JP2767955B2 - Semiconductor chip mounting method - Google Patents

Description

Description: FIELD OF THE INVENTION The present invention relates to a method for mounting a semiconductor chip having an electrode portion such as a dot, and more particularly, to correcting a displacement of a semiconductor chip which is bonded to a substrate in a horizontal line. Means for doing so.

(Prior art) For example, the reading and writing units of a facsimile
It is manufactured by bonding a plurality of semiconductor chips in which dot-shaped electrode portions such as D and LED are arranged side by side at a pitch to a substrate by a die bonding device in a horizontal line.

Electrodes such as dots are placed at a predetermined pitch in XYθ
It is necessary to align them correctly without any positional deviation in the direction. If there is a positional deviation, a reading error or a writing error occurs. However, the positions of the electrode portions of the semiconductor chip are likely to be out of order due to bonding errors, external errors of the semiconductor chip, and the like.

Further, the semiconductor chip bonded in a horizontal line as described above is subsequently connected to the electrode portion of the substrate by a wire by a wire bonder. However, if there is an angular error in the suction surface of the die collet, for example, a large positional error occurs during the bonding of a large number of semiconductor chips, and the connection using the wires becomes impossible.

Therefore, conventionally, after bonding a semiconductor chip to a substrate using a die bonding apparatus, an operator observes dots with a microscope, and if there is a misalignment, moves the semiconductor chip with a pin or the like to correct it.

(Problems to be Solved by the Invention) However, since the above-mentioned conventional means performs correction manually by an operator, work efficiency and accuracy are not improved.

Accordingly, an object of the present invention is to provide a means capable of automatically and accurately correcting misalignment of a semiconductor chip bonded in a horizontal line to a substrate with good workability.

(Means for Solving the Problems) For this purpose, according to the present invention, the semiconductor chip sucked by the die collet of the bonding head is positioned by the XY direction moving device by reciprocating the bonding head in the lateral direction. In the method of mounting a semiconductor chip to be bonded to a substrate, a plurality of semiconductor chips are sequentially bonded to the substrate by the die collet of the bonding head as described above, and then, of the semiconductor chips thus bonded, By observing two semiconductor chips separated from each other by a camera and detecting an angle with respect to a reference line of a line connecting the two semiconductor chips, an angular error in the θ direction is detected. The angle error is corrected by rotating the The chips are sequentially bonded to the substrate.

(Operation) According to the above configuration, two semiconductor chips separated from each other out of a plurality of bonded semiconductor chips;
Preferably, the semiconductor chip at the start and end portions is observed by a camera, and the angle between the line connecting the two semiconductor chips with respect to the reference line is detected, thereby detecting an angular error in the θ direction.

Next, the die collet is rotated by θ relative to the substrate to correct the angular error, and the semiconductor chips are sequentially bonded to the substrate.

Example Next, an example of the present invention will be described with reference to the drawings.

FIG. 1 is an overall front view of a die bonding apparatus, in which a chip supply unit 1, a substage 2, an XY-direction moving device 3 for positioning a substrate 11, and a driving device BM such as a basic machine. The bond supply unit 4 is provided side by side. The chip supply unit 1 includes a wafer 5 and a die ejector 12 for pushing up a semiconductor chip P on the wafer 5.
However, the bond supply unit 4 is provided with a bond plate 6. 7, 8, 9 are sub-transfer heads mounted on the drive unit BM,
A bonding head and a bond application head. These three heads 7 to 9 reciprocate in the horizontal direction (Y direction) along the guide device 10, so that the sub-transfer head 7 transfers the chips P on the wafer 5 to the sub-stage 2, and Bonding head 8 is XYθ at substage 2
The chip P whose positional deviation in the direction has been roughly corrected is transferred and mounted on the substrate 11 on the XY-direction moving device 3 and the bond application head 9
Applies the bond of the bond plate 6 to the substrate 11.

22 is a 3 provided at the lower end of the bonding head 8
The semiconductor chip P is attracted to the dicollet 22. The bonding head 8 is provided with a motor 25 and gears 26 and 27 for rotating the die collet 22 by θ (horizontal rotation).

A driving unit 13 drives the die ejector 12 via a rod 19, and reciprocates the heads 7 to 9 in a lateral direction via a rod 14, a swinging member 15, a horizontal rod 16, and the like. The XY direction moving device 3 is composed of XY tables 17 and 18, and moves the substrate 11 mounted thereon in the XY direction.

FIG. 2 shows that the semiconductor chip P is attracted to the three-way die collet 22 at the lower end of the bonding head 8 and the substrate 11 is sucked.
This shows how it is implemented in Reference numerals 20 and 21 denote first and second cameras provided above the XY direction moving device 3 for observing the substrate 11 and the chip P, respectively. In FIG. 1, cameras 20 and 21 are omitted because the figure becomes complicated.

The bonding head 8 reciprocates in the lateral direction Y, and transfers the semiconductor chip P, which has been corrected for the positional deviation in the XYθ direction in the substage 2, to the substrate 11 on the XY direction moving device 3 for bonding. The substrate 11 includes heads 7 to 9
Is transported in the direction X perpendicular to the reciprocating direction Y of the XY-direction moving device 3, and is carried into and out of the XY-direction moving device 3. A,
B is a position reference mark formed at a corner of the substrate 11.

This means detects the displacement of the chip P in the XY direction and corrects it, and then detects the displacement in the θ direction and corrects it. Next, referring to FIG. An example of detection of a pitch shift and its correction method will be described.

Prior to mounting the chip P on the substrate 11, first, the first camera 20 observes the reference marks A and B, detects the displacement of the substrate 11, and drives the XY direction moving device 3 to The displacement is corrected (FIG. 3A).

Next, the bonding head 8 reciprocates between the sub-stage 2 and the XY-direction moving device 3, so that the first chip P1 on the sub-stage 2 is attracted to the three-way die collet 22 to be picked up, and the substrate 11 is picked up. (FIG. 2B). In this case, the first chip P1
Is mounted using the reference mark A observed as described above as a reference point. The three-way die collet is a die collet that restricts movement of the semiconductor chip adsorbed thereto in three directions and allows only one direction of movement. In FIG. 2, it is adsorbed by the die collet 22. The chip P can move only in the left end direction.

Next, in the same manner, the second chip P2 is
And is bonded to the substrate 11 (FIG. 3 (c)).
Next, the bonding head 8 waits at the raised position, and drives the XY-direction moving device 3 in that state, and the chip P1 and the chip P2 are moved.
Is moved below the second camera 21, and the pitch of the dots S1 and S2 as the electrode portions of the adjacent chips P1 and P2 is observed. FIG. 4D is a plan view during observation, and the pitch between adjacent dots S1 and S2 is t + Δt, and the predetermined pitch t
It is much larger by Δt.

Next, the chip P2 is moved below the die collet 22 again, and the chip P2 is pressed down by the die collet 22,
The substrate 11 is moved by a distance Δt in the X1 direction (FIG. 4E). Then, the first chip P1 approaches the chip P2 by the same distance Δt, and this error Δt is corrected. In order to advantageously correct the error Δt, it is desirable to set the movement programming of the XY-direction moving device 3 so that the error Δt occurs on the plus side. This is because, if an error occurs on the minus side, when bonding the chip P2 to the substrate 11, there is a possibility that the chip P2 may not be able to be bonded to the previous chip P1. Note that the error Δt
When the correction is made, the bond 23 (see FIG. 3E) for bonding the chip P2 to the substrate 11 is not cured.

After mounting chip P2 in this way,
The third chip P3 is bonded onto the substrate 11 (FIG. 2 (f)), and the pitch of the adjacent dots S3 and S4 of this chip P3 and the preceding chip P2 (see FIG. 2) is similarly determined. Make corrections. Such work is performed on the following chips P4, P5 ...
For each of the chips P1 to Pn
Are mounted side by side on the substrate 11 with the pitch of the dots S aligned.

By the way, recognition systems such as the cameras 20 and 21 have a fixed error. Further, the driving means such as the XY direction moving device 3 and the driving unit 13 also have a certain error due to backlash, thermal expansion and the like. Therefore, as described above, even if the pitches of the adjacent dots S1, S2 and S3, S4... Are individually and accurately corrected each time, due to these constant errors, each dot S is as follows. It has a displacement.

That is, as shown in FIG. 4, the dot S1 of the first chip P1 is Δx1,
Δy1 and the second chip P2
Dot S3 has a displacement of Δx2, Δy2 from the ideal position S3 ′. This displacement is caused by each chip P1, P2, P3 ...
And the dot Sm of the n-th chip Pn is Δxn, Δyn from the ideal position Sm ′.
Has a cumulative error of Therefore, the constant error Δx,
Even if Δy is very small, as many chips P are mounted side by side, the accumulated errors Δxn and Δyn greatly exceed the allowable error.
Due to the wire bonder, each dot S and each electrode portion of the substrate 11 cannot be connected by a wire. Next, an example of a method of detecting such a constant error and correcting the same will be described with reference to FIG.

When mounting the chips P1, P2,... On the substrate 11, the dots S1, S3, S5 at the ends of the chips P1, P2,. ..Accumulated error (Δx,
Δy), (Δx2, Δy2), (Δx3, Δy3)... Then, it is detected whether or not the accumulated error does not exceed an allowable error (for example, 20 μ). If the accumulated error does or does not exceed the allowable error, the driving of the XY direction moving device 3 is performed so as to eliminate the error. Is controlled, and each chip P is mounted while correcting the position of the substrate 11.

FIG. 5 is a graph showing an example of this.
If it is confirmed that the accumulated error of the fifth chip P5 has approached the allowable error, when mounting the sixth chip P6, the XY direction moving device 3 is corrected by adding a correction in the direction to eliminate the accumulated error. Is driven to move the substrate 11. Similarly, when mounting the fourteenth chip P14, the correction in the reverse direction is performed. This correction may be made in any manner, for example, when the third chip P3 is mounted (that is, when the third chip P3 is still sufficiently within an allowable error), the correction may be made.
In short, it is sufficient to add a correction so that each dot does not cause a large displacement due to a fixed error. Although FIG. 5 shows the error Δx in the X direction, the same applies to the error Δy in the Y direction.

In this way, the present means observes the positional deviation of each dot S due to the fixed error by the camera 21 and corrects it based on the observation result. Absent.

By the way, for example, the angle error θ
In the case where there is an error, an angle error occurs in the chip P sucked by this, and the die P1 to Pn
Are sequentially mounted as described above, as shown in FIG. 6, the bonding lines R of the chips P1 to Pn are displaced from the reference line N, which is a correct bonding line, by the angle θ. In this case, the displacement of the chip P1 at the start end is small, but since this error is gradually accumulated as a fixed error, the chip Pn at the end is largely displaced.
When the displacement is increased in this way, it is impossible to connect the chip and the electrode portion a of the substrate 11 by wires, which is subsequently performed by a wire bonder. Therefore, next, means for detecting the above-described angle error θ and correcting the angle error θ will be described.

First, the reference marks A and B are observed by the cameras 20 and 21, and the reference line N is recognized from the line M connecting the marks A and B. Next, by detecting a line R connecting the center of the chip P1 at the start end and the center of the chip Pn at the end, and detecting the angle θ between the line R and the line N, the angular error of the inner surface of the die collet 22 is detected. Detect θ. The centers of chips P1 and Pn are
For example, these dots S can be easily detected by observing them. The angle θ can be obtained by detecting a line R connecting two chips separated from each other and detecting an angle between the line R and the reference line N. Therefore, the two chips can be arbitrarily selected. In this case, if the two chips are separated as much as possible, the angle θ can be accurately detected.
It is desirable to detect a line R connecting the chip P1 at the start end and the chip Pn at the end.

When the angular error θ is detected as described above, the motor 25 is driven to rotate the die collet 22 in the θ direction, thereby correcting the angular error θ. And the next board
From 11, the chips P 1 to Pn can be bonded in a horizontal line along the reference line N by bonding the chips with the dicollet 22 whose angle has been corrected.

Needless to say, it is only necessary to correct the angular error θ of the die collet 22 relative to the substrate 11, and therefore, by rotating the substrate 11 in the θ direction with respect to the die collet 22, the angular error θ is corrected. You may make it.

As described above, the present method firstly mounts the first chip on the substrate 11.
The angle error θ is detected by performing bonding of P1 to Pn, and the chips P1 to Pn are bonded from the next substrate 11 by the die collet 22 in which the angle error θ is corrected.

Although the above embodiment has been described with reference to an example in which the dot P is formed on the chip P, a chip in which an electrode portion of a circuit pattern is formed instead of the dot S can be similarly applied.

(Effects of the Invention) As described above, according to the present invention, it is possible to correct misalignment of a semiconductor chip bonded to a substrate and bond a large number of chips in a horizontal line correctly along a reference line. .

[Brief description of the drawings]

1 shows an embodiment of the present invention. FIG. 1 is a front view of a die bonding apparatus, FIG. 2 is a perspective view during bonding, and FIGS. 3 (a) to 3 (f) are portions during bonding. FIG. 4 is a front view of a semiconductor chip during bonding, FIG. 5 is a graph, and FIG. 6 is a plan view of a substrate. 3 XY direction moving device 8 Bonding head 11 Substrate 20, 21 Camera 22 Die collet P Semiconductor chip

Claims (2)

(57) [Claims] 1. A semiconductor chip mounting method in which a semiconductor chip adsorbed by a die collet of the bonding head is bonded to a substrate positioned by an XY direction moving device by reciprocating a bonding head in a lateral direction. In the method, after a plurality of semiconductor chips are sequentially bonded to the substrate by the die collet of the bonding head as described above, two semiconductor chips separated from each other among the semiconductor chips bonded in this manner are separated by a camera. By observing, by detecting the angle of the line connecting the two semiconductor chips with respect to the reference line, the angle error in the θ direction is detected. Then, the die collet is rotated θ relative to the substrate, and After correcting the angle error, make sure to bond the semiconductor chip to the substrate sequentially. A semiconductor chip mounting method which is characterized in that the. 2. The mounting of a semiconductor chip in which a semiconductor chip adsorbed by a die collet of the bonding head is bonded to a substrate positioned by an XY direction moving device by reciprocating a bonding head in a lateral direction. In the method, after bonding the semiconductor chip to the substrate by the die collet of the bonding head as described above, the XY direction of the electrode portion of the semiconductor chip bonded to the substrate and the electrode portion of the semiconductor chip bonded to the substrate before that Then, the semiconductor chip is pressed by the die collet and the XY direction moving device is driven to correct the XY direction error. After bonding to the substrate, the semiconductor chip thus bonded Observing two semiconductor chips apart from each other with a camera and detecting an angle of a line connecting the two semiconductor chips with respect to a reference line, an angular error in the θ direction is detected. A semiconductor chip is sequentially bonded to the substrate after correcting the angular error by rotating the semiconductor chip relative to the substrate by θ.