JPH0845970A - Die bonding device - Google Patents

Abstract

PURPOSE:To enable a semiconductor chip to be accurately mounted on a stem at a right position always. CONSTITUTION:The picture of a collet 1 is picked up by a TV camera 2 from below before the collet 1 sucks a semiconductor chip 19, a relative deviation of the picture image pattern picked up by the camera 2 from a previously stored reference pattern is calculated through a recognition section 7, the amount of deviation is outputted from the recognition section 7 to a control section 8 to actuate pulse motors 10a and 10b or a pulse motor 11 to correct the collet 1 in original position, and then the collet 1 sucks up the semiconductor chip 19 and transfers it by a prescribed distance of X-Y, and thus the chip 19 is bonded.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a die bonding apparatus for adsorbing and holding a semiconductor chip and adhering and fixing the semiconductor chip to a mounted member such as a lead frame or a stem.

[0002]

2. Description of the Related Art Conventionally, in a die bonding apparatus, a semiconductor chip transferred from an adhesive sheet or tray to a stage is vacuum-sucked by a collet, and a lead frame or a stem, which is a mounted member, is sucked by the collet. The semiconductor chip is transferred, and then the collet is lowered to press and attach the semiconductor chip to the surface of the mounted member.

Further, when the die, which is the semiconductor chip, is attached to the member to be mounted, it has to be always attached to the member to be mounted at a fixed position. Particularly, in a semiconductor device having severe high frequency characteristics, a high accuracy of die bonding position of ± 20 μm to ± 25 μm is required.

In order to perform such precise positioning, for example, an image of the sucked semiconductor chip is picked up from below by a television camera, and an external image pattern of the picked-up semiconductor chip and a reference pattern when the collet is sucked accurately Comparing with the calculated amount of displacement, the displacement is corrected to the relative distance between the original position of the collet and the mounted member, and after moving the collet, the semiconductor chip is transferred and positioned on the mounted member, or , The collet is made of a light-transmitting material, the pattern of the surface of the adsorbed semiconductor chip is imaged by the upper television camera, and the amount of deviation is calculated by comparing with the previously stored reference pattern of the semiconductor chip, As described above, the movement distance is corrected by the amount of displacement and positioning is performed.

[0005]

FIGS. 5A and 5B are views showing a collet and a semiconductor chip for explaining the problems of the conventional die bonding apparatus. In the conventional position recognition method for determining the position by the above-mentioned outer shape image pattern, it is difficult to manufacture by matching the center position of the collet with the center position of the surface pattern of the semiconductor chip,
Even if it is manufactured with high accuracy, the positioning accuracy is at most ± 50 μm. Therefore, further positioning accuracy cannot be expected.

In the latter method of recognizing the position of the semiconductor chip adsorbed and held by the collet on the surface pattern, as shown by the chain double-dashed line in FIG. Although there is no problem as long as it is adsorbed on the collet 1, it is often adsorbed while being inclined to the tapered surface 1a of the collet 1, as shown by the solid line in FIG. In this case, when the collet 1 is lowered and the semiconductor chip 19 is pressed against the mounted member 20 to return the inside of the exhaust hole 1b to the atmosphere, the semiconductor chip 19 is moved to the center of the collet 1 by the self-centering action by the pressing force of the tapered surface 1a. Imitate.

That is, when the semiconductor chip 19 tilted by θ degrees and adsorbed by the collet 1 is placed on the mounted member 20 and released from vacuum, the semiconductor chip 19 becomes sin θ × semiconductor chip due to the self-centering action of the collet 1. It will be attached with a deviation of the plate thickness t. For example, if the thickness t is 0.3 mm, 5
It becomes 26 μm just by tilting it, which exceeds the allowable value.

SUMMARY OF THE INVENTION Therefore, an object of the present invention is to provide a die bonding apparatus that can always attach a semiconductor chip to a mounted member at a more accurate position.

[0009]

A feature of the present invention is that in a die bonding apparatus for adsorbing and holding a semiconductor chip and adhering and fixing the semiconductor chip to a mounted member, a taper surface and an exhaust hole are formed inward from an end edge. A collet for holding and holding the semiconductor chip on the taper surface and holding the collet to move in the XY directions to position the collet on the mounted member and lower the collet to the mounted member. A moving mechanism for mounting the semiconductor chip, a television camera for imaging the end portion of the collet,
The relative movement amount until the pattern image of the end imaged by the television camera matches the previously stored reference pattern is calculated, and the displacement amount between the reference pattern of the collet and the image pattern is calculated. And a controller for correcting the amount of movement of the collet by the moving mechanism by the amount of displacement before the semiconductor chip is sucked. Further, it is desirable to include an illumination optical mechanism that projects illumination light coaxially with the light entering the television camera. Furthermore, if necessary, a rotation mechanism for rotating the collet around the axis is provided.

Another feature of the present invention is that the permissible limit value for clogging of the exhaust hole of the collet or the judgment value of the presence or absence of a flaw or chip on the tapered surface of the collet is stored in advance in the recognition section, and the television is stored. A die bonding apparatus that issues an alarm by determining whether an image pattern at the end portion of the collet input to a camera is within the allowable limit value for clogging of the exhaust hole and whether the taper surface is flawed or chipped. .

[0011]

Next, the present invention will be described with reference to the drawings.

1 (a) and 1 (b) are a plan view and a side view of a main part showing an embodiment of a die bonding apparatus according to the present invention. This die bonding device is shown in FIG.
As shown in, the tapered surface 1 formed inward from the end edge
a and an exhaust hole 1b, and a collet 1 for adsorbing and holding the semiconductor chip 19 on the taper surface 1a and an arm 3 for holding the collet 1 and moving in the XY directions to position the collet 1 on the stem 17. Collet 1
Is moved down to mount the semiconductor chip 19 on the stem 17, an XY drive mechanism 5 for moving the head 6 in the XY directions, a television camera 2 for capturing an image of the end 16 of the collet 1, and the television camera 2. A recognizing unit 7 that calculates a relative movement amount of the captured pattern image of the end portion 16 until it coincides with a reference pattern stored in advance and calculates a shift amount between the pattern image of the collet 1 and the reference pattern.
And a controller 8 that corrects the movement amount of the collet 1 by the XY drive mechanism 3 by the calculated shift amount before the semiconductor chip 19 is sucked.

XY for correcting the position of the collet 1
In addition to the drive mechanism 3, it is desirable to provide a pulse motor 11 capable of step feed for rotating the collet 1 around the axis. Although this will be described later, it is easier to correct the amount of deviation depending on the amount and direction of deviation rather than correcting the amount of deviation by dividing it into the XY component and by correcting the component in either the X or Y direction and the rotation angle component. The advantage is that

The television camera 2 includes a light source lamp 1
There is provided an illumination optical mechanism for reflecting the light from the mirror 3 on the half mirror 14 and projecting the reflected light to the end 16 of the collet 1 to transmit the light to the television camera 2 through the half mirror 14. By providing this illumination optical mechanism, it is possible to take an image with a sharp contrast even at the end 16 of the dark collet 1 without causing a shadow of the convex portion, and to obtain a sufficient amount of reflected light even if dust in the exhaust hole 1b described later is blocked. I am doing it.

In the figure, 9 is an intermediate stage,
This is a stage for temporarily placing the semiconductor chip 19 from the tray. The stage 4 is a stage on which the stem 17, which is a mounted member, is placed.

FIGS. 2A and 2B are diagrams showing image patterns at the ends of the rectangular collet and the round collet. In the case of the image shown in FIG. 2A, the image entering the television camera 2 is a case of a rectangular collet, and the images are taken at two positions on the diagonal of the position recognition units 15a and 15b. Further, as shown in FIG. 2B, in the case of the round collet, the position recognition 15c is the whole including the end of the collet 1.

As described above, the reference which has been processed into a black and white binarized image by the recognition unit 7 according to the intensity of the reflected light from the end 16 of the collet 1 and which has been input in advance into the memory circuit of the recognition unit 7 is used. The amount of deviation of the center position of the image pattern, which is compared with the pattern and received, with respect to the reference pattern is calculated.

FIG. 3 is a diagram showing an imaging area and slice lines for explaining a method of calculating the shift amount. In the case of the round collet, for example, the above-described calculation method, as shown in FIG. 3, first slices the entire region of the end portion 16 with a certain number of lines in the X and Y directions, and slices each slice line. Find the black and white pixel boundary points at. Then, the coordinates of the boundary points of the black and white pixels are averaged over all the slice lines to obtain the coordinates of the center of gravity of the image of the end portion 16. Next, the shift amount between the center of gravity of the reference pattern input in advance and the center of gravity of the captured image is calculated.

The amount of deviation calculated in this way is sent to the control unit 8 to move the head 6 including the arm 2 in the XY direction.
The pulse motors 10a and 10b of the drive mechanism 3 have a shift amount of X
And the correction pulse corresponding to the distance divided into the Y component is input, the head 6 is moved by the XY drive mechanism 3, and the origin position of the collet for determining the position to be sucked by the semiconductor chip 19 and the position to be die-bonded is determined. .

If there is a large deviation in either the X direction or the Y direction but a small deviation in the other directions,
Higher accuracy can be ensured by correcting the pulse motor 11 after the correction in the X direction or the Y direction. Since the collet rotary shaft has a lower load than the pulse motors 10a and 10b for XY drive, the pulse motor 11 can be directly connected to the collet rotary shaft and the pulse motor 11 can be directly connected to the collet rotary shaft without interposing a gear having a backlash. This is because it can be directly connected to the rotating shaft.

FIGS. 4A and 4B are views showing the state of the end portion of the collet for explaining another embodiment of the die bonding apparatus of the present invention. In addition, this die bonding apparatus has an advantage that the abnormality of the end portion of the collet can be detected in advance by including the television camera 2 that observes the end portion of the collet 1.

For example, as shown in FIG. 4 (a), when there is a chip or flaw 21 on the taper surface 1a of the collet that makes it impossible to adsorb to the semiconductor chip, the outer diameter of the taper surface 1a is shown. It is stored in the storage unit of the recognition unit 7 that the existence of black dots or continuous black dots is not possible in the ring-shaped region between the boundary point and the boundary point indicating the inner diameter, and the collet 1
The image pattern is scanned with a slice line to determine whether there are black spots or continuous black spots in the ring-shaped area, and if there is, an alarm is issued. If a new collet is exchanged in advance when an alarm is issued, the collet will not drop or hold uncertainly during the transfer of semiconductor chips.

Further, as shown in FIG. 4 (b), the exhaust hole 1
When the clogging 22 is generated in b and the suction force is weakened, if the boundary point coordinates of the limit line 23 indicating the minimum hole are input in the storage unit of the recognition unit 7 in advance, the clogging of the exhaust hole 1b of the collet 22
If the boundary point indicated by is inside the limit line 23, an alarm is issued.

The sequence control of the apparatus is designed so that the inspection of the end of the collet is performed before the above-mentioned position recognition, and every time one lot of semiconductor chips is mounted or a predetermined number of semiconductor chips are mounted. Good to do.

[0025]

As described above, according to the present invention, the relative position of the television camera that captures the bottom surface of the collet is compared with the image pattern captured by the television camera and the previously stored reference pattern. By recognizing the displacement and correcting the original position of the collet with the calculated displacement, even if the collet is not accurately attracted to the collet or the collet with inadequate manufacturing accuracy, the precise position of the mounted member is always maintained. There is an effect that a semiconductor chip can be attached to.

[Brief description of drawings]

FIG. 1 is a plan view and a side view of a main part showing an embodiment of a die bonding apparatus according to the present invention.

FIG. 2 is a diagram showing image patterns at ends of a rectangular collet and a round collet.

FIG. 3 is a diagram showing an imaging area and slice lines for explaining a method of calculating a shift amount.

FIG. 4 is a diagram showing a state of an end portion of a collet for explaining another embodiment of the die bonding apparatus of the present invention.

FIG. 5 is a diagram showing a collet and a semiconductor chip for explaining the problems of the conventional die bonding apparatus.

[Explanation of symbols]

 1 Collet 1a Tapered surface 1b Exhaust hole 2 Television camera 3 Arm 4 Stage 5 XY drive mechanism 6 Head 7 Recognition part 8 Control part 9 Intermediate stage 10a, 10b, 11 Pulse motor 12 Lens 13 Light source lamp 14 Half mirror 15a, 15b, 15c Position recognition part 16 End part 17 Stem 19 Semiconductor chip 20 Mounted member 21 Flaw 22 That is 23 Limit line

Claims (4)

[Claims] 1. A die bonding apparatus for adsorbing and holding a semiconductor chip and adhering and fixing the semiconductor chip to a member to be mounted, wherein a taper surface and an exhaust hole are provided inside from an end edge and the semiconductor chip is formed on the taper surface. A collet that attracts and holds the collet; and a moving mechanism that holds the collet and moves in the XY directions to position the collet on the mounted member and lower the collet to mount the semiconductor chip on the mounted member. , A relative amount of movement of the television camera that captures an image of the end of the collet and a pattern image of the edge captured by the television camera until the reference pattern is stored in advance, and the relative movement amount of the collet is calculated. A recognition unit that calculates the amount of positional deviation between the reference pattern and the image pattern; and A die bonding apparatus, comprising: a controller that corrects the movement amount of the dot with the shift amount. 2. The die bonding apparatus according to claim 1, further comprising an illumination optical mechanism that projects illumination light coaxially with light entering the television camera. 3. The die bonding apparatus according to claim 1, further comprising a rotating mechanism that rotates the collet around an axis. 4. An allowable limit value for clogging of the exhaust hole of the collet or a judgment value for the presence or absence of a flaw or chip in the tapered surface of the collet is stored in the recognition unit according to claim 1 in advance, and the television camera is stored. A die for issuing an alarm by determining whether or not the exhaust hole is within the allowable limit value for clogging of the exhaust hole and whether or not there is a flaw or chip in the tapered surface from the image pattern of the end portion of the collet input to the die. Bonding equipment.