JP2007242988A - Image acquisition method of semiconductor chip - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image acquisition method of a semiconductor chip which contributes to a reduction of a labor in inspection of the semiconductor chip. <P>SOLUTION: In this method, a semiconductor wafer is divided into a plurality of semiconductor chips by extending a film 2 stuck to the semiconductor chip, and in this state, an image for inspection of the semiconductor chip is acquired. This method comprises a first step of acquiring an image of the semiconductor chip on the film 2 at a relatively low resolution, and a second step of acquiring the image of the semiconductor chip on the film 2 at a relatively high resolution. This method further comprises a third step of deciding an acquisition region of the image by the second step based on the image acquired by the first step. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to an image acquisition method for a semiconductor chip.

  The LED or other semiconductor chip is manufactured in a wafer state in which a large number of semiconductor chips are arranged vertically and horizontally.

  The wafer is separated into individual semiconductor chips as described below.

  FIG. 1 is a diagram for explaining a process of separating a wafer into semiconductor chips.

  As shown in FIG. 1, with the wafer 1 attached to the film 2, the film 2 is stretched around as shown by the arrow in FIG. Then, the individual semiconductor chips constituting the wafer 1 are separated from each other.

  As shown in the perspective view of FIG. 2, the film 2 in a state where the chip group 3 made up of a large number of separated semiconductor chips is attached is applied to one open end of the ring-shaped frame 4 with an appropriate tension. It is pasted.

  When the semiconductor chip is separated by stretching the film 2 as described above, the semiconductor chip is cracked, chipped, scratched, adjacent semiconductor chips are left unseparated, or separated. There may be a problem that the subsequent semiconductor chips are overlapped with each other. The semiconductor chip in which these problems occur must be handled as a defective product.

  For this reason, the separated semiconductor chip is subjected to an inspection as to whether or not a defect has occurred.

  Conventionally, inspection of a semiconductor chip after separation has been performed visually by an operator.

  It should be noted that no prior art document relating to the inspection of the semiconductor chip after separation could not be found.

  However, the size of the semiconductor chip is very small, and the number of semiconductor chips constituting one wafer is very large. Therefore, it is very laborious for an operator to visually inspect the semiconductor chip after separation. Was necessary.

  The present invention has been made to solve the above-described problems, and an object of the present invention is to provide a method for acquiring an image of a semiconductor chip that contributes to reducing labor in the inspection of a semiconductor chip.

  In order to reduce the labor of inspection of a semiconductor chip, it is conceivable to perform inspection by adopting an image recognition technique.

  In order to inspect the semiconductor chip using the image recognition technology, an image of the chip group 3 on the film 2 is acquired with the film 2 attached on the ring 4 as shown in FIG. Analyzing images.

  Acquisition of an image is performed by, for example, scanning a line sensor camera on the film 2.

  Here, the semiconductor chip is fine. For this reason, it is necessary to acquire a high-resolution image for precise inspection of the semiconductor chip.

  Since the chip group 3 includes an extremely large number of semiconductor chips, high-resolution images are acquired for all the semiconductor chips in the chip group 3, and further, the defect inspection of the semiconductor chip is performed using all the acquired images. It takes a lot of time to do.

  By the way, it may be recognized that there is a clear defect in the semiconductor chip without using a high-resolution image.

  That is, for example, when a crack or chip generated in a semiconductor chip is large, adjacent semiconductor chips are not separated from each other, or separated semiconductor chips are overlapped with each other Can recognize a defect of a semiconductor chip only by using a relatively low resolution image.

  Therefore, the semiconductor chip image acquisition method of the present invention is a method of acquiring an image for inspection of a semiconductor chip in a state where the semiconductor wafer is divided into a plurality of semiconductor chips by stretching a film to which the semiconductor wafer is attached. A first process of acquiring an image of the semiconductor chip on the film at a relatively low resolution, a second process of acquiring an image of the semiconductor chip on the film at a relatively high resolution, and the first process And a third step of determining an image acquisition area in the second step based on the image acquired in the step.

  In the semiconductor chip image acquisition method of the present invention, the first step is to divide an area where a chip group composed of a plurality of semiconductor chips exists on the film into a plurality of areas, and to divide the divided areas into individual areas. A fourth process for determining the inclination of the arrangement of the semiconductor chips in the chip group based on the image obtained by the first imaging process in the first process. If it is determined in the fourth step that the semiconductor chip array is tilted, the fifth step of correcting the posture of the film to correct the tilt is performed, and then the second step is performed. It is preferable to redo the process 1 from the beginning.

  In the semiconductor chip image acquisition method of the present invention, the first step is to divide an area where a chip group composed of a plurality of semiconductor chips exists on the film into a plurality of areas, and to divide the divided areas into individual areas. When there are semiconductor chips that consist of a plurality of imaging processes that sequentially acquire images each time, and in the image obtained by one imaging process that constitutes the first process, a part of the semiconductor chip protrudes from the image In the imaging process performed after the one imaging process, imaging in the imaging process performed next to the one imaging process is performed so that an image including the whole of the protruding semiconductor chip can be captured. It is preferable to set the range.

  Similarly, in the semiconductor chip image acquisition method of the present invention, in the second step, an area where a chip group consisting of a plurality of semiconductor chips exists on the film is divided into a plurality of areas, and the division is performed. A semiconductor chip consisting of a plurality of imaging processes for acquiring images sequentially for each region, and a part of the semiconductor chip protruding from the image in the one imaging process constituting the second process. If present, in the imaging process performed after the one imaging process, the imaging process performed after the one imaging process so that an image including the entire protruding semiconductor chip can be captured. It is preferable to set the imaging range at.

  The semiconductor chip image acquisition method of the present invention preferably further comprises a sixth step of determining the coordinates of each semiconductor chip based on the image acquired in the first step.

  According to the present invention, a first process of acquiring an image of a semiconductor chip on a film with a relatively low resolution, a second process of acquiring an image of a semiconductor chip on a film with a relatively high resolution, And a third process for determining an image acquisition area in the second process based on the image acquired in the process of, for example, the semiconductor chip has an obvious defect only by using a low-resolution image. If this can be recognized, it is possible to omit obtaining a high-resolution image for the semiconductor chip. Alternatively, it is possible to omit the determination of the presence of a blank area in which no semiconductor chip is present from a low-resolution image, and obtaining a high-resolution image for the blank area. Therefore, it is possible to shorten the time required for the inspection of the defect of the semiconductor chip.

  Embodiments according to the present invention will be described below with reference to the drawings.

  In the present embodiment, as a preferred example of an image acquisition method for a semiconductor chip according to the present invention, an image acquisition method for an LED (Light Emitting Diode) chip will be described.

  FIG. 3 is a plan view of the LED chip.

  As shown in FIG. 3, the LED chip 7 includes a substrate 8 and a plus electrode 9 and a minus electrode 10 formed on the substrate 8.

  Such an LED chip 7 is manufactured in a state of a wafer 1 (see FIG. 1) in which a large number of LED chips 7 are arranged vertically and horizontally.

  As shown in FIG. 1, the wafer 1 is separated into individual LED chips 7 by stretching the film 2 around the film 2 in a state where the wafer 1 is attached to the film 2.

  Furthermore, the film 2 in a state in which the chip group 3 composed of a large number of separated LED chips 7 is attached, as shown in FIG. 2, is applied to one open end of the ring-shaped frame 4 with an appropriate tension. It is pasted.

  In the present embodiment, for example, an image of the LED chip 7 on the film 2 is acquired using a line sensor camera. For example, by illuminating the film 2 from below and imaging the LED chip 7 from above the film 2 with a line sensor camera, an image can be suitably captured. A line sensor camera that can switch the resolution between a low resolution and a high resolution is used.

  Since the LED chip 7 is fine, it is necessary to obtain a high-resolution image for precise inspection of the LED chip 7.

  Since the chip group 3 includes an extremely large number of LED chips 7, a high-resolution image is acquired for the LED chips 7 of the chip group 3, and further, all the acquired images are used to detect defects in the LED chip 7. It takes a lot of time to perform the inspection.

  By the way, it may be recognized that the LED chip 7 has a clear defect without using a high-resolution image.

  That is, for example, cracks and chips generated in the LED chip 7 are large, adjacent LED chips 7 are not separated from each other, or separated LED chips 7 are overlapped with each other. In such a case, the defect of the LED chip 7 can be recognized only by using a relatively low resolution image.

  Therefore, in the present embodiment, first, an image of the LED chip 7 on the surface of the film 2 is acquired with a relatively low resolution (for example, a resolution of about 20 μm).

  Next, based on the acquired low-resolution image, an area on the surface of the film 2 that does not require acquisition of an image with a relatively high resolution (for example, a resolution of about 1 μm) is determined by image analysis.

  That is, for example, for the LED chip 7 in which a defect can be clearly recognized only by confirming a relatively low resolution image, it is not necessary to examine the presence or absence of the defect again using a relatively high resolution image. For this reason, it is determined that the existence area of the LED chip 7 in which the defect is clearly recognized only by confirming the relatively low resolution image is an area where it is not necessary to acquire the relatively high resolution image.

  Next, an image of the LED chip 7 in a region excluding a region where it is determined that acquisition of an image at a relatively high resolution is unnecessary is acquired at a relatively high resolution.

  And the presence or absence of a defect is detected about each LED chip 7 using a comparatively high resolution image.

  In addition, processing for acquiring a relatively low resolution image, processing for determining an area that does not require acquisition of a relatively high resolution image by image analysis based on the obtained relatively low resolution image, and relatively low resolution image And the process of detecting the presence or absence of defects in each LED chip 7 using a relatively high resolution image are automatically performed by computer processing.

  FIG. 4 is a plan view showing the chip group 3 in a state of being stuck on the film 2.

  As shown in FIG. 4, relatively low resolution imaging by the line sensor camera is performed by dividing the area where the chip group 3 is present on the film 2 into a plurality of areas, and sequentially setting the line sensor camera for each area. This is done by scanning.

  That is, for example, when the chip group 3 is divided into four areas, first, an image of the first area 3A of the chip group 3 is acquired by scanning the line sensor camera along the arrow A in FIG. The image of the second region 3B of the chip group 3 is acquired by scanning the line sensor camera along the arrow B, and the third region 3C of the chip group 3 is scanned by scanning the line sensor camera along the arrow C. The image of the fourth region 3D of the chip group 3 is obtained by scanning the line sensor camera along the arrow D, and a relatively low resolution image is obtained for the entire region of the chip group 3. Can be acquired.

  Here, FIG. 5 is a plan view showing the arrangement of the LED chips 7 constituting the chip group 3. In FIG. 5, only some LED chips 7 are denoted by reference numerals. In FIG. 5, only a part of the chip group 3 is shown for simplicity.

  The LED chips 7 are ideally arranged in an orderly and horizontal arrangement on the film 2 as shown in FIG.

  When the LED chips 7 are arranged in an orderly manner as described above, for example, the first region 3A (FIG. 3) and the second region 3B (FIG. 3) are divided along the boundary line 6 shown in FIGS. Then, each region is sequentially imaged at a relatively low resolution, and similarly, the images of the third region 3C and the fourth region 3D may be acquired sequentially.

  By the way, as a result of stretching the film 2 to separate the LED chips 7, the arrangement of the LED chips 7 may be disturbed.

  For example, FIG. 6 is a plan view showing a case where the arrangement of the LED chips 7 is tilted.

  In the case of the example shown in FIG. 6, the first region 3A (see FIG. 3) and the second region 3B (see FIG. 3) are divided along the boundary line 6 shown in FIG. 6, and each region is imaged separately. In this case, since the arrangement of the LED chips 7 is tilted, it becomes difficult to inspect the defects.

  Further, it is preferable that each LED chip 7 determines its coordinates based on the acquired low-resolution image and manages in the subsequent process. However, if the arrangement of the LED chips 7 is inclined, the determination of the coordinates is performed. Is inconvenient.

  Therefore, in the present embodiment, the inclination of the array of the chip group 3 is determined by the image recognition process based on the image of the first area 3A acquired first among the four areas 3A to 3D shown in FIG.

  For example, when the arrangement of the chip group 3 is tilted as shown in FIG. 6, this can be determined based on the image of the first region 3A, so that the film 2 is formed in a ring shape as shown in FIG. By rotating the entire frame 4 (FIG. 2) in the direction of arrow E, the inclination of the array of the chip group 3 can be corrected.

  Here, the ring-shaped frame-like body 4 is horizontally arranged on a stage that can rotate on a horizontal plane.

  Therefore, by rotating this stage by a driving means such as a motor, the film 2 can be rotated together with the ring-shaped frame 4 in the direction of arrow E to correct the posture of the film 2.

  After correcting the posture of the film 2 in this way, the image of the first region 3A is acquired again at a relatively low resolution, and subsequently, the images of the second to fourth regions 3B to 3D are also relatively relatively similar. Obtain sequentially in low resolution.

  Incidentally, FIG. 5 shows an example in which the LED chips 7 are regularly arranged vertically and horizontally and there is no inclination, and FIG. 6 shows an example in which the LED chips 7 are regularly arranged although the chip group 3 is inclined. However, in practice, the LED chips 7 on the film 2 are often irregularly arranged.

  FIG. 8 is a plan view showing only one row irregularly arranged among the rows of LED chips 7 constituting the chip group 3.

  In the row of LED chips 7 shown in FIG. 8, the LED chip 7A is shifted from the row.

  Therefore, for example, the first region 3A (FIG. 3) and the second region 3B (FIG. 3) are divided along the boundary line 6 shown in FIG. 8, and each region is sequentially imaged at a relatively low resolution. Then, the image of the LED chip 7A cannot be captured in one image.

  Here, it is necessary to capture the entire image of each LED chip 7 in one image. This is because, when images of a part of the LED chip 7 are acquired in a cut by a plurality of scans and the acquired images are synthesized later, the image is affected by the error of the mechanical operation of the line sensor camera. This is because an error occurs and inconvenience occurs in defect inspection.

  Therefore, as shown in FIG. 8, when there are LED chips 7A arranged so as to straddle the boundary line 6, these LED chips 7A are obtained, for example, in the imaging of the first region 3A (FIG. 3). It is determined that there is no image (unprocessed), and it is included in the image of the second region 3B (FIG. 3) obtained by the next imaging.

  That is, after acquiring the image of the first area 3A (FIG. 3) with the boundary line 6 shown in FIG. 8 as the boundary, the image of the second area 3B (FIG. 3) with the boundary line 8 shown in FIG. To get.

  That is, for example, when there is an LED chip 7A partially protruding from the image of the first area 3A (FIG. 3), the LED chip 7A is included in the image of the second area 3B (FIG. 3) to be imaged next. The processing for setting the range of the second region 3B is automatically performed by computer processing so that the whole image is included.

  As described above, by acquiring images of a plurality of regions (for example, four regions of the first to fourth regions 3A to 3D), relatively low resolution images can be obtained for all the regions of the chip group 3. .

  Subsequently, processing for determining the coordinates of each LED chip 7 is automatically performed by computer processing based on the relatively low resolution image obtained as described above.

  That is, for example, by obtaining the geometric center of gravity of each LED chip 7 from the image, the position of the center of gravity can be used as the coordinates of each LED chip 7. Note that a technique for obtaining a geometric center of gravity of an object from an image and converting it to a coordinate is a well-known technique in image analysis technology.

  Further, as will be described below, a process for determining a defect of the LED chip 7 from an image having a relatively low resolution is automatically performed by image analysis.

  FIG. 9 is a diagram illustrating an example of a relatively low resolution image obtained by actual imaging.

  It can be determined that the LED chip 7B and the like shown in FIG. 9 are clearly cracked only by analyzing a relatively low resolution image, and can be determined to be defective.

  Further, the LED chip 7C and the like shown in FIG. 9 can be clearly determined that the two LED chips 7 are overlapped only by analyzing a relatively low resolution image, and can be determined to be defective. .

  Further, the LED chip 7D and the like shown in FIG. 9 clearly show that the two LED chips 7 remain unseparated only by analyzing a relatively low resolution image, and are determined to be defective. can do.

  As described above, the coordinates of the LED chip 7 that is clearly found to be a defective product from the relatively low resolution image are managed, and the inspection of the defect using the high resolution image is omitted.

  Although illustration is omitted, for example, when all of the LED chips 7 existing in a partial region of the chip group 3 are clearly defective from a relatively low resolution image, In that region, a relatively high resolution image is not acquired in the first place.

  That is, by analyzing a relatively low resolution image, it is possible to determine an area on the surface of the film 2 that does not require acquisition of a relatively high resolution image and to determine that acquisition of a relatively high resolution image is unnecessary. Only the image of the semiconductor chip in the region excluding the formed region may be acquired with a relatively high resolution.

  As for acquisition of images at a relatively high resolution, the chip group 3 is divided into a plurality of regions, and images are acquired sequentially for each of the divided regions, as in the case of acquiring images at a relatively low resolution. I do.

  Further, regarding acquisition of an image with a relatively high resolution, if a part of the LED chip 7A protrudes from the image as shown in FIG. 8, the next imaging includes the entire LED chip 7A. An imaging range is set so that an image can be taken.

  Further, the wafer is not limited to a rectangular shape as shown in FIG.

  FIG. 10 is a plan view showing a chip group 3 obtained by separating a circular wafer into individual LED chips 7.

  Also in this case, as shown in FIG. 4, for example, images of four regions of the first to fourth regions 3A to 3D are sequentially acquired at a relatively low resolution, and the LED chip is based on the obtained image. 7 defect determination and determination of an image acquisition area with relatively high resolution.

  Here, as shown in FIG. 10, there are blank areas 11 in which the LED chips 7 do not exist around the circular chip group 3, and therefore these blank areas 11 also have an image acquisition area with a relatively high resolution. As a result, loss occurs.

  Therefore, when acquiring an image of the circular chip group 3 as shown in FIG. 10, based on the determination based on the image acquired at a relatively low resolution, an image should not be acquired for the blank area 11 as much as possible. The computer automatically performs a process of determining a plan for an image acquisition area with a relatively high resolution.

  That is, for example, areas that do not include the blank area 11 as much as possible, such as the fifth area 3E, the sixth area 3F, the seventh area 3G,... Shown in FIG. By acquiring this image, it is not necessary to acquire a high-resolution image as much as possible in the blank area 11.

  According to the first embodiment as described above, for example, when it is possible to recognize that the LED chip 7 has an obvious defect only by using a low-resolution image, the LED chip 7 has a high-resolution image. Therefore, the time required for inspecting the defect of the LED chip 7 can be shortened.

  Further, when the blank area 11 is present around the chip group 3, it is not necessary to obtain a high-resolution image as much as possible for the blank area 11, and the time required for the inspection of the defect of the LED chip 7 can be shortened. .

  In the above embodiment, the method for acquiring the image of the LED chip 7 has been described. However, the present invention is not limited to this example, and can be similarly applied to a method for acquiring an image of another semiconductor chip.

It is a top view for demonstrating the process of isolate | separating a wafer into a LED chip. It is a perspective view which shows the ring-shaped frame-like body in which the film of the state in which the chip group which consists of many isolate | separated LED chips was affixed was affixed. It is a top view of a LED chip. It is a top view which shows the chip group of the state affixed on the film. It is a top view which shows the arrangement | sequence of the LED chip which comprises a chip group. It is a top view which shows the case where the arrangement | sequence of an LED chip inclines. It is a top view which shows the state which corrected the inclination of the arrangement | sequence of a LED chip. It is a top view which shows only one row | line | column arrange | positioned irregularly among the row | line | columns of the LED chip which comprises a chip group. It is a figure which shows an example of the comparatively low resolution image obtained by actual imaging. It is a top view which shows the chip group obtained by isolate | separating a circular-shaped wafer into each LED chip.

Explanation of symbols

1 wafer (semiconductor wafer)
2 Film 3 Chip group 7 LED chip (semiconductor chip)

Claims (5)

In a method of acquiring an image for inspection of a semiconductor chip in a state where the semiconductor wafer is divided into a plurality of semiconductor chips by stretching a film to which the semiconductor wafer is attached.
A first step of acquiring an image of a semiconductor chip on the film at a relatively low resolution;
A second step of acquiring an image of the semiconductor chip on the film at a relatively high resolution;
A third step of determining an acquisition area of the image by the second step based on the image acquired by the first step;
An image acquisition method for a semiconductor chip, comprising: The first process includes a plurality of imaging processes in which a region where a chip group including a plurality of semiconductor chips is present on the film is divided into a plurality of regions, and an image is sequentially acquired for each of the divided regions. Consists of
Performing a fourth step of determining the inclination of the arrangement of the semiconductor chips in the chip group based on the image obtained by the first imaging step in the first step;
If it is determined in the fourth process that the arrangement of the semiconductor chips is tilted, the first process is performed after the fifth process of correcting the posture of the film to correct the tilt. 2. The method of acquiring an image of a semiconductor chip according to claim 1, wherein the process is started again from the beginning. The first process includes a plurality of imaging processes in which a region where a chip group including a plurality of semiconductor chips is present on the film is divided into a plurality of regions, and an image is sequentially acquired for each of the divided regions. Consists of
When a semiconductor chip partially protruding from the image is present in the image obtained by one imaging process constituting the first process, the imaging process performed after the one imaging process The imaging range in the imaging process performed next to the one imaging process is set so that an image including the whole of the protruding semiconductor chip can be captured. An image acquisition method for the semiconductor chip described. The second process includes a plurality of imaging processes in which a region where a chip group including a plurality of semiconductor chips is present on the film is divided into a plurality of regions, and an image is sequentially acquired for each of the divided regions. Consists of
If there is a semiconductor chip partially protruding from the image obtained by one imaging process constituting the second process, the imaging process performed after the first imaging process The imaging range in the imaging process performed after the one imaging process is set so that an image including the entire protruding semiconductor chip can be captured. The image acquisition method of the semiconductor chip as described in any one of Claims.   5. The semiconductor chip according to claim 1, further comprising a sixth process of determining coordinates of each semiconductor chip based on the image acquired by the first process. 6. Image acquisition method.

Images