JP2004140084A - Pick-up method and device for picking up semiconductor chip - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To reduce chip searching operations in frequency and to improve a pick-up operation in efficiency. <P>SOLUTION: A semiconductor chip pick-up device is equipped with a first image sensing means (first camera 3) which images a pick-up target chip C and outputs an image I1 as image information and a first image processing means 4 which recognizes coordinates of a barycenter and an imperfect external appearance on the image I1. a second image sensing means (second camera 6) which images other chips C around the target chip C and outputs an image I2 as image information, a second image processing means 8 which determines whether the other chips C bear bad marks or not, recognizes the coordinates of a barycenter on the image I2, and forms map data of information concerning the other chips C as candidates for pick-up target chips when it is determined that the other chips C are not defective, and a control means 13 which moves a table 1 for correction on the basis of table correcting information obtained from the first image processing means 4 and moves the table 1 on the basis of the map data obtained from the second image processing means 8. <P>COPYRIGHT: (C)2004,JPO

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a semiconductor chip pickup method and a pickup device thereof, and more particularly, to a pickup device used in the manufacture of electronic components and a method of picking up a chip by image recognition.
[0002]
[Prior art]
2. Description of the Related Art Conventionally, when manufacturing a semiconductor device, a large number of chips are arranged and arranged on a wafer, and the chips arranged on a table together with the wafer are image-recognized and only good chips are picked up.
[0003]
Specifically, first, a plurality of chips are fetched one by one as images by a camera, and the fetched imaging signals are output to an image processing unit. Then, the image processing device performs image processing on a chip in the image recognition area in the camera view.
[0004]
Next, the image data of the chip obtained by the image processing is stored in the frame memory, and the quality of the chip is determined by recognizing the image data. Here, the quality of the chip is determined mainly on appearance defects such as chipping.
[0005]
Then, the chips determined to be good by the image recognition are picked up by vacuum suction using a special jig called a collet.
[0006]
Next, as shown in FIG. 20, taking a case where a plurality of chips C are arranged on a wafer table (not shown) as an example, a captured image and a pickup state when pickup is performed by a conventional pickup method. , And the movement locus of the wafer will be described with reference to FIGS.
[0007]
Before starting the pickup process, an inspection process for roughly inspecting whether the product is defective is performed in advance. The chips C2 and C3 determined to be defective in this inspection process are provided with a bat mark "●" indicating that they are defective.
[0008]
FIG. 21 is a captured image of the first chip C1. First, the quality of the chip C1 in the image recognition area a is determined by performing image recognition. If it is determined that the chip C1 is non-defective with no defective appearance such as chipping, the wafer table is moved so that the center of gravity of the chip C1 is positioned at the pickup position O. After the correction movement, the chip C1 is picked up by a collet. Then, when the chip C1 is picked up, the state shown in FIG. 22 is obtained.
[0009]
Next, as shown in FIG. 23, the wafer table is moved to the left on the X-axis by a predetermined pitch feed amount, and an image of the image recognition area a is captured. Then, image processing is performed on the image recognition area a, and the acceptability of the chip C2 located in the image recognition area a is determined. Here, since the chip C2 has a bat mark “●” in advance, it is determined that the chip C2 is defective, and no pickup is performed.
[0010]
Then, as shown in FIG. 24, the wafer table is moved to the left on the X-axis in the same manner as described above, and the image in the image recognition area a is captured. Then, image processing is performed in the image recognition area a, and it is determined that there is no chip C in the image recognition area a.
[0011]
Next, as shown in FIG. 25, the wafer table is moved to the left on the X-axis, and an image in the image recognition area a is captured. Then, image processing is performed in the image recognition area a, and it is determined that there is no chip C in the image recognition area a.
[0012]
Next, as shown in FIG. 26, the wafer table is moved to the left on the X-axis, and an image in the image recognition area a is captured. Then, image processing is performed in the image recognition area a, and it is determined that there is no chip C in the image recognition area a.
[0013]
In this way, as shown in FIGS. 24 to 26, when it is determined that there is no chip C consecutively three times, the pitch feed direction of the X axis is set to the reverse (rightward). Further, as shown in FIG. 27, the image is moved upward in the Y-axis by a predetermined pitch feed amount, and the image is captured. Then, image processing is performed in the image recognition area a, and it is determined that there is no chip C in the image recognition area a.
[0014]
Next, as shown in FIG. 28, the wafer table is moved rightward on the X-axis by a predetermined pitch feed amount, and an image in the image recognition area a is captured. Then, image processing is performed in the image recognition area a, and it is determined that there is no chip C in the image recognition area a.
[0015]
Next, as shown in FIG. 29, the wafer table is moved rightward on the X-axis in the same manner as described above, and the image in the image recognition area a is captured. Then, image processing is performed on the image recognition area a, and the acceptability of the chip C3 located within the image recognition area a is determined. In this pass / fail determination, since the chip C3 has a bat mark, it is determined that the chip C3 is defective and no pickup is performed.
[0016]
Next, as shown in FIG. 30, the wafer table is moved rightward on the X-axis, and an image in the image recognition area a is captured. Then, image processing is performed on the inside of the image recognition area a, and the acceptability of the chip C4 located within the image recognition area a is determined. If it is determined that the chip C4 is non-defective, the wafer table is corrected and moved so that the center of gravity of the chip C4 is located at the pickup position O, and the chip C4 is picked up by a collet. When the chip C4 is picked up, the state shown in FIG. 31 is obtained.
[0017]
Prior art document information related to this type of pickup method includes the following.
[0018]
[Patent Document 1]
JP-A-2-155243
[0019]
[Problems to be solved by the invention]
However, in the pick-up method, regardless of the quality of the chip C, all the chips C are positioned one by one in the image recognition area a, imaged and image-recognized, and the quality is determined. , The pickup operation is executed, and the chip C of the defective product is not picked up, and proceeds to the next chip C. Since the moving direction of the wafer table is not changed until it is determined that there is no chip C three times in a row, an operation to search for the chip C at the edge of the wafer where no chip exists from the beginning occurs as shown in FIG. However, there is a problem that the operation efficiency of the pickup operation is reduced.
[0020]
In order to reduce this search operation, there is a method in which chip position data on a wafer is created in advance by an inspection device before the semiconductor wafer is placed in the pickup device, and the pickup device moves the wafer table based on this data. is there. However, in this case, the pickup of the wafer may not be performed correctly because the shape of the wafer sheet changes and the chip position changes due to the influence of the needle push-up at the time of pickup or a change in time.
[0021]
SUMMARY OF THE INVENTION It is an object of the present invention to provide a semiconductor chip pickup method and apparatus capable of reducing the number of chip search operations and performing an efficient pickup operation.
[0022]
[Means for Solving the Problems]
In order to solve the above-mentioned problems, a semiconductor chip pickup method of the present invention is a semiconductor chip pickup method for sequentially picking up a predetermined chip to be picked up among a plurality of chips arranged on a table by a pickup means. Using a first camera having a predetermined magnification whose axes are matched and a second camera having a lower magnification than the first camera, the first camera images the chip to be picked up, and the second camera picks up the chip to be picked up. A plurality of chips around the target chip are imaged, and the presence / absence of a chip around the chip to be picked up, the presence / absence of a bat mark in the peripheral chip, and the coordinates of the center of gravity are recognized based on the image taken by the second camera. Chips with no bat mark based on the The information is stored as a pickup candidate, the table is moved based on the information, and the presence / absence of a chip to be picked up, the presence / absence of appearance defect such as chipping, and the barycentric coordinates are determined by an image taken by the first camera. When the table is recognized and it is determined that there is no appearance defect, the table is corrected and moved based on the recognized barycenter coordinates, and the table is picked up.
[0023]
According to this pickup method, the coordinates of the center of gravity of the chip to be picked up and the presence or absence of appearance defects such as chipping can be recognized with high accuracy by the first camera, and at the same time, the coordinates of the center of gravity of a plurality of chips to be picked up by the second camera in the future. , The presence or absence of a bat mark can be recognized. Therefore, one of the centroid coordinates of a plurality of good chips (pickup target chip candidates) obtained from the second camera is selected as a chip to be picked up next, and a pickup position (an image capturing position by the first camera). The table can be moved quickly.
[0024]
In this pickup method, the first camera has an angle of view capable of capturing an image of one of the plurality of chips on the table in the field of view, and the second camera has the first camera in the field of view. It is preferable that the angle of view includes at least 3 × 3 rows of chips including the chip to be picked up by a camera. In this way, the number of chip search operations can be reduced, and it is preferable in terms of accuracy in determining the presence or absence of a bat mark, camera performance, price, and the like.
[0025]
Further, among the chips in the image picked up by the second camera, a plurality of chips existing in the moving direction of the pick-up execution line having the picked-up chip are determined as good or bad, and the barycentric coordinate data of the chips determined as good are picked up. As the execution column map data, the coordinate data of the chip closest to the pickup position in the execution column map data is set as the next chip candidate to be picked up, and after picking up the chip to be picked up, based on the coordinate data of the chip candidate to be picked up. Preferably, the table is moved. With this configuration, the imaging by the first camera is performed only on substantially non-defective chips, so that the number of chip search operations can be reduced.
[0026]
Further, the plurality of chips are arranged on a wafer, arranged on the table together with the wafer, and among chips in an image captured by the second camera, adjacent to a pickup execution row having the chip to be picked up. The plurality of chips located in the next column to be judged pass / fail is determined, and the center-of-gravity coordinate data of the chip determined to be non-defective is taken as the next column map data. Is preferably determined. By doing so, it is possible to eliminate the chip search operation that is performed a plurality of times at the wafer end as in the related art.
[0027]
In addition, among the chips in the image captured by the second camera, there is no pick-up target chip candidate in the moving direction of the pick-up execution row having the pick-up target chip, and the pick-up execution row map data exists. When the target chip candidate is smaller than the distance obtained by subtracting the distance between the chips X from the visual field X dimension of the second camera, it is preferable to move the table based on the next row map data created during the front row pickup. By doing so, the number of chip search operations can be further reduced.
[0028]
In addition, among the chips in the image captured by the second camera, if there is no candidate chip to be picked up in the moving direction of the pickup execution row having the chip to be picked up and no map data of the pickup execution row exists, or If there is real column map data, but the picked-up chip candidate is separated by more than the distance obtained by subtracting the distance between chips X from the visual field X dimension of the second camera, the movement obtained by subtracting the distance between chips from the visual field of the second camera. It is preferable to move the table by an amount. This can further reduce the number of chip search operations.
[0029]
Further, the coordinates of the center of gravity of the chip are obtained by subjecting the captured images of the first camera and the second camera to image processing by image processing means, and the information is updated every time the table is moved. Is preferred. In this way, the chip position can be correctly recognized even when the sheet shape of the wafer changes due to the influence of needle raising at the time of pickup, a change in time measurement, or the like.
[0030]
Further, in the present invention, as a pickup device to which the pickup method is applied, there is provided a semiconductor chip pickup device for sequentially picking up a plurality of chips arranged on a table by a pickup means, wherein the pickup object arranged on the table is First imaging means having a lens of a predetermined magnification for imaging the chip and outputting the image information as image information, and recognition of presence / absence of appearance defects such as barycentric coordinates and chipping of the chip to be picked up by an image taken by the first imaging means. A first image processing means for performing the processing, and a second lens having a lower magnification than the first imaging means for imaging a plurality of chips present around the chip to be picked up arranged on the table and outputting the image information as image information. A plurality of channels are provided by an imaging unit and an image captured by the second imaging unit. A second image processing means for recognizing the presence / absence of a bat mark and the coordinates of the center of gravity and determining map data as a candidate for a chip to be picked up when it is determined to be a good chip, and an input from the first image processing means. The table is corrected and moved based on the obtained table correction information so that the pickup target chip is located at the pickup position, and the next pickup target chip candidate is located at the pickup position based on the map data input from the second image processing means. Control means for moving the table so as to control the pickup means.
[0031]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, embodiments of the present invention will be described with reference to the drawings.
FIG. 1 shows a pickup device for performing a semiconductor chip pickup method according to an embodiment of the present invention. This pickup device generally includes a wafer table 1, a first camera 3, a first image processing unit 4, a second camera 6, a second image processing unit 8, a collet 10, and a control unit for controlling these. 13
[0032]
The wafer table 1 arranges a large number of chips C together with wafers W arranged vertically and horizontally at predetermined intervals. The wafer table 1 has a table driving unit 2 fixed so that the center thereof is the coordinate origin, and is arranged such that the center Wo of the wafer W is located at the coordinate origin. The table drive unit 2 moves vertically and horizontally (X direction and Y direction) in accordance with an instruction from the control unit 13 and can move diagonally by simultaneously moving both directions of the X direction and the Y direction. .
[0033]
The first camera 3, which is the first imaging means, has a high-magnification lens, is disposed so that its focal point is located at the center of the wafer table 1, and enlarges the chip C located at the focal point to a high magnification. The image is output to the first image processing unit 4 as image information. In other words, the first camera 3 has a high-magnification angle of view at which one chip C can be magnified and imaged in the field of view. Therefore, an image can be captured in fine pixel units, and the target chip C can be accurately recognized.
[0034]
The first image processing unit 4 recognizes the presence / absence of the chip C to be picked up, the presence / absence of an appearance defect such as chipping, and the center of gravity coordinates, based on the captured image I1 captured by the first camera 3, and converts the table correction data. The recognition result is output to the control unit 13. The first image processing unit 4 is connected to a first image monitor 5 so that an image captured by an operator can be confirmed on the first image monitor 5.
[0035]
The second camera 6, which is a second imaging unit, has a lens with a lower magnification than the first camera, and images a plurality of chips C arranged on the wafer table 1 at a lower magnification to image. The information is output to the second image processing unit 8 as information. Specifically, the second camera 6 is disposed so as to have the same focal point as the first camera 3 and the optical axis coincide with each other by the pair of reflectors 7A and 7B. As shown in FIG. 2, the second camera 6 according to the present embodiment includes, among the plurality of chips C arranged on the wafer table 1 in the field of view, the chip to be picked up in the field of view of the first camera 3. The angle of view is set such that five chips C in the X direction and three chips C in the Y direction can be imaged. In the present embodiment, since the second camera 6 is configured so that the optical axis of the second camera 6 coincides with that of the first camera 3, the pickup target chip C is located at the center (the pickup position O) in the field of view of the second camera 6. It is located in.
[0036]
Based on the captured image I2 input from the second camera 6, the second image processing unit 8 determines whether there is a chip C present around the chip C to be picked up, The presence / absence of a bat mark “●” indicating the presence and the coordinates of the center of gravity are recognized, and map data of the chip candidate C to be picked up which is determined to be non-defective is created and output to the control unit 13. The second image processing unit 8 is connected to a second image monitor 9 so that an image captured by an operator can be confirmed by the second image monitor 9.
[0037]
The map data created by the second image processing unit 8 includes pick-up execution column map data and next column map data.
[0038]
The execution column map data includes, in the movement direction of the column (X direction) having the pickup target chip C currently performing the pickup operation, the pickup target chip candidate C that is determined to be non-defective in the search area a1. It is composed of barycentric coordinate data. That is, the second image processing unit 8 determines the area of the plurality of chips C existing in the search area a1, determines the acceptability of the chips C existing within the allowable area range based on the presence or absence of the bat mark, and determines that the chips C are non-defective. The determined barycenter coordinates of the chip C are used as execution column map data.
[0039]
The next column map data includes the center of gravity of the candidate chip C to be picked up, which is determined to be non-defective in the search area a2, in the next (lower) column adjacent to the pick-up execution column having the chip C to be picked up. It is composed of coordinate data. That is, the second image processing unit 8 determines the area of the plurality of chips C present in the search area a2, and determines the acceptability of the chips C within the allowable area range based on the presence or absence of a bat mark, and determines that the chips C are non-defective. The coordinates of the center of gravity of the chip C thus obtained are used as the next column map data.
[0040]
When the wafer table 1 is moved to the left, the range of the search area a1 is such that the X direction is the right field of view RX from the pickup position O and the Y direction is the center of the pickup position O as shown in FIG. This is a range obtained by adding the distance PY between the upper and lower chips C to the outer Y dimension CY of the chip C. Conversely, when the wafer table 1 is moved to the right, as shown in FIG. 4, the X direction is the left field of view LX from the pickup position O, and the Y direction is the external Y dimension CY of the chip C centered on the pickup position O. Is the range obtained by adding the distance PY to the upper and lower chips C.
[0041]
As shown in FIG. 5, the range of the search area a2 is the entire range AX of the captured image I2 in the X direction, and the Y direction is the external Y dimension CY of the chip C and the distance PY between the upper and lower chips C. This is the range of the addition. The base end (upper end) in the Y direction is half below the outer dimension CY of the chip C from the pickup position O, that is, the lower end of the chip C in the pickup execution row located at the pickup position O.
[0042]
The collet 10, which is a pickup means, is driven by a collet driving unit 11, and vacuum-adsorbs the chip C present at the pickup position O of the wafer table 1, and transports the chip C to the chip mounting unit 12 in the vacuum-adsorbed state. It is.
[0043]
The control unit 13 controls the collet driving unit 11 and the table driving unit 2 based on table correction data input from the first image processing unit 4 and map data input from the second image processing unit 8. is there. Specifically, in the present embodiment, four controls are performed according to the status of the chips C on the wafer table 1.
[0044]
In the first control, the position of the wafer table 1 is corrected and moved based on the table correction data input from the first image processing unit 4, and the chip C to be picked up is picked up by the collet 10.
[0045]
In the second control, among the plurality of chips C in the image I2 captured by the second camera 6, the chip C existing in the moving direction of the pickup execution row having the pickup target chip C is determined to be good or defective. The coordinate data of the chip C closest to the pickup position O is set as the next chip candidate C to be picked up, and the wafer table 1 is moved and positioned.
[0046]
The third control is to check whether or not the end (limit) of the wafer W has been reached. If it is determined that the end has been reached, the wafer table 1 is moved in the Y direction to change the pickup execution line, At the time of the turnback, the pickup target chip candidate C closest to the end of the row is moved and positioned so as to be located at the pickup position O.
[0047]
Specifically, in order to perform the third control, a limit radius Wr is set in advance for the wafer W as shown in FIG. As a result, a circular wafer limit Wl having the origin at the center Wo of the wafer W is set.
[0048]
Then, when the wafer table 1 is moved to the left side, when there is no chip candidate C to be picked up as a good chip on the right side of the pick-up execution row, and as shown in FIG. When the judgment value based on the relationship between the check point lp1 for confirming and the pickup position O is larger than the limit radius Wr, and based on the relationship between the check point lp2 for confirming the limit of the next row and the pickup position O. When the determination value becomes larger than the limit radius Wr, it is determined that the end of the wafer W has been reached, and the turning operation is performed. The coordinate positions of lp1 and lp2 are obtained by the following equation (1). In addition, the limit determination value based on these check points can be obtained by the following mathematical expression 2.
[0049]
(Equation 1)
lp1X = OX + second camera 6X field of view / 2, lp1Y = OY
lp2X = OX + second camera 6X field of view / 2, lp2Y = OY + Y movement pitch TY
[0050]
(Equation 2)
Wafer limit judgment value = √ (lpnX ² + LpnY ² )
[0051]
When the wafer table 1 is moved to the right, check points lp3 and lp4 are used instead of check points lp1 and lp2.
[0052]
The fourth control is a control method mainly when it is determined that the next chip candidate C to be picked up exists outside the field of view of the captured image I2 based on the next row map data created at the time of picking up the front row. In such a situation, and when the next chip candidate C to be picked up is smaller than the field of view of the second camera 6 by a distance reduced by the distance between the chips C, the next chip candidate C based on the next row map data created at the time of the front row pickup is used. Is determined, and the wafer table 1 is directly moved and positioned so that the chip candidate C to be picked up is located at the pickup position O. Further, in the above situation, when the next chip candidate C to be picked up is separated from the field of view of the second camera 6 by a distance that is smaller than the field of view of the second camera 6 by the distance of the chip C from the field of view of the second camera 6. The wafer table 1 is moved by the reduced movement amount, and a preliminary operation for creating the next column map data is performed. The preparatory operation is performed in the same manner even when there is no map data of the pickup execution column.
[0053]
Specifically, in the fourth control, in the image I2 picked up by the second camera 6, there is no chip candidate C to be picked up in the moving direction of the pickup execution row and there is no map data of the pickup execution row, or When there is map data of the picked-up execution row by the next row map data created at the time of picking up, but there is no effective map data when the distance is greater than the distance X obtained by subtracting the X distance TX of the chip C from the second camera view X dimension AX. Judge. Then, the wafer table 1 is moved by a movement amount obtained by subtracting the distance TX between the chips C from the X dimension AX of the second camera's visual field AX, and only the second camera 6 is processed to perform the preliminary operation for creating the next column map data. I do.
[0054]
For example, when the wafer table 1 moves rightward and is in the state shown in FIG. 8, chips C10 to C12 and C19 to C23 are captured in the captured image I2 of the second camera 6. However, since the chips C11 and C12 existing in the same row as the chip to be picked up C10 are marked with bad marks, they are determined to be defective, and there is no chip to be picked up C in the captured image I2. On the other hand, for the picked-up chip candidate C16 located on the picked-up row, coordinate data at the time of picking up the front row is created, and these are present as picked-up row map data. The chips C19 to C23 in the captured image I2 have completed the creation of coordinate data as the next column map data, but have not completed the creation of coordinate data for the chip C24 and subsequent chips.
[0055]
Here, as shown in FIG. 9, when the wafer table 1 is moved and positioned so that the coordinate position of the next pickup target chip candidate C16 becomes the pickup position O, pass / fail determination of chips C25 to C29 in the next row is performed. And the coordinate data can be recognized. However, the pass / fail judgment of the chip C24 and the coordinate data cannot be recognized.
[0056]
Therefore, in the present embodiment, as shown in FIG. 8, the distance between the next chip candidate C16 to be picked up and OX is longer than the distance obtained by subtracting the distance TX between chips X from the second camera view X dimension AX. In this case, as shown in FIG. 10, the wafer table 1 is moved by a moving amount obtained by subtracting the distance TX between the chips C from the second camera view X dimension AX. This prevents a missing portion from being generated in the next column map data.
[0057]
Further, in the fourth control, as shown in FIG. 11, in the captured image I2 of the second camera 6, there is no pickup candidate chip candidate C on the pickup execution line, but the pickup object chip candidate C15 during the front row pickup. In the case where the coordinate data of C18 to C18 exists as the execution column map data, the distance between OXs of the next chip candidate to be picked up C15 is obtained by subtracting the distance X between the chips 12 from the second camera view X dimension AX. Shorter than the distance. Therefore, as shown in FIG. 12, the moving position of the wafer table 1 is directly determined as the position of the chip candidate C15 to be picked up, based on the next line map data created at the time of picking up the front line.
[0058]
Next, the operation of the pickup device will be described.
As shown in FIG. 13, when the pick-up operation is started by the operation of the operator, the pick-up device first performs the pick-up operation shown in steps S1 to S6 and the map data creation operation shown in steps S7 to S9. Perform parallel processing.
[0059]
Specifically, in step S1, it is determined whether or not the wafer table 1 has been positioned (second control or fourth control) based on the map data by the second image processing unit 8. When the positioning is performed based on the map data, the process proceeds to step S2. When the positioning is not performed based on the map data, the process skips steps S2 to S6 and proceeds to step S10.
[0060]
Here, since the map data by the second image processing unit 8 does not exist at the beginning of the pick-up operation, the process skips steps S2 to S6 and proceeds to step S10. If the map data obtained by the second image processing unit 8 includes the chip candidate C to be picked up after the second time after any of steps S12, S16, and S17 described later, the process proceeds to step S2. Become. Further, after the second time, if the pickup target chip candidate C does not exist in the map data obtained by the second image processing unit 8, the process skips steps S2 to S6 and proceeds to step S10. That is, after the second time, when the process returns to step S1 via steps S12 and S16, the process proceeds to step S2, and when the process returns to step S1 after step S17, the process proceeds to step S10.
[0061]
In step S2, after the image I1 is captured by the first camera 3, in step S3, the first image processing unit 4 that has received the image I1 executes an image recognition process for the chip C to be picked up.
[0062]
Next, in step S4, it is determined whether or not the pickup target chip C recognized by the first image processing unit 4 is non-defective, and if it is non-defective, the process proceeds to step S5, and if not, the process proceeds to step S10. .
[0063]
In step S5, the wafer table 1 is corrected and moved (first control) based on the recognition processing in step S3, and the coordinates of the center of gravity of the chip C to be picked up and the pickup position O are matched. After that, in step S6, the collet 10 is operated via the collet driving unit 11, the target chip C is picked up, and the process proceeds to step S10.
[0064]
On the other hand, in step S7, after the image I2 is captured by the second camera 6, in step S8, the second image processing unit 8 which has received the image I2 reads the images of the plurality of chips C located around the chip C to be picked up. Perform recognition processing.
[0065]
Next, in step S9, map data including the execution column map data and the next column map data is created based on the peripheral chip C recognized by the second image processing unit 8, and the process proceeds to step S10.
[0066]
In step S10, the currently stored map data is corrected (updated and stored). More specifically, when the process proceeds to step S10 after steps S5 and S6, since a predetermined chip C to be picked up has been picked up, the data of the chip C is deleted from the data map. When the process proceeds to step S10 after step S4, the chip C is not picked up, but is determined to be defective, so that the data of the chip C is similarly deleted from the data map. In addition, the process proceeds from step S7 to step S10 via step S9, and if there is a chip C to be picked up without a bad mark around the chip C to be picked up, the barycenter coordinates of the chip C are set to the next coordinates. Add to column map data. At this time, if the next column map data of the barycenter coordinate of the chip C exists, it is compared with the old data, and if the coordinate difference is within the allowable range set in both XY, it is determined that the chip C is the same. Then, the old data is deleted and updated and stored.
[0067]
Next, in step S11, it is detected whether or not there is valid map data of the pickup execution row. If there is valid map data, the process proceeds to step S12, the wafer table 1 is moved (second control) based on the map data, and the process returns to the first step. If there is no valid map data, the process proceeds to step S13.
[0068]
In step S13, it is determined whether or not the edge (limit) of the wafer W has been reached (third control). When it is determined that the end has been reached, the process proceeds to step S14, and when it is determined that the end has not been reached, the process proceeds to step S17.
[0069]
In step S14, the return direction (Y direction) of the movement direction of the wafer table 1 is set, and in step S15, the map data sequence is changed so that the next column map data becomes the effective column map data. Thereafter, in step S16, the wafer table 1 is moved stepwise in the Y direction, and the process returns to the first step (steps S1, 7).
[0070]
If the wafer W has not reached the end, in step S17, the wafer table is step-moved in the X direction (fourth control), and the process returns to the first step.
[0071]
Next, the movement operation of the wafer table 1 in accordance with the flowchart of FIG. 13 will be specifically described with reference to FIG. In this example, the wafer table 1 is set to move to the left, and starts from a state where there is no map data.
[0072]
First, when the pick-up operation is started by an operator's operation, since map data has not been created at this time, steps S1 to S6 are skipped, the second camera 6 captures the image I2, and creates map data ( Steps S7 to S9) are performed.
[0073]
FIG. 14 shows a captured image I2 of the second camera 6 at this time. When the image I2 is captured (step S7), the second image processing unit 8 performs an area determination for the chips C1 and C2 located in the search area a1, which is the pickup execution column map data, and both chips C1 and C2 have an allowable area. It is determined to be within the range. Thereafter, bat mark recognition is performed on these chips C1 and C2. Thereby, it is determined that the chip C1 has no bad mark, it is determined that the chip C1 is the target chip C of the execution column map data, the center of gravity of the chip C1 is recognized, and the obtained coordinate data is used as the execution column map data. Remember. The chip C2 is determined to have a bat mark, and is excluded from the target chip C in the execution column map data.
[0074]
In addition, the area determination is performed on the chips C3 to C8 located in the search area a2, which is the target of the next row map data, based on the same captured image I2, and the chips C3 to C7 are determined to be within the area allowable range. Judge as out of range. Thereafter, bat mark recognition is performed for each of the chips C3 to C7. As a result, it is determined that the chip C3 has a bat mark, and is excluded from the target chip C in the map data. Further, the chips C4 to C7 are determined to have no bad mark, are determined to be the target chip C in the next row, and the center of gravity is recognized for these chips C4 to C7, and the obtained coordinate data is used as the next row map data. Remember.
[0075]
When the map data created in this manner is stored (step S10), since the effective map data is present in the pickup execution row, the limit calculation process for the wafer W is not performed, and it is determined that the time for turning back the wafer table 1 has not been reached (step S10). S11). Then, the wafer table 1 is moved based on the map data created as described above (step S12).
[0076]
Thereafter, when returning to the first step, the pickup target chip C1 closest to the pickup position O in the pickup execution column map data is positioned at the pickup position O as shown in FIG.
[0077]
Then, the first camera 3 captures the image I1 of the chip C1 to be picked up as shown in FIG. 16 (step S2), and based on the image I1, the first image processing unit 4 determines whether the chip C1 has an external defect such as the presence or absence, chipping or the like. The presence / absence and barycentric coordinates are recognized. If it is determined that the chip C1 has no defective appearance and is non-defective, the wafer table 1 is corrected and moved (step S5), and a pickup operation (step S6) is performed.
[0078]
When the chip C1 is picked up in this way, the data of the chip C1 is deleted from the picked-up execution column map data (step S10).
[0079]
Further, in parallel with the processing by the image I1 captured by the first camera 3, the second image processing unit 8 uses the captured image I2 of the second camera 6 shown in FIG. The area is determined for C2, and the area is determined to be within the allowable range. Then, bat mark recognition is performed on the chip C2, the chip C2 is determined to have a bat mark, and is excluded from the map data target chips.
[0080]
Further, the area determination is performed on the chips C3 to C8 located in the search area a2 based on the same captured image I2, and the chips C3 to C7 among them are determined to be within the allowable area. Then, bat mark recognition is performed for these chips C3 to C7, the chip C3 is determined to have a bat mark, and is excluded from the map data target chip C. Further, the chips C4 to C7 are judged to be non-defective, the positions of the centers of gravity are recognized, and the coordinate data is stored as the next column map data.
[0081]
Note that the coordinate data of each of the chips C4 to C7 in the map data is compared with the previously created old data, and when the coordinate difference is within the allowable range set in advance for both XY, it is determined that the chips are the same, and the old data is determined. Delete (update and store).
[0082]
When the data of the chip C1 is deleted as described above, as shown in FIG. 17, there is no valid map data in the pickup execution row. Therefore, the limit of the wafer W is checked (step S13).
[0083]
When it is determined that the wafer W is at the limit by checking the limit of the wafer W, the wafer table 1 is set to be turned back (step S14), and the next column map data is set as the pickup column map data (step S15).
[0084]
Thereafter, based on the newly set execution column map data, the pick-up target chip candidate C4 located at the end of the wafer table 1 among the pick-up target chip candidates C4 to C7 is determined as the movement position, as shown in FIG. Then, step movement is performed in the Y direction (step S16).
[0085]
As described above, the pickup device of the present invention performs the pickup by accurately determining the acceptability of the chip C to be picked up by the first camera 3 and positioning the center of gravity coordinate to the pickup position O without fail. At the same time, map data of the pick-up execution row having the pick-up target chip and the pick-up target chip candidate C arranged in the next adjacent row are created, and the wafer table 1 is moved based on the map data. In addition, when the pickup execution row is changed to the next row, the pickup target chip candidate located at the end of the next row to be changed is moved so as to be located at the pickup position O.
[0086]
Therefore, as shown in FIG. 19, the number of times of searching operation of the chips C generated at the edge of the wafer W or the like can be reduced, and the wafer table 1 can be moved efficiently. As a result, the speed of the pickup operation of the chip C can be increased.
[0087]
Moreover, since the coordinates of the center of gravity of each chip C are obtained by performing image processing on the captured images I1 and I2 of the first camera 3 and the second camera 6 for each operation by the image processing units 4 and 8, respectively, The chip position can be correctly recognized even if the sheet shape of the wafer W changes due to the influence of the needle push-up, a time change, or the like.
[0088]
Note that the pickup method and the pickup device of the present invention are not limited to the configuration of each of the above embodiments, and various modifications are possible.
[0089]
For example, in the above-described embodiment, the angle of view is set so that the second camera 6 can image 5 × 3 rows of chips C (5 in the X direction and 3 in the Y direction). May be the angle of view at which the image can be taken. Even in this case, the same operation and effect as described above can be obtained. Further, it may be possible to image a plurality of chips C of 5 × 3 rows or more, and in this case, the number of the preliminary operations and the like can be reduced. However, when the angle of view is such that an excessively large number of chips C can be taken in, the effect of reducing the number of chip search operations is improved, but the accuracy of bat mark presence / absence determination, camera performance, price, and the like are increased. Therefore, the range (number) in which the second camera 6 can capture an image may be determined as desired in consideration of these points. That is, in the present invention, the second camera 6 only needs to have an angle of view capable of imaging at least the chips C in 3 × 3 rows.
[0090]
【The invention's effect】
As is apparent from the above description, the pickup method and apparatus according to the present invention use the first camera and the second camera whose optical axes are coincident with each other, and the appearance of the chip for executing the pickup, such as the barycentric coordinates and chipping. The presence or absence of a defect is recognized with high accuracy by the first camera, and at the same time, the second camera is used to recognize the coordinates of the center of gravity of a plurality of chips to be picked up in the future and the presence or absence of bat marks. That is, the next chip candidate to be picked up can be recognized while picking up one chip to be picked up, so that the search step can be reduced.
[0091]
In addition, since the coordinates of the center of gravity of the chip are updated each time the table is moved, it is possible to correctly recognize the chip position even if the sheet shape changes due to the effect of needle push-up occurring at the time of picking up the chip to be picked up, a time change, or the like. it can.
[Brief description of the drawings]
FIG. 1 is a schematic configuration diagram showing a pickup device according to an embodiment of the present invention.
FIG. 2 is a plan view showing a search area for processing an image captured by a second camera.
FIG. 3 is a plan view showing a search area a1 when a wafer table is moved to the left by a second image processing unit.
FIG. 4 is a plan view showing a search area a1 when the second image processing unit moves the wafer table to the right.
FIG. 5 is a plan view showing a search area a2 by a second image processing unit.
FIG. 6 is a plan view showing a method of setting a wafer mitt.
FIG. 7 is a plan view showing a wafer limit confirmation method.
FIG. 8 is a process diagram for explaining a moving method when the next chip candidate to be picked up is separated in the picked-up column map data.
FIG. 9 is a process chart for explaining an erroneous moving method when the next chip candidate to be picked up is separated in the picked-up column map data.
FIG. 10 is a process chart for explaining the moving method of the present embodiment when the next chip candidate to be picked up is distant in the picked-up column map data.
FIG. 11 is a process chart for explaining another moving method when the next chip candidate to be picked up is distant in the picked-up execution column map data.
FIG. 12 is a process chart for explaining another moving method of the present embodiment when the next chip candidate to be picked up is distant in the picked-up execution column map data.
FIG. 13 is a flowchart showing an operation of the pickup device of the embodiment.
FIG. 14 is a plan view showing an image captured by a second camera at the beginning of the start of the pickup operation.
FIG. 15 is a plan view showing an image captured by a second camera in a step subsequent to FIG. 14;
FIG. 16 is a plan view illustrating an image captured by a second camera in a step subsequent to that of FIG. 15;
FIG. 17 is a plan view showing an image captured by a second camera in a step subsequent to FIG. 15;
FIG. 18 is a plan view showing an image captured by a second camera in a step subsequent to that of FIG. 17;
FIG. 19 is a schematic diagram showing a movement locus of a wafer table by the pickup device of the embodiment.
FIG. 20 is a plan view showing an example of a chip arrangement for explaining a conventional pickup method.
FIG. 21 is a plan view showing a captured image in a first step of a conventional pickup method.
FIG. 22 is a plan view showing a state where a chip is picked up in the state of FIG. 21;
FIG. 23 is a plan view showing a captured image in the next step of FIG. 21;
FIG. 24 is a plan view showing a captured image in a step subsequent to FIG. 23;
FIG. 25 is a plan view showing a captured image in a step subsequent to FIG. 24;
FIG. 26 is a plan view showing a captured image in a step subsequent to that of FIG. 25;
FIG. 27 is a plan view showing a captured image in a step subsequent to FIG. 26;
FIG. 28 is a plan view showing a captured image in a step subsequent to FIG. 27;
FIG. 29 is a plan view showing a captured image in a step subsequent to FIG. 28;
FIG. 30 is a plan view showing a captured image in a step subsequent to that of FIG. 29;
FIG. 31 is a plan view showing a state where a chip is picked up in the state of FIG. 30;
FIG. 32 is a schematic view showing a movement locus of a wafer table by a conventional pickup method.
[Explanation of symbols]
1. Wafer table 2. Table drive unit
3. First camera (first imaging unit) 4. First image processing unit
5 First image monitor 6 Second camera (second imaging unit)
7A, 7B: Reflector 8: Second image processing unit
9: second image monitor 10: collet
11: Collet drive unit 12: Chip mounting unit
13 ... Control unit
C: Chip W: Wafer
I1: Image captured by the first camera I2: Image captured by the second camera
a1, a2: Search area O: Pickup position

Claims (8)

A semiconductor chip pickup method for sequentially picking up a predetermined pickup target chip among a plurality of chips arranged on a table by a pickup means,
Using a first camera having a predetermined magnification whose optical axes are matched and a second camera having a lower magnification than the first camera, the first camera images the chip to be picked up, and the second camera picks up the chip to be picked up. Image multiple chips around the chip,
Based on the image captured by the second camera, the presence / absence of a chip around the chip to be picked up, the presence / absence of a bat mark in the chip around the chip, and the coordinates of the center of gravity are recognized. The information is stored as a chip candidate, and the table is moved based on the information,
Based on the image captured by the first camera, the presence / absence of a chip to be picked up, the presence / absence of appearance defects such as chipping, and the center of gravity coordinates are recognized, and if it is determined that there is no appearance defect, the table is updated based on the recognized center of gravity coordinates. A method for picking up a semiconductor chip, wherein the semiconductor chip is corrected and moved and picked up. The first camera has an angle of view capable of imaging one of a plurality of chips on the table in a field of view of the first camera,
2. The semiconductor according to claim 1, wherein the second camera includes the chip to be picked up by the first camera in a field of view, and has an angle of view capable of imaging at least 3 × 3 rows of chips. 3. How to pick up chips. Among the chips in the image picked up by the second camera, a plurality of chips existing in the moving direction of the pick-up execution line having the pick-up target chip are judged to be good or bad, and the barycenter coordinate data of the chip determined to be non-defective is picked up. As the map data, the coordinate data of the chip closest to the pickup position in the execution column map data is set as the next chip candidate to be picked up. 3. The method for picking up a semiconductor chip according to claim 2, wherein A plurality of the chips are arranged on a wafer, and are arranged on the table together with the wafer;
Among the chips in the image captured by the second camera, a plurality of chips located in the next row adjacent to the pickup execution row having the pickup target chip are determined to be good or bad, and the barycentric coordinate data of the chips determined to be good are determined. 4. The semiconductor chip pickup according to claim 2, wherein the next column map data is used, and a position to move to the next column is determined at the time of turning back at the wafer end based on the next column map data. Method. Among the chips in the image captured by the second camera, there are no candidate chips to be picked up in the direction of movement of the pick-up execution row having the pick-up target chips, and there is pick-up execution row map data. 4. The apparatus according to claim 2, wherein when the number of candidates is smaller than the distance obtained by subtracting the distance between the chips X from the visual field X dimension of the second camera, the table is moved based on the next row map data created at the time of picking up the front row. 5. The method for picking up a semiconductor chip according to claim 4. When there is no candidate chip to be picked up in the moving direction of the pickup execution line having the chip to be picked up among the chips in the image captured by the second camera, and no map data of the pickup execution line exists, or If the map data exists, but the chip candidate to be picked up is separated by more than the distance obtained by subtracting the distance between the chips X from the visual field X dimension of the second camera,
6. The semiconductor chip pickup method according to claim 2, wherein the table is moved by a movement amount obtained by subtracting a chip distance from the second camera field of view. The coordinates of the center of gravity of the chip are obtained by subjecting the captured images of the first camera and the second camera to image processing by image processing means, and the information is updated every time the table is moved. The method for picking up a semiconductor chip according to any one of claims 2 to 6. A semiconductor chip pickup device for sequentially picking up a plurality of chips arranged on a table by a pickup means,
A first imaging unit having a lens with a predetermined magnification for imaging a chip to be pick-up arranged on the table and outputting it as image information;
First image processing means for recognizing, based on an image taken by the first imaging means, the presence / absence of appearance defects such as barycentric coordinates and chipping of the chip to be picked up;
A second imaging unit having a lower magnification lens than the first imaging unit that images a plurality of chips existing around the pickup target chip arranged on the table and outputs the image information as image information;
Based on the image captured by the second image capturing means, the presence / absence of bat marks of a plurality of chips and the coordinates of the center of gravity are recognized, and if it is determined that the chip is non-defective, map data of the information is generated as a chip candidate to be picked up. Image processing means;
The table is corrected and moved based on the table correction information input from the first image processing means so that the chip to be picked up is located at the pickup position, and the next pickup is performed based on the map data input from the second image processing means. And a control means for controlling the pickup means by moving the table so that the target chip candidate is located at the pickup position.

Images