JP2658405B2 - Method for manufacturing semiconductor device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Summary] The present invention relates to a method of manufacturing a semiconductor device, and particularly to a method of determining whether or not there is a defective mark on a chip prior to die bonding of the chip. The purpose of the present invention is to provide a method for reliably determining the presence or absence of a defective mark on a chip having dimensions and various patterns. [1] Place a full-cut wafer on a table and move the table to The center of the chip to be determined to determine the presence or absence of a chip is set below the camera, the image is captured by illuminating from below, the outer shape of the chip is recognized from the image, the center position is calculated from the outer shape, A method of manufacturing a semiconductor device in which an image is captured by illuminating the chip from above, and the presence or absence of a defective mark at the center position of the chip to be determined is detected by a mark detection unit; A wafer is placed on a table, and an image of one corner of a chip to be determined to determine the presence or absence of a defective mark is first captured by a camera.
Recognizing the position of the one corner end, then moving the table to capture the image of the diagonal corner at the diagonal of the one corner, recognizing the position of the diagonal corner end, and then recognizing the position of the one corner.
An image of the center of a line segment connecting the corner end and the diagonal corner end is fetched, and the presence or absence of a defective mark at the center is detected and determined by a mark detection unit to manufacture a semiconductor device.

[Industrial applications]

The present invention relates to a method for manufacturing a semiconductor device, and more particularly to a method for determining whether or not there is a defect mark on the upper surface of a chip before die bonding the chip.

When a good chip is selected and die-bonded from a full-cut wafer including a defective chip to which a defective mark (usually an ink mark) is applied, an area including one chip of the full-cut wafer set on the XY table is selected. An image is taken with a CCD camera, the outer shape of the chip is recognized by image processing, the position data of the center of the chip is calculated from the position data of the outer edge of the chip, and then the presence or absence of a defective mark at the center is detected and determined. A non-marked good chip is selected and die-bonded.

At this time, it is necessary to reliably determine the presence or absence of a defective mark on chips having various patterns and various dimensions.

[Conventional technology]

A conventional method of recognizing a defective mark on a chip of a full-cut wafer and a subsequent die bonding method will be described.

The schematic diagram of the defective mark recognition device II shown in FIG. 6 is a schematic diagram of a conventionally used defective mark recognition device, in which 1 is a full cut wafer, 3 is an XY table, and 31 is an XY table (fixed portion). ), 32 is a table, 41 is an upper light source, 5 is a CCD camera, 51 is a lens, 6 is an image processing unit, 61 is an arithmetic processing unit, 62 is a mark detection unit, 7 is an XY table driving unit, and 8 is a control. Represents a part.

FIG. 7 shows a full-cut wafer with a defective mark, 1 is a full-cut wafer, 2 is a chip, and 21 is a defective mark.

The defective mark is preliminarily tested for each chip in a wafer state by a measurement prober, and is applied to the center of the upper surface of the defective chip in a substantially circular shape with ink.
It is desirable that the defective mark be as small as possible so as not to contaminate surrounding chips and to reduce ink consumption, as long as the mark can be confirmed.

The full-cut wafer 1 has a UV tape (the adhesive strength decreases when exposed to ultraviolet light) adhered to the back surface, the scribe line is cut in the middle of the UV tape, and each chip is scribed with a width of about 0.1 mm. They are separated by lines.

First, the full-cut wafer 1 is placed on the XY table 3,
The XY table 3 is moved to a position where an image of a chip for which the presence or absence of a defective mark is to be determined is captured. Upper light source 41
Irradiates light on the upper surface of the full-cut wafer 1, and an image including the entire chip is captured by the CCD camera 5. The image is decomposed into, for example, 255 × 255 pixels and taken in.

FIG. 8 schematically shows an image of a chip taken into the CCD camera 5, 2 indicates a chip to be determined, and 21 indicates a defective mark. The lens 51 has a magnification capable of capturing an image of one entire chip.

The image of the full-cut wafer 1 by the upper light source 41 includes the scribe line portion and the defect mark 21 at the center of the chip.
The part is dark because there is little reflected light, and the other part is bright because there is much reflected light. An electric signal corresponding to the brightness of the image is sent to the image processing unit 6. In the image processing section 6, a certain signal level is set in advance, and the image signal larger than the predetermined level is determined as white and the smaller image signal is determined as black, and the entire image is binarized. Set as a set.

Recognition of the outer shape of the determined chip 2 can be known by the image processing unit 6 from, for example, a boundary where a series of pixels in the X direction and the Y direction is inverted from white to black or from black to white.

The center of the chip having the defective mark 21 can be calculated by the arithmetic processing unit 61 as an intermediate point between the left end and the right end and an intermediate point between the upper end and the lower end of the outer shape of the chip 2 to be determined.

The black and white of the pixel at the center of the chip 2 to be determined is determined. If there is black, it is determined that there is a defective mark. If there is white, it is determined that there is no defective mark.

If it is determined that there is a defective mark, the control unit 8 sends XY
An instruction is issued to the table driving unit 7 to set the XY table 3 to 1
Move by the distance of the chip so that the next chip is below the lens 51. Then, the presence or absence of a defective mark is determined in the same manner as described above.

If it is determined that there is no defective mark, the chip is transported to a position where die bonding is to be performed.

Fig. 9 is a schematic diagram of a die bonder, 1 is a full cut wafer, 3 is an XY table, 32 is a table, 5 is a CCD camera, 9 is a chip transfer arm, 91 is a lead frame supply unit, and 92 is a lead frame. , 93 is a lead frame, 94 is an adhesive supply section, 95 is a bonding arm, 96 is an intermediate table, 97 is a wafer supply section, and 98 is a lead frame transport section.

The chip determined to have no defective mark is transferred to the intermediate table 96 by vacuum suction of the chip when the tip of the chip transfer arm 9 comes to that position. After the precise positioning on the intermediate table 96, the bonding arm
The wafer is carried by 95 to a lead frame 93, where die bonding is performed.

By the way, there are various patterns on the chip surface of the full-cut wafer 1, and there is a portion which does not reflect the light from the upper light source well and appears dark, and even the same type of chip has a different chip surface between manufacturing lots. There are chips that make it difficult to recognize scribe lines and defective marks due to different light reflection states.

FIG. 10 (a) schematically shows an example of a chip in which such a defective mark is difficult to recognize.

In addition to the scribe line and the black portion of the defective mark 21, a slightly black pattern exists as noise, which makes it difficult to set the binarization levels of white and black. That is, in order to completely determine these noises as white, it is necessary to set the determination level closer to the black side, and in this case, a part of the scribe line and the defective mark 21 may be mistaken as white. It is.

Furthermore, in recent years, the chip size has been increased due to the development of large-capacity memories and ASICs, and if the entire chip outline was to be displayed on the screen of the CCD camera 5, the magnification of the lens 51 would have to be reduced. The scribe line of the image captured by the camera 5 becomes thin, and the defect mark 21 also becomes small.

FIG. 10 (b) shows an example of a chip in which such a defective mark is difficult to recognize.

The scribe line of the image of the CCD camera 5 becomes thinner,
Depending on how the light is applied, the screen becomes extremely thin or cut off, making recognition difficult. The defective mark 21 also becomes smaller, and the image becomes thinner, thereby lowering the certainty of recognition.

[Problems to be solved by the invention]

Therefore, it is becoming difficult to reliably detect the presence or absence of a defective mark only by the conventional method of recognizing a defective mark.

The present invention provides [1] a method for reliably determining the presence or absence of a defective mark in a chip having black pattern noise on its surface,
And [2] It is an object of the present invention to provide a method for reliably determining the presence or absence of a defective mark in a large chip in which a scribe line is thin and a defective mark is small.

[Means for solving the problem]

The above-mentioned problems are as follows: [1] A defective mark in the center of the top surface of the chip
A full-cut wafer 1 including a plurality of chips separated by a scribe line and including a defective chip having 21 is placed on a mounting portion 33 through which light is transmitted, and a lower light source 42 provided below the mounting portion 33 is used. Illuminated, an image of a region including at least the chip 2 to be determined of the full-cut wafer 1 is captured by a camera 5 provided above the mounting portion 33, and the image is binarized by an image processing section 6, and Obtaining the position data of the outer edge of the chip 2, calculating the position data of the center of the chip 2 by the arithmetic processing unit 61 from the position data of the outer edge of the chip 2, Illuminating with an upper light source 41 provided above the placement unit 33, capturing an image of the chip 2 to be determined by the camera 5, and binarizing the image by the image processing unit 6 to obtain detection data; From the detection data, the calculated h And a step of detecting and determining the presence or absence of a defective mark in the position data of the center of the chip by the mark detection unit 62. [2] Including a defective chip having the defective mark 21 in the center of the upper surface of the chip A full-cut wafer 1 including a plurality of chips separated by a scrape line is placed on a table 3 and illuminated by an upper light source 41 provided above the table 3 so that one of the chips 2 to be determined
An image of the corner portion is captured by a camera 5 provided above the table 3, and the image is
Digitizing and obtaining the position data of the one corner end; and moving the table 3 relative to the camera 5 to
Capturing an image including a diagonal corner portion at a diagonal of the corner end into the camera 5, binarizing the image by the image processing unit 6, and obtaining position data of the diagonal corner end; Calculating the center of a line segment connecting the end position data and the position data of the diagonal corner end by the arithmetic processing unit 61 and defining it as the position data of the center of the chip 2 to be determined; The image of the position data portion at the center of the chip 2 to be determined is moved relative to the camera 5 and is taken into the camera 5, and the image is binarized by the image processing unit 6, and And a step of detecting and determining the presence or absence of a defective mark in the position data of the central portion of the chip 2 by the mark detection unit 62.

[Action]

[1] An upper light source 41 for illuminating the full cut wafer 1 mounted on the mounting portion 33 from above and a lower light source 42 for illuminating from below the full cut wafer 1 are provided.

First, the lower light source 42 is turned on. The portion of the scribe line becomes bright because light is transmitted upward, and the portion of the chip 2 to be determined becomes dark because the light is blocked. Therefore, it is easy to set a level for binarizing the scribe line portion with white and the chip 2 to be determined with black, and as the outer shape of the chip 2 to be determined, the left end, right end, upper end, and lower end are set to white. It is easy to recognize as a black border. Then, it is easy to calculate the center position of the chip 2 to be determined as the middle between the left end and the right end and the middle between the upper end and the lower end.

Next, only the upper light source 41 is turned on to determine the presence or absence of the defective mark 21. Since the center position of the defective mark 21 should already be known, if the defective mark 21 is present, the position should be the darkest. Even if there is a black noise pattern, it is not as black as the defective mark. By shifting the black-and-white determination level to the black side, a black noise pattern is classified into white, and the presence or absence of a defective mark can be reliably determined.

[2] When recognizing the presence or absence of a defective mark on a large chip, the magnification of the lens is increased to partially capture the chip image. By doing so, the scribe line becomes thicker,
Defective marks can be enlarged.

Since the size of the chip and the thickness of the scribe line are known in advance, it is possible to set so that one corner of a certain chip is captured in the image of the camera 5. In order to recognize one corner from the image, for example, the binarized pixel is scanned in the X direction and the Y direction, and one pixel is calculated from the X coordinate and the Y coordinate of the position where white to black or black to white is inverted. The position of the corner can be recognized and the position data of one corner end can be obtained.

Next, the table 3 is moved to bring the diagonal corner at one corner diagonally below the camera 5. Also in this case, since the chip size is known in advance, it is possible to move the table 3 and bring the diagonal corner below the camera 5 even if the full-cut wafer 1 has a slight bending or twist. It is. In order to recognize the diagonal corner from the image, for example, the binarized pixel is scanned in the X direction and the Y direction, and the X coordinate and the Y coordinate of the position where white to black or black to white is inverted are read. The position of the diagonal corner can be recognized, and the position data of the diagonal corner end can be obtained.

Assuming that the center position of the chip 2 to be determined is in the middle of a straight line connecting the position of one corner end and the position of the diagonal corner end,
It is easy to calculate the center position of the chip from the position coordinates of the corner end, the amount of movement of the table 3 and the position coordinates of the diagonal corner end.

Next, the table 3 is moved, the center of the chip 2 to be determined is brought under the camera 5, and an image is captured. If there is a defective mark in the center, the image is large, so that the presence or absence of the defective mark can be reliably determined.

〔Example〕

1 to 3 are diagrams for explaining the embodiment I.
This is an example of recognizing a defective mark of a chip having black pattern noise as shown in FIG.

FIG. 1 is a schematic diagram of a defect mark recognition device I, wherein 1 is a full cut wafer, 3 is an XY table, 31 is an XY table (fixed portion), 32 is a table, 33 is a mounting portion, and 41 is a mounting portion. Upper light source, 42 is a lower light source, 5 is a camera, a CCD camera, 51 is a lens, 6 is an image processing unit, 61 is an arithmetic processing unit, 62 is a mark detection unit, 7 is an XY table drive unit, and 8 is a control unit. Represents

 FIG. 2 shows a procedure I for determining the presence or absence of a defective mark.

FIG. 3 shows an image of the chip, FIG. 3 (a) shows an image of the chip by illumination of the lower part of the wafer, and FIG. 3 (b) shows an image of the chip by illumination of the upper part of the wafer.

 This will be described below with reference to FIGS. 1 to 3.

First, the full-cut wafer 1 is placed on the left and right ends of the chip Y
XY so that the top and bottom are parallel to the X direction.
The XY table 3 is mounted on the mounting portion 33 of the table 3, and the XY table 3 is moved to determine whether or not the defective mark 21 exists. Set 2.

With the lower light source 42 turned on and the upper light source 41 turned off, 1
An image of a region including the chip is captured by the CCD camera 5. Since light passes through the portion of the scribe line of the full-cut wafer 1 and the light is blocked at the portion of the chip to be determined, an image of the chip to be determined by the lower illumination of the wafer is as shown in FIG.

Images are captured by the CCD camera 5. The image is decomposed into pixels, and an electric signal corresponding to each pixel is sent to the image processing unit 6. A signal level for binarizing an image into white and black is set in the image processing unit 6 in advance, and when a signal higher than that level comes, white, and when a signal lower than that level comes, black is set. Judgment is performed to binarize the image into black and white.

Next, a series of pixels in the X direction is scanned, and a boundary at which white to black or black to white is inverted is recognized as a left end or a right end of the chip to be determined.

Next, a series of pixels in the Y direction is scanned, and a boundary where white is inverted to black or black to white is recognized as an upper end or a lower end of a chip.

Next, the position data of the left end and the right end are calculated by the arithmetic processing unit 61,
Then, position data at the center is calculated from the position data at the upper end and the lower end, and is determined as the center of the chip 2 to be determined.

Next, with the lower light source 42 turned off and the upper light source 41 turned on, an image of an area including one chip is captured by the CCD camera 5. FIG. 3 (b) schematically shows an image of a chip obtained by illuminating the upper part of the wafer, wherein 2 denotes a chip to be determined, and 21 denotes a defective mark.

The image captured by the CCD camera 5 is decomposed into pixels, and an electric signal corresponding to each pixel is sent to the image processing unit 6.
A signal level for binarizing an image into white and black is set in the image processing unit 6 in advance, and when a signal higher than that level comes, white, and when a signal lower than that level comes, black is set. Judgment is performed to binarize the image into black and white. At this time, even if there is a black pattern noise, the binarization level is set to the black side in order to determine it as white.

The mark detection unit 62 detects and determines whether there is black or white at the position of the chip center obtained in the previous stage. If there is black, it is determined that there is a defective mark, and if there is white, it is determined that there is no defective mark.

4 and 5 are diagrams for explaining the embodiment II.
This is an example of recognizing a defective mark of a large-sized chip as shown in FIG.

The defect mark recognition devices used are the defect mark recognition device shown in FIG. 1, the defect mark recognition device II shown in FIG.
Either may be used.

 FIG. 4 shows a procedure II for determining the presence or absence of a defective mark.

FIG. 5 shows an image of the chip, and FIG.
(B), (c), and (d) show an overall view, an image of one corner, an image of a diagonal corner, and an image of a chip center, respectively.

 Hereinafter, description will be made with reference to FIGS. 4 to 6.

As shown in the overall view of FIG. 5 (a), the image captured by the CCD camera 5 is an image of one corner, an image of a diagonal corner,
Shoot separately from the image centered on the chip.

First, the full-cut wafer 1 is placed on the left and right ends of the chip Y
XY so that the top and bottom are parallel to the X direction.
The chip 2 mounted on the table 3 and moving the XY table 3 to determine the presence or absence of the defective mark 21 is
The position is set so that the image of the one corner can be captured by the CCD camera 5.

Next, the upper light source 41 is turned on and the CCD camera 5 captures an image of one corner.

FIG. 5B shows an image of one corner captured by the CCD camera 5.

The image is decomposed into pixels, and an electric signal corresponding to each pixel is sent to the image processing unit 6. A signal level for binarizing an image into white and black is set in the image processing unit 6 in advance, and when a signal higher than that level comes, white, and when a signal lower than that level comes, black is set. Judge and change the image to black and white 2
Value.

Next, a series of pixels in the X direction is scanned, and a boundary where white is inverted to black or black to white is recognized as the left end of the chip 2 to be determined.

Next, a series of pixels in the Y direction is scanned, and a boundary where white is switched to black or black to white is recognized as the upper end of the chip 2 to be determined.

Next, the position of one corner end is recognized as the intersection of a line extending the left end in the Y direction and a line extending the upper end in the X direction.

Next, the XY table 3 is moved so that a diagonal corner on one diagonal of the chip 2 to be determined comes below the CCD camera 5. Since the dimensions of the chip 2 to be determined are known in advance, even if the full-cut wafer 1 has a slight bending or twist, the XY table 3 is moved to a position where the CCD camera 5 takes in the diagonal corner at least. It is possible to do.

Next, the diagonal corner is imaged by the CCD camera 5. FIG. 5C shows an image of a diagonal corner captured by the CCD camera 5.

The image is decomposed into pixels, and an electric signal corresponding to each pixel is sent to the image processing unit 6. A signal level for binarizing an image into white and black is set in the image processing unit 6 in advance, and when a signal higher than that level comes, white, and when a signal lower than that level comes, black is set. Judge and change the image to black and white 2
Value.

Next, a series of pixels in the X direction is scanned, and a boundary where white is inverted to black or black to white is recognized as the right end of the chip 2 to be determined.

Next, a series of pixels in the Y direction is scanned, and a boundary where white is inverted to black or black to white is recognized as a lower end of the chip 2 to be determined.

Next, the position of the diagonal corner end is recognized as the intersection of the line extending the right end in the Y direction and the line extending the lower end in the X direction.

Next, the position of the diagonal corner end, the position of the one corner end determined previously, and the X from the one corner to the diagonal corner
Based on the amount of movement of the Y table 3, the arithmetic processing unit 61 calculates a center point on a line connecting the one corner end and the diagonal corner end, and defines that as the center of the chip.

Next, X is set so that the center of the chip is just below the lens 51.
-Move the Y table 3. Then, an area including the center of the chip is imaged.

 FIG. 5 (d) shows an image at the center of the chip.

The image is decomposed into pixels, and an electric signal corresponding to each pixel is sent to the image processing unit 6. A signal level for binarizing an image into white and black is set in the image processing unit 6 in advance, and when a signal higher than that level comes, white, and when a signal lower than that level comes, black is set. Judgment is performed to binarize the image into black and white. Then, the mark detection unit 62 detects and determines whether there is black or white at the position of the chip center. If there is black, it is determined that there is a defective mark, and if there is white, it is determined that there is no defective mark.

〔The invention's effect〕

As described above, according to the present invention, a scribe line becomes thinner when a black and white pattern is present as noise on a chip and it is difficult to recognize a defective mark, and when one chip is taken into the screen at a time, The determination of the presence or absence of a defective mark can be reliably performed even for a chip to be determined in which the determination of the presence or absence of the defective mark becomes difficult due to the small size of the defective mark on the screen.

Therefore, when a good chip without a defective mark is selected from the full-cut wafer and the chip is die-bonded to the lead frame, the good chip is surely selected and the chip is die-bonded to the lead frame. be able to.

[Brief description of the drawings]

1 to 3 are views for explaining the embodiment I, FIG. 1 is a schematic diagram of a defective mark recognition device I, FIG. 2 is a procedure I for determining the presence or absence of a defective mark, FIG. 4 (b) are images of the chip, FIGS. 4 and 5 are diagrams for explaining the embodiment II, FIG. 4 is a procedure II for judging the presence or absence of a defective mark, and FIGS. 5 (a) to 5 (d). Is an image of a chip, FIG. 6 is a schematic diagram of a defect mark recognition device II, FIG. 7 is a full cut wafer having a defect mark, FIG. 8 is an image of a chip taken into a CCD camera, and FIG. FIGS. 10 (a) and 10 (b) are schematic diagrams showing an example of a chip in which a defective mark is difficult to recognize. In the drawing, 1 is a full-cut wafer, 2 is a chip, which is a chip to be determined, 21 is a defective mark, 3 is a table, an XY table, 31 is an XY table (fixed part), 32 is a table, 33 is a mounting section, 41 is an upper light source, 42 is a lower light source, 5 is a camera, a CCD camera, 51 is a lens, 6 is an image processing section, 61 is an arithmetic processing section, 62 is a mark detection section, and 7 is X. -Y table drive unit, 8 is a control unit, 9 is a chip transfer arm, 91 is a lead frame supply unit, 92 is a lead frame storage unit, 93 is a lead frame, 94 is an adhesive supply unit, and 95 is a bonding arm. Reference numeral 95 denotes an intermediate table, 97 denotes a wafer supply unit, and 98 denotes a lead frame transfer unit.

Claims (2)

(57) [Claims] 1. A mounting portion (1) including a defective chip having a defective mark (21) at the center of the upper surface of the chip, and a full cut wafer (1) including a plurality of chips separated by scribe lines. 33), illuminated by a lower light source (42) provided below the mounting portion (33), and the camera (5) provided above the mounting portion (33). Capturing an image of a region including at least the chip to be determined (2), binarizing the image by an image processing unit (6), and obtaining position data of the outer edge of the chip to be determined (2); A step of calculating position data of a central portion of the chip to be determined (2) from a position data of an outer edge of the determination chip (2) by an arithmetic processing section (61); Illuminated by the upper light source (41), and the image of the chip (2) is determined by the camera. 5) obtaining the detection data by binarizing the image by the image processing unit (6), and detecting the presence or absence of a defective mark in the position data of the center of the chip calculated from the detection data. And a step of detecting and determining in the section (62). 2. A full cut wafer (1) including a defective chip having a defective mark (21) at the center of the upper surface of the chip and including a plurality of chips separated by a scribe line is placed on a table (3). Illuminated by an upper light source (41) provided above (3), an image of one corner of the chip to be determined (2) is captured by a camera (5) provided above the table (3), and the image is taken. A step of obtaining the position data of the one corner end by binarizing by the image processing unit (6); and moving the table (3) relative to the camera (5) to diagonally diagonally form the one corner end. Capturing the image including the diagonal corner portion into the camera (5), binarizing the image by the image processing unit (6), and obtaining position data of the diagonal corner end, Position data and the diagonal corner A step of calculating the center of a line segment connecting the end position data by the arithmetic processing unit (61) and defining the center as the position data of the center of the chip to be determined (2); The image of the position data portion at the center of the chip to be determined (2) is taken into the camera (5) by moving relative to 5), and the image is binarized by the image processing unit (6). And the chip to be determined (2)
Detecting and determining the presence or absence of a defective mark in the position data at the center of the semiconductor device by the mark detection unit (62).