JP2647053B2 - Method for manufacturing semiconductor device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for manufacturing a semiconductor device, and more particularly to a method for judging the quality of a pellet in a die bonding step of a diced wafer having a mirror pellet.

[0002]

2. Description of the Related Art In a pre-processing step in which a circuit is formed on a wafer, there is a high possibility that a defective product is formed even when a circuit is formed in a peripheral portion of the wafer. Usually not done. Since such a pellet has a flat surface and is covered with an unpatterned Al film, the pellet almost totally reflects light. This type of pellet is commonly called a mirror pellet.

A mark (defective mark) is attached to such a mirror pellet in a pellet sorting step, and care is taken to prevent bonding in a die bonding step. FIG. 3 is a flowchart showing a conventional pellet recognition method in a die bonding step of a wafer having a mirror pellet, and FIG.
FIG. 4 is a diagram illustrating a relationship between a wafer sensing position and an optical image luminance when determining the presence or absence of a defective mark according to the flowchart of FIG. 3.

In the die bonding step, only good pellets are picked up from the diced wafer 5 shown in FIG. 4 and die bonding is performed. A mirror pellet 1 and a patterned pellet 2 are mixed in the wafer. Defective marks are given to the full-size mirror pellet (pellet B) and the defective pattern pellet (pellet C).

[0005] Here, the mirror pellet 1 is a metal film whose entire surface is flat and almost totally reflects when irradiated with light, and has a different reflectance from the patterned pellet 2. Therefore, when the defective mark 3 of the same size as the defective mark 3 attached to the patterned pellet 2 is attached to the mirror pellet 1, the detection size of the defective mark is different under the same illumination light amount, and the determination is made at the same mark size threshold value I. In this case, it is recognized that there is no defective mark, and is determined to be non-defective.

Therefore, in the pre-process of die bonding, the defective mark on the mirror pellet 1 is enlarged, and the defective mark is re-hit on the mirror pellet 1 so that the quality can be determined with the same illumination light amount and the same threshold value. Was. That is, the full-size mirror pellet 1 has a re-strike point failure mark 4 larger than the failure mark 3 attached to the patterned pellet 2.

As shown in FIG. 3, in step S201, the quantity of illumination light, the luminance threshold I, and the mark size threshold I are set. In step S202, the outer shape and position of the pellet are recognized based on the luminance threshold value I. If the outer shape of the pellet is defective, the process proceeds to step S20.
Going to step S5, if there is no external defect, step S203
Proceed to.

In step S203, the detected mark size is compared with the mark size threshold value I. If the detected mark size is larger, it is determined that there is a defective mark, and the flow advances to step S205 to determine the detected mark size. If it is smaller, it is determined that there is no defective mark, and step S
Proceed to 204. After the die bonding is performed in step S204, the process proceeds to step S205. In step S205, it is determined whether or not the work has been completed for all the pellets in the wafer. If the work has not been completed, the process skips to the next pellet in step S206. If completed, the process ends.

Now, assuming that the pellets A, B, C, and D are arranged as shown in FIG. 4 and the determination proceeds from left to right, the determination is first performed for the pellet A. In step S202, it is determined that the outer shape is defective, and the process returns to step S202 via steps S205 and S206, and proceeds to the determination on the pellet B.
In step S202, it is determined that the external shape is good, and
At 203, the presence or absence of a defective mark is determined. At this time, since the detected mark size is larger than the mark size threshold value I, it is determined that there is a defective mark, and step S2 is performed.
Go to 05.

Next, a determination on the pellet C is made. After the external shape is determined to be good in step S202, the presence or absence of a defective mark is determined in step S203. At this time, since the detected mark size is larger than the mark size threshold value I, it is determined that there is a defective mark, and the process proceeds to step S205. Next, the determination on the pellet D is performed. After it is determined in step S202 that the external shape is non-defective, it is determined in step S203 that there is no defective mark, and die bonding is performed in step S204. Hereinafter, the same operation is performed for all the pellets of the wafer.

There is also known a method of detecting a mirror pellet by changing the amount of illumination light without attaching a defect mark to the mirror pellet, and performing die bonding while removing the mirror pellet. FIG. 5 is a flowchart showing a method for manufacturing such a semiconductor device.
No. 43 was proposed. First, in step S301, the illumination is controlled to the amount of light that can be correctly recognized by the non-defective pellet. In step S302, the quality of the pellet is recognized. Here, a marking is attached to the defective pellet.

If it is a non-defective product, the pellets are die-bonded in steps S304 to S307, and if it is a defective product, the process proceeds to step S310. After die bonding is performed in step S307, in step S308, the wafer table is moved by one pitch, and in step S309, it is determined whether or not pitch driving for one wafer has been completed. If not completed, the process returns to step S301.

In step S310, a determination is made as to the pellet size of the pellet determined to be defective. If the pellet size is small, the process proceeds to step S308; otherwise, the process proceeds to step S3.
Proceed to 11. In step S311, the illumination is controlled to a light amount suitable for obtaining binary image data of the mirror pellet, which is registered in advance.

In step S312, it is recognized whether or not it is a mirror pellet. In step S313,
If it is not a mirror pellet, the process proceeds to step S308. If it is a mirror pellet, the process proceeds to step S314.
After removing the mirror pellet from the wafer in Step S318, the process proceeds to Step S308.

[0015]

In the conventional pellet recognition method described with reference to FIG. 3 and FIG. 4, when a defect mark of the same size is attached to the mirror pellet and the pattern-bearing pellet, the mirror pellet 1 There is a problem that it is necessary to perform a re-strike of a defective mark on the mirror pellet 1 in a pre-process of die bonding because it is impossible to distinguish between the non-defective product and the non-defective product 2 having the pattern.

In another conventional method for recognizing pellets described with reference to FIG. 5, each time a defective product is generated, the amount of illumination light is switched to that for a mirror pellet, and thereafter, it is adapted again to a pellet having a pattern. Since the light amount has to be controlled, there is a problem that the number of steps increases and the operation takes time.

The present invention has been made in view of the above-mentioned drawbacks of the prior art, and has as its object to identify a mirror pellet without performing operations such as re-pointing a defective mark or controlling the amount of illumination light. It is intended to improve the overall workability.

[0018]

According to the present invention, there is provided, according to the present invention, a method for picking up a good pellet from a diced wafer having a mirror pellet and a defective pellet with a marking, and performing die bonding. In the method of manufacturing a semiconductor device, the reflected light from each pellet is subjected to binarized image recognition based on two types of luminance thresholds, and the pellet size and mark size are detected to determine the quality of the pellet. And a method of manufacturing a semiconductor device.

[0019]

Next, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a flowchart showing a method for recognizing a pellet in a die void process according to an embodiment of the present invention. FIG. 2 is a flowchart showing a wafer sensing position and an optical image when a defect mark is determined and a mirror pellet or a good pellet is determined. It is a figure showing the relation with brightness.

First, in step S101, the quantity of illumination light, luminance thresholds I and II, and mark size thresholds I and II are set. The brightness threshold value I is set to a level at which the pattern pellet 2 can be properly recognized, and the brightness threshold value II is set to a level at which the pattern pellet 2 cannot be recognized and the mirror pellet can be recognized. Is done. Also, it is set that the mark size threshold value I <the mark size threshold value II.

In step S102, the outer shape and position of the pellet are detected at the luminance threshold value I. If the outer shape of the pellet is defective, the process proceeds to step S106. If the outer shape of the pellet is good, the process proceeds to step S103. Step S10
At 3, the detected mark size detected at the luminance threshold value I is compared with the mark size threshold value I. If the detection mark size <the mark size threshold value I, there is no defective mark 3, that is, it is determined to be a good product or the mirror pellet 1, and the process proceeds to step S104. If the detection mark size ≧ the mark size threshold value I, there is a defective mark. That is, it is determined that there is a defective pellet with a pattern, and the process proceeds to step S106.

In step S104, the detected mark size detected at the luminance threshold value II is compared with the mark size threshold value II to determine whether the pellet is a mirror pellet or a non-defective pellet. That is, if the detection mark size> the mark size threshold value II, it is determined that the pellet is a non-defective product, and the process proceeds to step S105.
And the process proceeds to step S106 (the luminance threshold value II).
In the case of binarization, it is recognized that a mark is attached to the entire surface of a good pellet.)

In step S105, the pellets determined to be non-defective are die-bonded, and the flow advances to step S106. In step S106, it is determined whether or not the determination for all the pellets for one wafer has been completed. If completed, the process ends. Otherwise, the determination is performed for the next pellet via step S107.

Now, assuming that the pellets A, B, C, and D are arranged as shown in FIG. 2 and the determination proceeds from left to right, the determination is first performed for the pellet A. In step S102, it is determined that the outer shape is defective. After step S106 and step S107, the process returns to step S102, and the determination is performed on the pellet B. In step S102, the external shape is determined to be good, and in step S103, the presence or absence of a defective mark is determined.
In the mark detection at the luminance threshold value I, the mark on the mirror pellet cannot be detected, so it is determined that there is no defective mark, and the process proceeds to step S104.

In step S104, the detected mark size detected at the luminance threshold value II and the mark size threshold value II are compared.
Proceed to step S106. Then, the determination on the next pellet C is started via step S107.

The pellet C is determined to be non-defective in step S102, and the presence or absence of a defective mark is determined in step S103. The detected mark size detected at the luminance threshold value I is compared with the mark size threshold value I. Since the detected mark size ≧ the mark size threshold value I, it is determined that there is a defective mark in the pellet having a pattern, After steps S106 and S107, the process proceeds to the determination of the next pellet D.

After the pellet D is determined to be non-defective in step S102, in step S103
It is determined that there is no defective mark, and the process proceeds to step S104.
In step S104, since the detected mark size is larger than the mark size threshold II in the luminance threshold value II, it is determined that the pellet is a non-defective product, and die bonding is performed in step S105. Hereinafter, the same operation is performed for all the pellets of the wafer.

In the above embodiment, the defect mark 3 is attached to the mirror pellet, and the detected mark size is compared with the mark size threshold value II in step S104. May be changed. That is, it is abolished to put a defect mark on the mirror pellet.
However, in steps S103 and S104, the same determination as in the embodiment is performed.

[0029]

As described above, according to the present invention, in the step of judging the quality of the pellet at the time of die bonding, two steps are taken.
This method sets the type of brightness threshold and detects the pellet size and mark size based on the threshold value. It is not necessary to adjust the light amount of the pellet, and the non-defective product of the pellet can be determined only by the binarization recognition based on the two kinds of luminance thresholds, and the efficient die bonding operation can be performed.

[Brief description of the drawings]

FIG. 1 is a flowchart showing a work flow for judging pass / fail of a pellet according to an embodiment of the present invention.

FIG. 2 is a diagram illustrating a relationship between a wafer sending position and an optical image luminance of a pellet for explaining an embodiment of the present invention.

FIG. 3 is a flowchart showing a conventional work flow for judging the quality of pellets.

FIG. 4 is a diagram illustrating a relationship between a wafer sending position and an optical image luminance of a pellet for explaining a conventional example.

FIG. 5 is a flowchart showing the flow of another conventional example for determining the acceptability of pellets.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Mirror pellet 2 Patterned pellet 3 Defective mark 4 Re-strike point defective mark 5 Dicing wafer

Claims (4)

(57) [Claims] 1. A method of manufacturing a semiconductor device in which a good pellet is picked up from a diced wafer having a mirror pellet and a defective pellet having a defective mark and die bonding is performed, two types of reflected light from each pellet are used. A method of manufacturing a semiconductor device, comprising: performing binarized image recognition based on a luminance threshold value; detecting a pellet size and a mark size; 2. The method according to claim 1, wherein the two types of luminance threshold values are a first luminance threshold value capable of recognizing a non-marked portion of a pellet having a pattern, and a first luminance threshold value capable of recognizing a non-marked portion of a pellet having a pattern. 2. The method according to claim 1, wherein the threshold value is a second luminance threshold value at which a non-mark portion of the mirror pellet can be recognized. 3. A method according to claim 1, wherein the pellet size detected based on the binarized image data obtained by the first luminance threshold is within a predetermined range and the mark size obtained in the same manner is the first size. A first determining step of determining whether the pellet is smaller than a predetermined value and a second determining step of determining whether the pellet is a mirror pellet based on the binarized image data obtained by the second luminance threshold value. 3. The method of manufacturing a semiconductor device according to claim 2, further comprising the steps of: 4. A mirror mark is provided on the mirror pellet, and a pellet size detected based on the binarized image data obtained by the first luminance threshold value is within a predetermined range. A first determination step of determining whether or not the mark size obtained in the same manner is smaller than a first predetermined value; and a binary image data obtained by the second luminance threshold value. 3. The method according to claim 2, further comprising: a second determining step of determining whether the detected mark size is larger than a second predetermined value.