will be described with reference to FIG. When a camera with a fixed focus is used, and when chips or substrates having different thicknesses are handled, the recognition accuracy at the time of correcting a blurred image position is reduced. In order to solve this, it is necessary to adjust the Z axis of the camera using the Z stage of the X, Y, Z stages to which the camera is attached each time, but this adjustment is complicated and the Z axis adjustment is required. If so, the center of the camera is displaced in the X and Y directions due to yawing of the Z stage, and it becomes impossible to perform positioning with an accuracy of 10 μm or less.

In order to solve such a problem, the camera 22 itself is provided with an automatic focusing mechanism 76 as shown in FIG. 9, thereby not only making the Z-axis adjustment of the camera unnecessary, but also preventing the camera center from shifting. , With an accuracy of less than 10 μm. 74 is a support member for supporting the first camera 22. Such an autofocus mechanism is not limited to the first camera 22, but also includes the second camera 48 and the third camera.
It is desirable to provide in 50 as well.

Next, the contact sensor mounted on the bonding head 34 will be described. Since optical semiconductor chips are generally thinner than electric circuit chips, even a slight impact load at the time of chip suction and bonding causes cracks. In order to prevent this, the descent speed of the bonding head 34 needs to be extremely low. However, if the bonding head is lowered at an extremely low speed from the origin position of the Z axis, one operation time until a chip is picked up and bonded becomes longer. Would. In order to solve this, in the present embodiment, a contact sensor capable of detecting the movement of the Z axis is added to the bonding head 34, and a dummy chip slightly thicker than a chip to be handled in advance is used, and the suction nozzle 36 contacts the dummy chip. , That is, the position (Z1) in the Z direction is detected in advance.

When picking up chips, the bonding head 34 is lowered at the normal speed up to this Z1 point, and after exceeding this Z position, the speed can be switched to a very low speed to prevent cracks during chip suction become.

Also, at the time of bonding, the dummy chip is used to detect the height at which the dummy chip contacts the substrate. Then, the bonding head 34 is lowered at a normal speed at the predetermined height, and the speed is switched to a very low speed after the predetermined height is exceeded, thereby making it possible to prevent cracks during bonding.

Further, assuming that the difference between the thickness including the bump height of the dummy chip and the chip actually connected is ΔZ, and the height of the bump connecting the chip and the substrate after melting is H, the final Z-axis of the bonding head is obtained. By setting the position Z2 as Z2 = Z1 + ΔZ + H, it is possible to arbitrarily control the height of the bump connecting the chip and the substrate after melting.

In order to achieve an alignment accuracy of 10 μm or less, it is necessary to minimize distortion of the entire apparatus due to heat generated from the heat column 18 and the power supply section of the main body. Therefore, by disposing a heat shield plate made of a material having low thermal conductivity, for example, Kovar or ceramic, so as to surround the heat column 18 and the power supply portion of the main body, it is possible to prevent distortion of the device caused by thermal distortion. Become. Furthermore, this structure also makes it possible to shorten the time required to reach a thermally stable state. Further, X, Y, θ arranged immediately below the heat column 18
The same effect can be obtained by forming the constituent members such as the stages 12, 14, 16 and the camera support member 74 with a material having low thermal conductivity.

Although the above-described embodiment has been particularly described with respect to flip chip bonding of an optical semiconductor chip, it is needless to say that the present invention is applicable not only to optical semiconductor chips but also to flip chip bonding of ordinary electric circuit chips.

Advantageous Effects of the Invention Since the present invention is configured as described above in detail, it is possible to achieve high-accuracy positioning of 10 μm or less, particularly in flip chip bonding of an optical semiconductor chip.

[Brief description of the drawings]

FIG. 1 is an explanatory diagram for explaining how to obtain θ1, θ2, FIG. 2 is an explanatory diagram for explaining how to obtain a correction amount in the x and y directions, and FIG. 3 is a flip chip bonding according to an embodiment of the present invention. FIG. 4 is a schematic diagram of the apparatus, FIG. 4 is an explanatory diagram for explaining image comparison between a bump drawing and an actual bump, FIG. 5 is an explanatory diagram of a chip pickup method, FIG. FIG. 7, FIG. 7 is a sectional view showing a nozzle tip structure according to a preferred embodiment of the present invention, FIG. 8 is a schematic configuration diagram of a scrubbing mechanism on the substrate stage side, and FIG. 10 ... board mounting table, 18 ... heat column, 20 ... board, 22 ... 1st camera, 24 ... chip mounting table, 32 ... optical semiconductor chip, 34 ... bonding head, 36 ... for suction Nozzle, 48: Second camera, 50: Third camera, 52: Stage controller, 54: CRT, 70: Suction nozzle, 72: Piezoelectric element, 76: Auto focus mechanism

Continuation of the front page (56) References JP-A-64-42829 (JP, A) JP-A-2-114928 (JP, U) (58) Fields investigated (Int. Cl. ⁶ , DB name) H01L 21 / 60 311

Claims (8)

(57) [Claims] A first camera (22) for observing a pattern on a substrate (20) and a second camera (48) for observing bumps on a chip (32) are provided, and the first and second cameras are used by using a reference scale. Camera (22,4
8), the offset distance between the center point of the first camera (22) and the second camera (48) on the XY plane is determined, and the image taken by the first camera (22) is monitored ( 54), the absolute position coordinates of the pattern on the substrate (20) are read from the image based on the absolute magnification of the first camera (22), and the monitor (54) is switched to switch the second camera (48). ) Is displayed on the monitor (54), and the absolute position coordinates of the bumps of the chip (32) are read from the image based on the absolute magnification of the second camera (48). Bonding head (34)
Is moved by the sum of the offset distance and the shift amount between the absolute position coordinates of the pattern and the bump. 2. In the chip image projected on the monitor (54) by the second camera (48), the coordinates of the center point of each bump are obtained from the outer frame of the plurality of bumps, and the coordinates of the center point are obtained from the coordinates of each center point. Find the center of gravity of the bumps on the chip (3
And 2) selecting a pattern on the substrate corresponding to the plurality of bumps in the substrate image projected on the monitor (54) by the first camera (22), The center coordinates of each pattern are obtained from the outer frames of the plurality of patterns obtained, and the center of gravity of the plurality of patterns is obtained from each center point coordinate to obtain absolute position coordinates on the substrate (20) side. The amount of movement ΔX, ΔY, and Δθ for completely overlapping the chip (32) and the substrate (20) is calculated from the coordinate shift amount, and the bonding head (34) sucking the chip (32) is moved to the offset distance X. By moving the component + ΔX, the Y component of the offset distance + ΔY, and the substrate mounting table (10) by Δθ, the position of the bump of the chip (32) and the pattern of the substrate (20) during flip chip bonding are aligned. Alignment method for flip chip bonding of claim 1, wherein the door. 3. The flip-chip bonding device according to claim 2, wherein the first camera is arranged such that the origin of the image obtained by the first camera coincides with the center of rotation of the substrate mounting table. Alignment method. 4. The positions and shapes of bumps and patterns are input to a computer in advance, and the input image is compared with images from the first and second cameras (22, 48) to determine the shapes of bumps and patterns. 4. The method according to claim 1, wherein the positional relationship is recognized and corrected. 5. A substrate mounting table (10) movable in X, Y, and θ directions and capable of being heated, and a chip mounting table (24) configured to be movable in X, Y, and θ directions.
A bonding head (34) having chip suction means (36) movable between the chip mounting table (24) and the substrate mounting table (10), and mounted on the substrate mounting table (10). A first camera (22) for observing the pattern of the picked-up substrate (20), a second camera (48) for observing bumps of the chip (32) sucked by the chip sucking means (36), Monitor means (54) for displaying images of the second and second cameras (22, 48); and switching means for switching the images of the first and second cameras (22, 48) and displaying the images on the monitor means (54). Means for obtaining an offset distance between the center points of the first camera (22) and the second camera (48) on the XY plane; absolute position coordinates of a pattern on the substrate (20); ) The absolute position coordinates of the upper bump are read, and the deviation between the absolute position coordinates is calculated. Flip-chip bonding system, characterized in that it consists of a means for. 6. The flip chip bonding apparatus according to claim 5, wherein a vibration means (72) is provided on said substrate mounting table (10). 7. A bonding head for picking up a chip is mounted on an actual chip (32) at the tip of the bonding head (34).
A contact sensor for detecting a first height at which the dummy chip contacts the thicker dummy chip and a second height at which the dummy chip contacts the substrate (20) during bonding; 6. The flip chip according to claim 5, further comprising control means for switching the driving of the bonding head to a low speed when the first height and the second height are detected during suction and bonding. Bonding equipment. 8. The chip suction means (36) of the bonding head (34) is constituted by a chip suction nozzle (70) whose tip (70b) has a convex spherical shape having a center point on the center axis of the nozzle bore. A flip chip bonding apparatus characterized in that: