JPH02137339A - Pellet bonding device - Google Patents

Abstract

PURPOSE:To improve reliability of a product in case of pellet bonding and to shorten coping time by providing means for detecting the position of an image frame input from a TV camera and inspecting paste coating state after coating on the basis of the detection result. CONSTITUTION:Illumination means 6 for illuminating and imaging a frame, TV cameras 5a, 5b, 5c, and paste coating inspecting means having image processing means for inspecting coating state such as paste coating amount after coating, coating position and coating angle on the basis of the detection results are provided. The paste coating inspecting means is so constructed as to check the states such as paste coating state, pellet position deviation, etc., on the basis of the inspection result of the image processing means. Thus, the reliability of a product is improved, and regulating time can be shortened.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to a pellet bonding device in a semiconductor assembly process, and in particular, a pellet bonding device suitable for improving product reliability and shortening one setup time during pellet bonding. 'jAht.

[Prior Art] In a conventional pellet bonding apparatus, wafer position detection is disclosed in, for example, Japanese Patent Application Laid-Open No. 60-24513.
As described in Japanese Patent No. 9, a method has been proposed in which detection is performed every time a pellet is picked up.

Regarding the detection of defective pellets, for example,
As described in Publication No. 1-129839.

A method has been proposed in which detection is performed by providing an ink mark detection area.

Furthermore, frame position detection is performed before pellet bonding.

Furthermore, regarding the paste application inspection, the operator will make adjustments and
A method has been implemented in which images from a camera installed above the frame feeder are monitored.

[Problems to be Solved by the Invention] In the above-mentioned conventional technology, after applying paste to the tab of the frame, there is a problem with automation of inspecting the amount of paste applied and the application position, and correcting the amount of paste applied and the application position. This makes it impossible to early detect defects in the amount of paste applied, such as no paste application, poor application, excessive application, and misalignment of all pastes or multiple pastes, and to reduce the defective rate. There was a problem.

Furthermore, since the position of the wafer is detected every time a pellet is picked up, there is a problem in that it takes a long time to detect the position.

In addition, the detection of defective pellets requires a lot of time to set up because the ink mark detection area is set by the operator, and because only one pellet is detected per recognition, there are many There is a problem in that it takes a long time to bond, especially if there are a number of defective pellets.

Also, since frame position detection is not performed before applying paste, blocks may occur due to misalignment of paste application.

There are problems such as dirt on the bonding head, dirt on the pellet, and uneven wettability, resulting in cracks and cracks in the pellet, and in order to prevent this, workers have to spend a lot of time counting and making adjustments. There was an unavoidable problem.

Pellet bonding also has a problem in that it takes a lot of time to make adjustments when changing product types.

An object of the present invention is to provide a pellet bonding device that can improve product reliability and shorten setup time.

[Means for Solving the Problems] In order to achieve the above-mentioned object 1, the pellet bonding apparatus of the present invention includes an illumination means and a TV camera that illuminate and image a frame, and a position of the frame from an image input from the TV camera. is detected, and based on the detection result, the amount of paste applied after application is determined. (a) A paste application inspection means having an image processing means for inspecting application conditions such as cloth position and application angle.

In addition, the paste application inspection means is configured so that it is possible to check the paste application state and the status of pellet misalignment based on the inspection results by the image processing means, in order to shorten setup time by omitting adjustment work by one operator. It is what was done.

In addition, the image processing means detects the four sides of the rectangular tab of the frame to enable automation of teaching.
The center of gravity of the tab is determined from the detected four sides, and the position of the frame is thereby detected.

Furthermore, since it is difficult to illuminate the image processing means from below with a frame feeder, a mirror is provided below the frame, and the illumination from the illumination means is totally reflected by the mirror to detect the position of the frame. It is structured as follows.

In order to achieve the above object, the pellet bonding apparatus of the present invention includes an illumination means and a TV camera that illuminate and image the frame, and detects the position of the frame from the image input from the TV camera, and based on the detection result. With or without pellets.

The apparatus is equipped with a pellet bonding inspection means having an image processing means for inspecting the bonding state of the pellets such as misalignment and chipping.

In order to achieve the above object, the pellet bonding apparatus of the present invention includes: illumination means and a TV camera that illuminate and image the frame; and an image processing means that detects the position of the frame from an image input from the TV camera.
The apparatus includes a paste application means for applying paste based on the detected position of the frame by the image processing means, and a pellet bonding means.

Further, in order to simplify the configuration, the image processing means is configured to be used in common with the image processing means of the paste application means and the pellet bonding inspection means.

In order to achieve the above object, the pellet bonding apparatus of the present invention includes: illumination means and a TV camera that illuminate and image the frame; and an image processing means that detects the position of the frame from an image input from the TV camera.
The apparatus includes a paste application means for applying paste based on the detected position of the frame by the image processing means, and a pellet bonding means for bonding pellets.

In order to achieve the above L1 target, the pellet bonding apparatus of the present invention includes an illumination means and a TV camera that illuminate and image the wafer, and detects two orthogonal dicing lines in the image input from the TV camera. and a wafer positioning detection means having an image processing means for calculating the intersection point position and the slope of the straight line, and a movement control means for controlling the movement of the wafer based on the detection result of the image processing means. .

In addition, in order to achieve the above object, the pellet bonding apparatus of the present invention includes: illumination means and a TV camera for illuminating and imaging the wafer; , and a defective pellet detection means for detecting defective pellets without inputting the ink mark detection area.

In addition, the image processing means is configured to be able to simultaneously detect the pellet to be inspected and subsequent pellets to be inspected, in order to speed up detection by omitting the detection of defective pellets each time the pellet is picked up. It is something.

In addition, in order to shorten the detection time, the image processing means is configured so that when an edge is detected, it can detect the next stage of pellets.

Further, in order to achieve the above object, the pellet bonding apparatus of the present invention includes at least one lighting means and a T.
A wafer position detection means configured to share a V camera and an image processing means, a defective pellet detection means detecting a defective pellet, a frame position detection means detecting a frame position, The apparatus is equipped with a paste application means for applying paste, a paste application inspection means for inspecting the state of the paste after paste application, a pellet bonding means for bonding pellets, and a pellet bonding inspection means for inspecting pellet bonding.

[Function] The present invention detects the position of the frame before applying the paste and inspects the condition of the paste after applying the paste, thereby improving the accuracy of the paste application position.

Improved reliability and shortened setup time.

Furthermore, since a mirror is installed under the frame to provide pseudo-transparent illumination and the center of gravity is calculated from the four sides of the tab, it is possible to improve positional accuracy.

Further, regarding wafer position detection, since representative point detection is performed only when wafers are replaced, highly accurate position detection can be performed.

Furthermore, regarding the detection of defective pellets, the setup time can be shortened by inspecting the wafer position to determine the defective pellet detection position and performing detection without providing an ink mark detection area.

Furthermore, since defective pellets are read in advance and the next stage is skipped when detecting edges, the detection time can be shortened.

Also, after bonding only when changing types.

Since pellet bonding is inspected and the condition of pellets is automatically checked, setup time for pellet bonding can be shortened.

In addition, an image of the frame is captured by an 'rv camera, and the position of the frame is detected from this image by an image processing section. Based on the detected position, the movement of the paste application nozzle and bonding head is controlled by a movement control means to accurately Since paste application and pellet bonding are performed on the machine, we prevent blocks caused by misalignment of paste application, dirt on the bonding head, cracks in the pellets caused by dirt on the pellets, and uneven wetting properties. Adjustment time by the operator can be shortened.

[Example] Hereinafter, drawings showing one embodiment of the present invention will be described.

FIG. 1 is a system configuration diagram showing a pellet bonding apparatus which is an embodiment of the present invention.

In FIG. 1, 1 is a recognition means, 2 is a drive control means, and 3 is a recognition means.
is a wafer, 4 is a wafer setting section, and 5 is a wafer TV camera 5a which serves as a camera selection means and inputs an image of the wafer 3.
Then, the pellet TV camera 5b formed at a magnification that can input an image of the entire pellet, and the tab TV camera 5c that inputs an image of the frame tab are selected by a command from the recognition means 1. Reference numeral 6 denotes illumination control means, which includes a bright field illumination 6a and a ring illumination 6b installed in one optical system 7a fixed above the wafer setting section 4, and the other optical system 7b fixed above the bonding position. A bright field illumination 6c and a ring illumination 6d installed in the recognition means 1 are controlled by commands from the recognition means 1. Reference numeral 8 denotes motor drive control means, which includes an X table motor 8a and a Y table motor 8b that move and position the wafer setting section 4 in the XY force directions, and move the paste application nozzle 12 and the second bonding head 13 in the XY force directions. X table motor 8c and Y table motor 8d for positioning
and are driven by the output signal from the drive control means 2. 9 is used as a first bonding head to pick up the pellet 17. 10 is an intermediate positioning pocket, and the picked up pellet 17 is precisely positioned mechanically. Reference numeral 11 denotes a frame feeder, which holds a block 14 so as to be inserted thereinto, and moves and positions a frame 15 mounted on the upper surface of the frame feeder to a predetermined position. The block 14 is formed into a mirror surface to totally reflect the illumination light.

Next, the operation will be explained using the operation sequence diagram shown in FIG.

The operation is performed by the cooperation between the pellet bonder drive control means 2 and the recognition means 1.

First, when the wafer 3 is supplied to the wafer setting section 4 by a wafer loader (not shown), the recognition means 1 selects in advance the bright field illumination 6a and the ring illumination 6b of one optical system 7a via the illumination control means 6. The wafer TV camera 50 is controlled to the optimum value determined by the camera selection means 5.
The image of the wafer 3 is manually generated, and the position of the wafer 3 is detected from the manually generated image. The position detection of the wafer 3 is as follows:
This is done only when wafer 3 is supplied. The drive control means 2 also controls the motor drive control means 8 based on the detection information from the recognition means 1.
The X table motor 8a and the Y table motor 8bt+- are driven via the wafer set section 4 to move the wafer 3 to the pickup position.

Thereafter, when the wafer 3 is positioned at the pick-up position, the recognition means 1 controls the bright field 1K (bright 6a and ring illumination 6b to optimal values, inputs an image of the entire pellet from the pellet TV camera 5b, and detects the pellet 17). As a result, only the pellets 17 determined to be good are picked up using the first bonding head 9 and accurately positioned in the intermediate positioning pocket 10.

On the other hand, when the frame 15 is positioned at the bonding position on the block 14 by the frame feeder 11, the recognition means 1 controls the bright field illumination 6c and ring illumination 6d of the other optical system 7a to optimum values, and also controls the tab TV camera 5c. The image of the tab 16 is input using the input image, the position of the tab 16 is detected from the input image, and the position information of the tab 16 is input to the drive control means 2. The drive control means 2 drives the X table motor 8C and the Y table motor 8dti via the motor drive control means 8 based on the position information of the tab 16 from the recognition means 1 to position the paste application nozzle 12. .

Thereafter, paste is applied onto the tab 16 from the paste application nozzle 12.

On the other hand, an image of the tab 16 is captured by the tab TV camera 5c and inputted to the recognition means 1, and the recognition means 1 inspects the amount of paste applied, the application position, etc.

Next, the recognition means 1 inspects the paste application amount and application position of the tab 16, and the results are fed back to the drive control means 2, and are used for adjusting the amount of paste discharged from the paste application nozzle 12, correcting the application position, etc. .

If the inspection results indicate that the product is defective, a warning will be issued to the operator.

Next, the pellet positioned in the intermediate positioning pocket 10 is adsorbed by the second bonding head 13, and the pellet is bonded onto the tab 16 by driving the X table motor 8C and the Y table motor 8d based on the tab position information. .

When bonding is completed, defects such as presence or absence of pellets on the tab 16, misalignment, chipping, and paste adhesion are determined. Similar to the paste test, this judgment test also uses tab TV.
The frame tab 16 is imaged by the camera 5c, and the image is input to the recognition means 1, which inspects the quality.

As a result, if it is determined to be defective, a warning is issued to the operator.

Further, in the case of a positional deviation of the pellet, a signal is input to the drive control means 2 to correct the position of the pellet.

In this way, when the pellet adheres to the tab 16,
The X-table motor 8C and the Y-table motor 8b are driven once by the output signal from the drive control means 2 via the motor drive control means 8, and the wafer cellar 1 part 4 moves to set the next pellet at the pickup position. do.

Thereafter, when there are no more good pellets on the wafer 3, the next wafer 3 is supplied to the wafer setting section 4 by the wafer loader and the above operation is performed again.

Note that when paste application and pellet bonding are performed at the same location as in this example, one TV
It suffices to detect the position once with a camera, but if paste application and ballen 1-bonding are performed at separate stations, install an 'rv camera above each station, and paste the paste at each station. The same effect as this embodiment can be obtained by detecting the position of the frame 15 before each coating and pellet bonding.

Next, the individual recognition items described above in the recognition means 1 will be specifically explained.

First, the position of frame 15 is detected.
The pellet is adhered to the central tab 16 as shown in FIG. The second optical system 7b used for detecting the frame position is also used for paste application inspection and pellet bonding inspection, and therefore has a magnification that allows input of the entire tab 16 in one field of view, as described above.

In order to detect the position of the frame 15 with high precision.

Since the outline of the tab 16 needs to be input accurately as an image, lighting becomes a problem.

When illumination is used from above, such as bright field illumination 6c and ring illumination 6d, as shown in FIG. 4 (aHb),
On the tab 16 side, the specular reflection component of the light is strong, so it becomes bright, whereas the background reflects diffusely and looks dark.However, because the edges of the frame 15 are round, it is not possible to accurately represent the actual external shape. To represent the external shape, use Figure 5(a)
As shown in (b), a silhouette image by so-called transmitted illumination illuminated from the back side of the frame 15 is optimal, but in reality it is difficult to illuminate the frame feeder 10 from below.

Therefore, in this embodiment, the block 14 is used as a mirror surface to completely reflect the illumination light as shown in FIGS. 6(a) and 6(b), thereby obtaining pseudo-transmitted light.

Next, the detection procedure will be explained with reference to FIG.

For detection, since the tab 16 is rectangular, using this,
First, the four straight lines of the tab 16 are detected, and the position of their center of gravity is calculated to improve accuracy.

Input the image on tab 1G, as shown in Figure 7(a),
A binary image is obtained and the approximate position of the tab 16 is detected from this binary image.

That is, as shown in FIG. 7(b), a binary image is
Scanning is performed in both directions, the length of a continuous 0/1 image string is determined, and the length is compared with the dimension of the tab 16 inputted in advance to find a pixel string corresponding to the tab 16.

The midpoint of the pixel rows in both the X and Y directions is the approximate position (temporary center of gravity) of the tab 16.

Next, the obtained temporary center of gravity and the scanning position I determined by teaching! c (relative distance from the center of gravity), scanning positions @ (eight locations) for detecting the edges of the tab 16 are determined as shown in FIG. 7(C).

After that, edge detection processing is performed at eight scanning positions,
Four straight lines forming a rectangle are calculated from the obtained coordinates, and the exact position of the tab 16 is determined as shown in FIG. 7(d).

In addition, prior to the edge detection process, the data necessary for this is taught in advance, but this work is only performed for the first product to be produced, and the data is stored on a floppy disk or the like so that it can be performed from the first time onwards. There's no need.

In addition, during teaching, inputting the illumination level, tab dimensions, determining the binarization threshold, and determining the edge detection scanning position are performed, but all of these can be input manually or determined automatically. is also possible. For example, the binarization threshold value is set as shown in FIG. 7(a), as shown in FIG.
) If you take the brightness histogram of the tab image, it will be divided into two peaks, the bright background and the dark tab 16, so there is a method such as taking the middle value (valley) between the two peaks.

As for the edge search scanning position, a position that forms a straight line must be selected. However, the tab 16 has a portion called a tab suspension lead, as shown in FIG. 8(b), and the number and position of the tab suspension lead differs depending on the product type. Therefore, as shown in FIG. 8(b), a small window is set up, eight locations where the 0 area and 1 area are divided in half are searched, and the tab 16 is set to the scanning position.
Calculate and store the relative distance to the center of gravity.

Teaching can be automated by using the methods described above.

Next, wafer position detection will be explained with reference to FIG.

As shown in FIG. 9(a), the wafer 3 is set in the wafer setting section 4 after being subjected to so-called dicing, in which the wafer 3 is scratched and divided into a plurality of pellets. The position and angle of the wafer 3 are determined by detecting the intersection of the dicing lines 18 from the enlarged image as shown in FIG. 9(b). This detection process is performed at 2 to 4 locations on the wafer 3 and calculated.

In this case, the wafer TV camera 5a is set to an enlargement magnification such that the dicing line 18 has three or more pixels.

In addition, the intersection detection process is performed from point 2 on the wafer 3 as far away as possible.
Since the accurate position of the wafer 3 can be detected by performing the process at four locations, the position detection process only needs to be performed once after setting the wafer 3, and does not need to be performed every time a pellet is picked up. .

Next, the intersection detection procedure of the dicing lines 18 is performed at the first θ.
This will be explained using figures.

After setting the wafer 3, input the image on the wafer 3 from the wafer TV camera 5a, perform straight line detection processing according to the procedure shown in FIG.
Check the status of. At this time, if two orthogonal dicing lines 18 are detected, the intersection position and the slope of the straight line are calculated.

In addition, if two straight lines cannot be detected, the XY bidirectional tables 8a and 8b are driven based on the detection information to remove the wafer.
Move and find the intersection.

This process is performed at 2 to 4 locations on the wafer 3, and the position and inclination of the wafer 3 are calculated based on the results.

Next, the straight line detection method shown in FIG. 11 will be explained with reference to the vertical straight line shown in FIG. 12.

A dicing line 1 is created by scanning a binarized straight line in the X direction as shown in FIG. 9(b) manually by the method already explained.
Find an O pattern corresponding to the width of 8.

(FIG. 11(1)) If it is not found, move the scanning position several pixels in the Y direction and scan again. If a similar pattern is detected for several pixels consecutively, it is regarded as part of a straight line (line segment). (FIG. 11(2)) If it cannot be detected, continue with FIG. 11(1).

If n or more line segments are found by searching for line segments within a certain pitch and a certain width, it is determined that the line is the target line. (If Figure 11 is not found, continue with Figure 11 (1).

If the target straight line is detected, the straight line distance is calculated using the least squares method from the detected plural line segments, and if a horizontal straight line is found, the intersection point is calculated.

The intensity and binarization threshold of the bright field illumination 6a and the ring illumination 6b during input are set and memorized in advance to values that allow the pellet 17 and the dancing line 18 to be clearly distinguished during teaching, and Used for position detection.

Next, detection of defective pellets will be explained with reference to FIG. 13.

Defective pellets include, as shown in FIG. 13, those that are missing because they are on the edge 20 of the wafer 3, and those that are determined to be defective in the pellet sorting process and are marked with ink marks 19.

The image for detecting a defective pellet as described above must contain the pellet 22 to be inspected within the field of view, and therefore is input at a lower magnification than in the case of detecting the position of the wafer 3. The intensities of the bright field illumination 6a and the ring illumination 6b during input are also memorized at values that allow the ink mark 19, wafer edge 20, and pellet 17 to be clearly distinguished during teaching, and the illumination control unit 6 at the time of detection.
Control with.

The same applies to the binarization threshold.

As a method for detecting the ink mark 19, the binary image is subjected to segmentation (area division) processing, and the ink mark 1
Check whether or not an O region with an area corresponding to 9 exists in the pellet to be inspected l-22, or scan the inside of the pellet to be inspected 22 to find a 0 region with a length corresponding to the ink mark 19. There are ways to find out whether or not.

The position of the pellet 22 to be inspected is calculated based on the position of the wafer 3, which is determined in advance.

Therefore, there is no need to set the detection area of the ink mark 19 during teaching.

The detection of the wafer edge 20 is performed by detecting 4 of the pellets 22 to be inspected.
A wafer edge detection area 21 is provided at the corner.

If the number of 0 counts in the area or the ratio of 0 in the area (0 force count/total area) exceeds a certain value, it is determined that the pellet is defective.

When detecting defective pellets, if multiple pellets 17 are included in one image as shown in FIG.
Not only the next pellet to be inspected 22 to be picked up, but also the subsequent pellets are inspected at the same time. This eliminates the need for manual image processing and inspection processing every time the image is picked up, thereby increasing speed. For example, in the case of FIG. 14, the pellet 22 to be inspected is defective, but the next two pellets on the right are good.

Therefore, when picking up wafer 3, pellet 1
It is sufficient to pick up the good pellets sent individually, and then pick up the good pellets without detecting the bad pellets at the next pickup.

Further, as shown in FIG. 15, if one or more pellets with a wafer edge in a certain direction are detected, it is determined that there are no good pellets in that stage, and the process skips to the next stage. This allows the takt time to be further shortened.

Next, a paste application inspection performed after paste is applied to the tab 16 will be described.

In this case, the input image from the tab TV camera 5c is the 16th
As shown in the figure, the bright field illumination 6c and the ring illumination 6d are set in advance during teaching so that the paste 23 and the tab 16 can be clearly distinguished, and the set values are stored.

Similarly, the binarization threshold value is set in advance during teaching, but in this case, automatic setting is possible by applying the threshold determination method using the histogram described above inside tab 16. .

Furthermore, it is also possible to automatically set the bright field illumination 6c and the ring illumination 6d so that the difference between the peaks and valleys of the histogram increases.

The procedure for paste application inspection will be explained.

A tab image is input from the tab TV camera 5c and binarized to obtain a binarized image as shown in FIG.

At this time, since the position of the tab 16 has already been detected, the paste detection area 24 is provided based on that information.

The inside thereof is segmented to detect all O areas with an area corresponding to the paste 23, or the paste detection area 24 is scanned and a 0 area corresponding to the paste 23 is detected.

Next, as shown in FIG. 16, the center of gravity of all the pastes or several pastes such as those present in the four corners is detected, and the center of gravity of the entire paste (nozzle center) is calculated based on these values.

Thereafter, the calculated value is compared with the tab gravity center value determined in advance, and if the deviation is equal to or greater than a certain value, it is determined that the positional deviation is defective.

Furthermore, if a plurality of paste gravity centers are detected, it is also possible to inspect the angular deviation of the entire paste in both the X and Y directions.

Next, it is determined whether each paste 23 is not coated, poorly coated, overcoated, etc. based on its area.

Further, by accurately extracting the outer shape of the paste 23 and comparing it with one circle, its roundness can be determined.

Further, as shown in FIG. 19, if the area of the paste 23 increases or decreases in a certain direction, it can be determined that the paste 1 application nozzle 12 is tilted as shown in FIG. can.

By performing the above-mentioned inspection immediately after applying the paste, it is possible to detect defects early and reduce the defective rate.

Next, a pellet bonding inspection performed after the pellet 17 is bonded to the tab 16 will be described.

In this case as well, the bright field illumination 6c and the ring illumination 6 are arranged so that the input image from the tab TV camera 5c can clearly distinguish between the tab 16 and the pellet 17 as shown in FIG.
The intensity of d and the binarization threshold are set in advance.

Next, the pellet bonding inspection method will be explained.

First, a pellet detection area 27 is provided as shown in FIG. 22 based on the inspected tab position information. Segmentation processing or scanning is performed inside the pellet 1.
7 areas are detected.

After that, the contour of the detected pellet 17 area is tracked, and the coordinate data or contour of the tracking point sequence is approximated to a polygon to obtain a polygonal vertex sequence (line segmentation point sequence 25).
Get data.

Use this data as the four sides (four straight lines) that make up the pellet's external shape.
into groups. The Hough transform is used for grooving. There are methods such as calculating the slope of a line segment and dividing it into 90° increments. Straight lines are individually determined for each of the four divided groups using the least squares method or the like.

Next, as shown in FIG. 23, the four intersections of the four straight lines are calculated, and the center of gravity of the pellet is determined. The center of gravity of the pellet is compared with the center of gravity of the tab, and the slope is calculated from the intersection to determine the displacement of the pellet.

After that, as shown in FIG. 24, for the tracking point sequence or the line segmentation point sequence of the contour of the pellet 17 that is not on the four sides of the pellet 17, two adjacent sides of the four sides of the pellet 17 are When perpendicular lines are drawn in the X and Y directions from the multiple tracking points or line division points above, the lengths of the two perpendicular lines, ht, are determined and the shorter one is chipped 26 If the limit value is exceeded, it is determined that a defect such as a chip 26 or paste adhesion exists in that part.

Since these inspections are also performed immediately after paste bonding, it is possible to prevent unnecessary defects by early detection of defects.

The above paste application inspection and bonding inspection are performed when changing products and are used as an automatic check of the paste condition and pellet condition, which can omit the position v1 adjustment by the operator and significantly shorten setup time. can be measured.

Therefore, according to the present embodiment, changes in high precision detection of flake 11 position, wafer position, paste application, high precision and high reliability of pellet bonding, high speed detection of defective pellets, and various processes at the time of product replacement are achieved. The setup time can be shortened.

In the above embodiment, there are three TV cameras 5, bright field illumination 6a, 6c, and ring illumination 6b.

Although two sets of 6d are reinstalled, the number is not limited to this, and the number can be changed at any time by configuring them so that they can be shared.

[Effects of the Invention] According to the present invention, since paste application and pellet bonding are performed after detecting the frame position, paste application position accuracy is improved, blocks caused by paste application deviation, prevention of bonding head contamination, and pellet contamination are prevented. This has the effect of improving product reliability by preventing pellet breakage and cracks that occur due to uneven wettability, and shortening the adjustment time required by operators. In addition, highly accurate frame position detection is possible by inputting an accurate image by using a mirror block when detecting the frame position, and by detecting the center of gravity of the tab from the four sides of the tab.

Further, by detecting the wafer position with high precision only when exchanging wafers, processing time can be reduced.

Defective pellet detection has the effect of shortening one setup time because it is not necessary to set the ink mark detection area during teaching. Furthermore, by adopting a method in which several pellets are determined to be defective in one recognition, and when an edge is detected, the method is skipped to the next stage, the detection time can be reduced.

In addition, by performing a paste application inspection and a pellet bonding inspection when replacing one type of product, and using the status of the paste and pellets as an automatic check, it is possible to omit equipment adjustment by the operator, which has the effect of shortening time.

[Brief explanation of the drawing]

FIG. 1 is a perspective view of a system configuration showing a pellet bonding apparatus that is an embodiment of the present invention, FIG. 2 is an operation sequence diagram, FIG. 3 is a perspective view of an example of a frame to which the present invention is applied, and FIG. (a) Figure 5 (a) Figure 6 (n) is an explanatory diagram showing the state of light when illuminating the tab, Figure 4 (b)
Figure 5 (b) Figure 6 (b) is a binary image diagram, Figure 7 is a frame position detection process diagram, Figure 8 is a teaching automation method diagram, and (a) is a brightness histogram diagram of the tab image. , Part (b) is an explanatory view showing the brightness area of the window installed on the outline of the tab, Part (c) is an enlarged explanatory view of the window, and Fig. 9(a) is a perspective view showing a wafer to which the present invention is applied. , FIG. 9(b) is an enlarged view of the pellet and dicing line shown in FIG. 9(a), FIG. 10 is a wafer position detection operation sequence diagram, FIG. 11 is a dicing line detection operation sequence diagram, and FIG. 12 13 to 15 are pellet image diagrams for explaining a defective pellet detection operation, and FIGS. 16 to 19 are tab image diagrams for explaining a paste application inspection operation. FIG. 20 is an explanatory diagram showing an example of the state during paste application;
FIG. 21 is an image of the tab after pellet bonding, FIGS. 22 and 23 are explanatory diagrams showing the bonding inspection operation procedure, and FIG. 24 is an explanatory diagram for explaining the chipping inspection method. DESCRIPTION OF SYMBOLS 1... Recognition means, 2... Drive control means, 3... Wafer, 4... Wafer setting part, 5... Camera selection means, 6... Lighting control means, 7... Optics Series, 8...
Motor drive-9 control means, 9... first bonding head, 10... intermediate positioning pocket, 11... frame feeder, 12... paste application nozzle, 13
...Second bonding head, l4...Block, 15...Frame, 16...Tab. 17... Beret.

Claims (1)

[Claims] 1. A pellet bonding apparatus that applies a paste to a frame and then bonds pellets picked up from a wafer to the paste, comprising an illumination means and a TV camera that illuminate and image the frame, and the TV camera. paste application inspection means having an image processing means for detecting the position of the frame from an image inputted from the frame and inspecting the application state such as the application amount, application position and application angle of the paste after application based on the detection result. Pellet bonding equipment. 2. A pellet bonding apparatus according to claim 1, wherein the paste application inspection means is configured to be able to check the paste application state and the positional deviation of the pellet based on the inspection result by the image processing means. 3. The image processing means according to claim 1 is configured to detect the four sides of the rectangular tab of the frame, determine the center of gravity of the tab from the detected four sides, and thereby detect the position of the frame. pellet bonding equipment. 4. The image processing means according to claim 3 is a pellet bonding apparatus comprising a mirror disposed below the frame and configured to detect the position of the frame by totally reflecting the illumination from the illumination means on the mirror. 5. A pellet bonding device that applies paste to a frame and then bonds pellets picked up from a wafer to the paste, including an illumination means and a TV camera that illuminate and image the frame, and a frame from the image input from the TV camera. A pellet bonding apparatus comprising a pellet bonding inspection means having an image processing means for detecting the position of the pellet and inspecting the bonding state of the pellet such as the presence or absence of the pellet, misalignment, chipping, etc. based on the detection result. 6. The pellet bonding inspection means according to claim 5,
A pellet bonding apparatus configured to be able to check the bonding state of pellets based on inspection results by an image processing means. 7. A pellet bonding apparatus, wherein the image processing means according to claim 6 is configured to share the image processing means according to claim 1. 8. The image processing means according to claim 6 is configured to detect the four sides of the rectangular tab of the frame, determine the center of gravity of the tab from the detected four sides, and thereby detect the position of the frame. Pellet bonding equipment. 9. A pellet bonding device that applies a paste to a frame and then bonds pellets picked up from a wafer to the paste, including an illumination means and a TV camera that illuminate and image the frame, and a frame from the image input from the TV camera. A pellet bonding apparatus comprising: an image processing means for detecting the position of a frame; a paste application means for applying a paste based on the position of the frame detected by the image processing means; and a pellet bonding means for bonding pellets. 10. A pellet bonding apparatus configured such that the image processing means according to claim 9 shares the image processing means according to claim 1 or 5. 11. The image processing means according to claim 9 is configured to detect the four sides of the rectangular tab of the frame, determine the center of gravity of the tab from the detected four sides, and thereby detect the position of the frame. pellet bonding equipment. 12. A pellet bonding apparatus according to claim 10, wherein the paste application means and the pellet bonding means are installed either at the same location or at different locations. 13. The image processing means according to claim 11 is a pellet bonding apparatus comprising a mirror disposed below the frame and configured to detect the position of the frame by totally reflecting the illumination from the illumination means on the mirror. . 14. A pellet bonding apparatus that applies a paste to a frame and then bonds pellets picked up from a wafer to the paste, including an illumination means and a TV camera that illuminate and image the wafer, and the T.
An image processing means that detects two orthogonal dicing lines in an image input from a V camera and calculates the intersection position and the slope of the straight line, and controls the movement of the wafer based on the detection results of the image processing means. A pellet bonding apparatus comprising a wafer position detection means and a movement control means. 15. A pellet bonding apparatus that applies a paste to a frame and then bonds pellets picked up from a wafer to the paste, including an illumination means and a TV camera that illuminate and image the wafer, and a TV camera that illuminates and images the wafer.
It has an image processing means that calculates the position of the pellet to be inspected from the image input from the camera based on the position of the wafer detected in advance, and detects defective pellets without inputting the ink mark detection area. Pellet bonding equipment equipped with pellet detection means. 16. A pellet bonding apparatus according to claim 15, wherein the image processing means is configured to be able to simultaneously detect a pellet to be inspected and a subsequent pellet to be inspected. 17. A pellet bonding apparatus according to claim 16, wherein the image processing means is configured to detect a next stage of pellets when an edge is detected. 18. A wafer configured to share at least one illumination means, a TV camera, and an image processing means in a pellet bonding apparatus that applies a paste to a frame and then bonds pellets picked up from a wafer to the paste. a wafer position detection means for detecting the position of the wafer, a defective pellet detection means for detecting a defective pellet, a frame position detection means for detecting the position of the frame, a paste application means for applying the paste, and a state of the paste after applying the paste. A pellet bonding apparatus comprising: a paste application inspection means for inspecting paste application inspection means; a pellet bonding means for bonding pellets; and a pellet bonding inspection means for inspecting pellet bonding.