JP2010034133A - Crack detecting device for polycrystalline silicon wafer - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To detect an internal crack of a polycrystalline silicon wafer by calibrating differences between patterns of a wafer top-surface image by reflected light and a wafer transmission image by transmitted light. <P>SOLUTION: Crystal grains are different in size between wafer top-surface image data obtained from reflected light of the polycrystalline silicon wafer 2 and wafer transmission image data obtained from infrared transmitted light of the polycrystalline silicon wafer 2. Consequently, a crack suspicious pixel which has a remarkable lightness difference due to an internal crack and a remarkable lightness difference due to a shift of a boundary of crystal grains etc., is detected from the difference values between the two image data. Adjacent crack suspicious pixels are linked and the crack is detected from a distribution state of the crack suspicious pixels. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a crack detection apparatus for a crystalline semiconductor, and more particularly to a crack detection apparatus for a polycrystalline silicon wafer.

  A silicon crystal semiconductor is widely used as a photoelectric conversion layer used in a solar cell panel. Monocrystalline and polycrystalline semiconductors are used for this silicon crystal semiconductor. Single crystal silicon semiconductors have high photoelectric conversion efficiencies, and polycrystalline silicon semiconductors have respective advantages in that manufacturing costs can be suppressed. However, in both types of silicon semiconductors, if cracks are inherent in the crystal, the cracks may expand due to some impact and eventually the whole wafer may be cracked. This crack is generated inside the silicon wafer during the manufacturing process. For this reason, it is very important to detect internal cracks in the silicon wafer.

As a method for detecting this internal crack, there is a method using the property that silicon transmits infrared rays. If the single crystal silicon wafer is irradiated with infrared rays, the infrared rays are scattered or absorbed by internal cracks in the single crystal silicon wafer, so that the infrared transmitted light that has passed through the cracks is transmitted through the single crystal silicon wafer other than the cracks. Infrared intensity is weaker than light. Thus, cracks are observed as dark portions in the infrared image transmitted through the single crystal silicon wafer. In this way, internal cracks in the single crystal silicon wafer can be detected. Patent Document 1 discloses an example of such a semiconductor wafer inspection apparatus.
JP 2006-351669 A

  However, in the case of a polycrystalline silicon wafer, it is more difficult to detect internal cracks using infrared transmitted light than in the case of a single crystal silicon wafer. This is because a polycrystalline silicon wafer is composed of a collection of crystal grains, and each crystal grain has a different plane orientation, so that the intensity of transmitted light transmitted through each crystal grain varies. Thus, bright and dark image data is observed for each crystal grain in the infrared image by the transmitted light of the polycrystalline silicon wafer. The crystal grains have various shapes, some of which are linear. In such a linear crystal grain, or a crystal grain boundary that is a boundary between crystal grains, it is difficult to distinguish from a crack.

  The present invention has been made to solve such a problem, and distinguishes cracks from polycrystalline silicon wafers from crystal grains, and detects cracks present in polycrystalline silicon wafers. An object is to provide a detection device.

  As a result of intensive studies, the inventors of the present application have obtained the following knowledge. In order to erase the pattern of crystal grains having the respective plane orientations, white light is irradiated on the polycrystalline silicon wafer, and the surface image of the polycrystalline silicon wafer obtained by this reflected light is transmitted by infrared transmitted light. If infrared images are subtracted, cracks can be detected.

  However, the patterns on the front and back surfaces of the polycrystalline silicon wafer are slightly different. In particular, if a polycrystalline silicon wafer is imaged with a high-resolution camera, a difference in pattern is detected. In addition, since the infrared image by the infrared transmitted light is an infrared image affected by the inside of the polycrystalline silicon wafer, the size of the crystal grains is different from the surface image of the polycrystalline silicon wafer obtained by the reflected light. Different images are obtained. Thus, simply comparing the surface image of the polycrystalline silicon wafer and the infrared transmission image erroneously detects, for example, a crystal grain boundary as a crack. In this regard, the inventor of the present application pays attention to the fact that the pixel distribution state is different between the crystal grain boundary and the crack in the pixel suspected of being cracked obtained by subtracting the reflected light image and the infrared transmission image. Invented a crack detecting device for a polycrystalline silicon wafer for detecting a crack in the polycrystalline silicon wafer.

In order to achieve such an object, the present invention has the following configuration.
That is, the invention described in claim 1 is a crack detection apparatus for a polycrystalline silicon wafer, wherein a light irradiation means for irradiating one surface of the polycrystalline silicon wafer with white light or monochromatic light, and the polycrystalline First imaging means for imaging reflected light of the surface of the silicon wafer that has been irradiated with the white light or monochromatic light, infrared irradiation means for irradiating the other surface of the polycrystalline silicon wafer with infrared light, and the polycrystalline silicon A second imaging means for imaging the infrared transmitted light transmitted from the other surface of the wafer to the one surface; wafer surface image data obtained from the first imaging means; and a wafer obtained from the second imaging means. A first calculation unit for differentiating the transmission image data for each pixel; a threshold calculation unit for calculating a threshold value for determining a pixel suspected of cracking from the difference value obtained by the first calculation unit; Based on the threshold value calculated by the calculation unit, the binarization processing unit for determining a pixel suspected of cracking from the difference value obtained by the first calculation unit, and the binarization processing unit And a crack discriminating unit for discriminating a crack from a distribution state of pixels suspected of being cracked.

  According to the above configuration, the wafer surface image data obtained from the reflected light of the polycrystalline silicon wafer and the wafer transmission image data obtained from the transmitted light of the polycrystalline silicon wafer are subtracted, and a pixel suspected of cracking from the difference value A threshold for discriminating is calculated, and a pixel suspected of cracking is discriminated based on this threshold. A simple difference due to the difference in crystal grain pattern between wafer surface image data and wafer transmission image data results in a brightness difference due to misalignment of crystal grain boundaries in addition to cracks. A pixel that is suspected of being cracked is determined by calculation. Furthermore, it is possible to more accurately determine a crack by determining whether or not it is a crack from the distribution state of pixels suspected of being cracked. Thus, when the crystal grain pattern of the wafer surface image data obtained from the reflected light of the polycrystalline silicon wafer does not match the crystal grain pattern of the wafer transmission image data obtained from the transmitted light of the polycrystalline silicon wafer. Can also detect cracks.

  The invention according to claim 2 is a crack detection apparatus for a polycrystalline silicon wafer, wherein one surface of the polycrystalline silicon wafer is irradiated with white light or monochromatic light, and the polycrystalline silicon wafer. First imaging means for imaging reflected light of the surface irradiated with the white light or monochromatic light, infrared irradiation means for irradiating infrared light on the other surface of the polycrystalline silicon wafer, and the polycrystalline silicon wafer. A second imaging means for imaging the infrared transmitted light that is transmitted from the other surface to the one surface; wafer surface image data obtained from the first imaging means; and a wafer transmission image obtained from the second imaging means. A first arithmetic unit that compares data for each pixel; and a pixel having a larger brightness difference when the first arithmetic unit performs difference between the wafer surface image data and the wafer transmission image data. A filter processing unit for emphasizing a difference value obtained by the first calculation unit, a second calculation unit for subtracting the image data processed by the filter processing unit and the wafer transmission image data for each pixel, and A threshold value calculation unit that calculates a threshold value for discriminating a pixel suspected of cracking from the difference value obtained by the second calculation unit, and a threshold value calculated by the threshold value calculation unit. A binarization processing unit that discriminates a pixel suspected of cracking from the obtained difference value, and a crack discrimination unit that discriminates a crack from the distribution state of pixels suspected of cracking discriminated by the binarization processing unit It is characterized by comprising.

  According to the above configuration, the wafer surface image data obtained from the reflected light of the polycrystalline silicon wafer and the wafer transmission image data obtained from the transmitted light of the polycrystalline silicon wafer are differentiated, and the difference value is enhanced according to the brightness difference. To enhance the contrast. A threshold for discriminating a pixel suspected of cracking is calculated from the difference value with enhanced contrast, and a pixel suspected of cracking is discriminated based on this threshold. A simple difference due to the difference in crystal grain pattern between wafer surface image data and wafer transmission image data results in a brightness difference due to misalignment of crystal grain boundaries in addition to cracks. A pixel that is suspected of being cracked is determined by calculation. Furthermore, it is possible to more accurately determine a crack by determining whether or not it is a crack from the distribution state of pixels suspected of being cracked. Thus, when the crystal grain pattern of the wafer surface image data obtained from the reflected light of the polycrystalline silicon wafer does not match the crystal grain pattern of the wafer transmission image data obtained from the transmitted light of the polycrystalline silicon wafer. Can also detect cracks.

  According to a third aspect of the present invention, in the crack detection apparatus for a polycrystalline silicon wafer according to the first or second aspect, the crack discriminating unit discriminates a crack by pattern matching with a predetermined template. Features.

  According to the above configuration, the pattern can be easily identified from the distribution state of the pixel suspected of being cracked by pattern matching between the pixel suspected of being cracked and the template.

  According to a fourth aspect of the present invention, in the crack detection apparatus for a polycrystalline silicon wafer according to the first or second aspect, the crack discriminating unit discriminates a crack based on a dispersion density of pixels suspected of being cracked. And

  According to the above configuration, the crack can be easily determined from the distribution state of the pixel suspected of being cracked by using the dispersion density of the pixel suspected of being cracked.

  According to a fifth aspect of the present invention, in the crack detection apparatus for a polycrystalline silicon wafer according to any one of the first to fourth aspects, the polycrystalline silicon wafer is checked for cracks based on the determination of the crack determination unit. And a wafer storage section for storing the polycrystalline silicon wafer after sorting.

  According to the above configuration, the polycrystalline silicon wafer can be sorted and stored depending on the presence or absence of cracks based on the discrimination of the crack discriminating section. Therefore, the polycrystalline silicon wafer having internal cracks after the crack inspection of the polycrystalline silicon wafer. And a polycrystalline silicon wafer having no internal crack can be easily distinguished.

  According to a sixth aspect of the present invention, in the polycrystalline silicon wafer crack detection device according to any one of the first to fifth aspects, the first imaging means and the second imaging means are line sensor cameras. It is characterized by.

  According to the above configuration, since the first imaging unit and the second imaging unit are line sensor cameras, it is possible to capture an image of the polycrystalline silicon wafer by reflected light and transmitted light while transporting the polycrystalline silicon wafer. Therefore, the crack inspection time can be shortened.

  According to the crack detection apparatus for a polycrystalline silicon wafer according to the present invention, even when the crystal grain pattern of the wafer surface image data does not match the crystal grain pattern of the wafer transmission image data, It is possible to provide a crack detection apparatus for a polycrystalline silicon wafer that distinguishes crystal grains from each other and detects a crack existing inside the polycrystalline silicon wafer.

Embodiments of the present invention will be described below with reference to the drawings.
FIG. 1 is a schematic diagram of a polycrystalline silicon wafer crack detection apparatus according to an embodiment, and FIG. 2 is a block diagram of an image processing unit according to the embodiment.

With reference to FIG. 1, the structure of the polycrystalline silicon wafer crack detection apparatus will be described.
The crack detection apparatus 1 for a polycrystalline silicon wafer includes a wafer transport unit 3 that transports a polycrystalline silicon wafer 2, a white light source 4 that emits white light from an oblique upper portion of the transported polycrystalline silicon wafer 2, and a white light source. The first line sensor camera 5 for imaging the upper surface of the polycrystalline silicon wafer 2 irradiated with white light by 4, the infrared tube 6 for irradiating infrared light from the lower part of the polycrystalline silicon wafer 2, and the polycrystalline silicon wafer 2 The image which detects the crack of the polycrystalline silicon wafer 2 with the 2nd line sensor camera 7 which images the transmitted infrared rays, the image data imaged with the 1st line sensor camera 5, and the image data imaged with the 2nd line sensor camera 7 The host computer 9 having the processing unit 8 and the presence or absence of cracks by the host computer 9 are used to determine the polycrystalline silicon wafer. It is composed 2 in the wafer storage section 10 for acceptance box and those with and without some of the cracks.

  In addition, the polycrystalline silicon wafer crack detection apparatus 1 includes a transfer driving unit 11 that drives the wafer transfer unit 3, and a first monitor 12 that displays an image captured by the first line sensor camera 5. A second monitor 13 for displaying an image picked up by the second line sensor camera 7, a third monitor 14 for displaying cracks detected by the host computer 9, and a large number of wafers from the wafer transfer section 3 to the wafer storage section 10. It includes a storage transport mechanism 15 that transports the crystalline silicon wafer 2 and a storage drive section 16 that slides the wafer storage section 10, and further controls the transport drive section 11, the storage transport mechanism 15, and the storage drive section 16. In the host computer 9.

  The polycrystalline silicon wafer 2 is produced by thinly slicing an ingot produced by a casting method or the like. In this embodiment, the polycrystalline silicon wafer 2 has a size of 156 mm square and a thickness of about 180 μm. As described above, since the polycrystalline silicon wafer 2 is thinner than the size in the planar direction, it has two surfaces, a front surface and a back surface.

  A plurality of rollers 18 are arranged in a row in the wafer transfer section 3 for transferring the polycrystalline silicon wafer 2, and the ends of the rollers 18 are connected to each other by a belt 19. As the roller 18 connected to the transport driving unit 11 rotates, the belt 19 rotates, and thereby the other rollers 18 also rotate and transport the polycrystalline silicon wafer 2. The rotation and rotation speed and rotation speed of the roller 18 are controlled by the transport driving unit 11. The polycrystalline silicon wafer 2 is placed on the wafer transport unit 3 from a wafer supply unit (not shown), and the polycrystalline silicon wafer 2 is transported from the left in FIG. 1 to the right (forward direction).

  The white light source 4 irradiates light to the polycrystalline silicon wafer 2 to be transported. In this embodiment, a metal halide lamp which is a kind of mercury lamp is used. The first line sensor camera 5 images the surface of the polycrystalline silicon wafer 2 irradiated with light, and the wafer surface image data, which is image data of the surface of the polycrystalline silicon wafer 2, is displayed on the first monitor 12 and the image processing unit 8. Send to. That is, the first line sensor camera 5 captures the surface image data by imaging the surface of the polycrystalline silicon wafer 2 obtained by the reflected white light. The first line sensor camera 5 uses a CCD line sensor, a CMOS line sensor, or the like. The white light source 4 is not limited to a metal halide lamp, and may be a fluorescent tube that emits white light or monochromatic light, or a lamp that emits other white light or monochromatic light. The white light source 4 corresponds to the light irradiation means in the present invention, and the first line sensor camera 5 corresponds to the first imaging means in the present invention.

The infrared tube 6 is arranged between the rollers 18 arranged in a row so that infrared rays are irradiated from the back surface of the polycrystalline silicon wafer 2. Thus, when the polycrystalline silicon wafer 2 is transported on the infrared tube 6, infrared light is irradiated to the back surface of the polycrystalline silicon wafer 2 and passes through the polycrystalline silicon wafer 2. By imaging the transmitted infrared rays with the second line sensor camera 7, an infrared transmission image of the polycrystalline silicon wafer 2 is acquired. The infrared transmission image data is sent to the second monitor 13 and the image processing unit 8. In the present embodiment, an infrared ray with an infrared wavelength of 900 nm or more is used. Similar to the first line sensor camera 5, the second line sensor camera 7 uses a CCD line sensor, a CMOS line sensor, or the like.
The infrared tube 6 corresponds to the infrared irradiation means in the present invention, and the second line sensor camera 7 corresponds to the second imaging means in the present invention.

  In this way, with respect to the polycrystalline silicon wafer 2 transported in the same direction, one picks up the reflected light obtained by irradiating the surface of the polycrystalline silicon wafer 2 with white light, and the other picks up the polycrystalline silicon wafer 2. Two images of the same position on the polycrystalline silicon wafer 2 are obtained by imaging the transmitted light irradiated with infrared rays from the back surface of the substrate.

  In addition, the first line sensor camera 5 and the second line sensor camera 7 employ about 4000 pixels, and both are line sensor cameras that react in the visible light region. It is more preferable that the sensitivity peak of the line sensor camera is shifted to the infrared region side. In the present application, since the image is captured as 8-bit image data per pixel, the brightness is handled as data divided into 256 gradations.

  The host computer 9 includes an image processing unit 8 that detects internal cracks in the polycrystalline silicon wafer 2 based on image data acquired by the first line sensor camera 5 and the second line sensor camera 7, a transport drive unit 11, and a storage unit. And a control unit 17 that controls the operation timing of the transport mechanism 15 and the storage drive unit 16. The control unit 17 may be configured as a programmable logic controller (PLC) separately from the host computer 9.

  The image processing unit 8 detects internal cracks in the polycrystalline silicon wafer 2 from the wafer surface image data obtained from the first line sensor camera 5 and the infrared transmission image data obtained from the second line sensor camera 7. As shown in FIG. 2, the image processing unit 8 stores a first memory unit 20 for storing wafer surface image data obtained from the first line sensor camera 5 and an infrared transmission image data obtained from the second line sensor camera 7. The second memory unit 21, the first calculation unit 22 that performs a difference calculation on the wafer surface image data and the infrared transmission image data sent from the first memory unit 20 and the second memory unit 21, and the difference image data A threshold calculation unit 23 that calculates a threshold for binarization, a binarization processing unit 24 that binarizes image data that has been differentiated by the threshold calculated by the threshold calculation unit 23, and a binarization process A labeling unit 25 for connecting image data and a crack determining unit 26 for determining a crack from the connected image data. The first memory unit 20 and the second memory unit 21 constitute a memory unit 27. The crack determined by the crack determination unit 26 is displayed on the third monitor 14, and a crack determination signal is sent to the control unit 17.

  The polycrystalline silicon wafer 2 determined by the image processing unit 8 for the presence or absence of internal cracks is transferred to the wafer storage unit 10 by the storage transfer mechanism 15. The storage and transport mechanism 15 attracts the polycrystalline silicon wafer 2 by the wafer holding unit 28 in accordance with an operation signal from the control unit 17, horizontally moves along the arm 29, and transports it to the stage 30 of the wafer storage unit 10. The wafer holding unit 28 may transport the polycrystalline silicon wafer 2 to the stage 30 by other methods besides the adsorption.

  The wafer storage unit 10 has two stages 30 on the pedestal 31 for storing the polycrystalline silicon wafer 2, and stores the polycrystalline silicon wafer 2 separately depending on the presence or absence of internal cracks. The control unit 17 in the host computer 9 sends an operation signal to the storage drive unit 16 depending on the presence or absence of an internal crack in the polycrystalline silicon wafer 2, and the storage drive unit 16 is arranged so as to be movable on the two rails 32. By driving the pedestal 31, the two stages 30 move, and the polycrystalline silicon wafer 2 having internal cracks and the polycrystalline silicon wafer 2 having no internal cracks are stored separately. Further, the polycrystalline silicon wafer 2 is separately stored by moving the storage stage 30 along the rail 32 via the pedestal 31, but this is not a limitation, and for example, the storage stage 30 remains fixed, The storage and transport mechanism 15 may move in the horizontal direction and store the polycrystalline silicon wafers 2 separately.

  Next, a sequence for detecting internal cracks in the polycrystalline silicon wafer 2 will be described with reference to FIGS. FIG. 3 is a flowchart showing a series of sequences for detecting cracks in the polycrystalline silicon wafer according to the embodiment. FIGS. 4 and 5 are explanatory diagrams for capturing an image of the polycrystalline silicon wafer 2 according to the embodiment. 6 to 8 are captured images of the polycrystalline silicon wafer 2 according to the embodiment.

(Step S1) Acquisition of Wafer Surface Image Data As shown in FIG. 4, when the polycrystalline silicon wafer 2 is transported in the forward direction and comes directly below the first line sensor camera 5, white light is irradiated by the white light source 4. The first line sensor camera 5 captures the wafer surface image of the polycrystalline silicon wafer 2. The wafer surface image data shown in FIG. 6 is displayed on the first monitor 12 and stored in the first memory unit 20 in the image processing unit 8.

(Step S2) Infrared Transmission Image Data Acquisition As shown in FIG. 5, when the polycrystalline silicon wafer 2 is further transported in the forward direction and the polycrystalline silicon wafer 2 covers the tube of the infrared tube 6, irradiation from the infrared tube 6 occurs. The infrared rays thus transmitted are transmitted through the polycrystalline silicon wafer 2 and received by the second line sensor camera 7. In this way, infrared transmission image data of the polycrystalline silicon wafer 2 shown in FIG. 7 can be obtained. A star-shaped crack is observed in the center of FIG. The infrared transmission image data is displayed on the second monitor 13 and stored in the second memory unit 21 in the image processing unit 8.

(Step S3) Difference Calculation The first calculation unit 22 performs a difference calculation for each corresponding pixel between the wafer surface image data and the infrared transmission image data stored in the first memory unit 20 and the second memory unit 21, respectively. The difference image data (difference value) is obtained. The difference image data is sent to the threshold value calculation unit 23.

  Here, FIG. 8 is a superposition of FIG. 6 which is wafer surface image data and FIG. 7 which is infrared transmission image data. As shown in FIG. 8, since the crystal grain boundary that is the boundary between crystal grains does not necessarily match between the wafer surface image data and the infrared transmission image data, the pattern of the crystal grains can be completely canceled even if the difference calculation is performed. Can not. Therefore, it is necessary to discriminate between the difference image data due to the crystal grain boundary mismatch and the difference image data due to the crack. As described above, the difference in brightness between the wafer surface image data and the infrared transmission image data is remarkably detected as a difference in brightness due to internal cracks and a difference in brightness caused by deviations in crystal grain boundaries.

(Step S4) Threshold Value Calculation Next, a threshold value calculating unit 23 calculates a threshold value for discriminating between the crystal grain pattern and the suspected crack pattern from the difference image data. As this threshold value calculation method, an area division average value method is illustrated with reference to FIG. In FIG. 9, in the difference image data, an area for nine pixels is divided as one area, and divided areas PB1 to PB9 are illustrated. And the average value of the difference value of each pixel in one area, for example, PB1, is calculated, and this is used as the threshold value in the PB1 area. In this way, the respective threshold values for the divided areas PB1 to PB9 are calculated. Thus, different threshold values are calculated depending on the divided areas. In this embodiment, the area for nine pixels is one divided area, but this is for the sake of simplicity. Actually, an area of about 50 pixels is used as one divided area, but the number of pixels constituting one divided area may be set as appropriate. In addition, the number of divided areas into which the difference image data is divided may be appropriately set according to the number of pixels constituting one divided area.

(Step S5) Binarization processing The threshold value of each divided region calculated by the threshold value calculation unit 23 is compared with the difference value of each pixel constituting the corresponding divided region, and a difference value larger than the threshold value is calculated. The corresponding pixel is assumed to be a crack suspicious pixel.

(Step S6) Labeling If there is a pixel that is also identified as a suspected crack pixel in the pixel area in the vicinity of 8, centering on the suspected cracking pixel determined by the binarization processing unit, pixels that are determined as suspected crack pixels A labeling process is performed to detect a single colony. Referring to FIG. 10, each cell indicates a pixel, pixels indicated by diagonal lines are pixels that are determined to be suspected cracks, and pixel colonies Co <b> 1 to Co <b> 3 that connect these pixels are detected. . At this time, a pixel colony with a small number of connected pixels may be determined as noise and may not be labeled. In the embodiment, if the number of connected pixels is 2 pixels or less, it is determined as noise and the labeling process is not performed. The number of pixels serving as a reference for whether or not to perform the labeling process is not limited to two pixels, and an optimal number of pixels may be set as appropriate.

(Step S7) Crack Determination Next, it is determined whether the pixel colony subjected to the labeling process is due to a crack or a crystal grain boundary. In this embodiment, a crack discrimination method by a pattern matching method is exemplified. As a rule of thumb, the pixel colonies detected from the cracks and the pixel colonies detected from the crystal grain boundaries differ in the distribution state of the connected pixels. Referring to FIG. 10, Co1 and Co2 are pixel colonies detected from the crystal grain boundaries, while Co3 is a pixel colony detected from cracks.

  The pixel colonies caused by cracks are binarized by the deviation of the crystal grain boundaries, while the suspected cracked pixels after binarization are aggregated in a relatively narrow area and have linearity. Later suspected crack pixels have variations and flatness. Focusing on the difference between the two distribution states, it is possible to determine whether the pixel is a crack or a crystal grain boundary based on the pixel distribution state of the pixel colony.

  In this embodiment, a houndstooth (checkered flag) template images PM1 and PM2 shown in FIGS. 11 and 12 are used as objects for pattern matching. Each of these two types of template images is pattern-matched with a pixel colony. At this time, as shown in FIG. 13, a template image having the same size as the circumscribed rectangle of the pixel colony is used. And, the higher correlation with each template image is set as the correlation value R of the colony, and if this correlation value R is smaller than a threshold value determined in advance by an empirical rule, the rate of aggregation of pixels in the colony is large. Detect as a crack. If the correlation value R is larger than a threshold value determined in advance based on empirical rules, the ratio of pixel variation in the colony is large, so that it is recognized as a deviation of the crystal grain boundary. The following formula is illustrated as calculation of correlation value R.

ｆij：画素コロニー画素値
ｔij：テンプレート画像画素値

fij: Pixel colony pixel value tij: Template image pixel value

  The polycrystalline silicon wafer 2 in which the crack is detected displays the infrared transmission image on the third monitor and notifies the operator by displaying the determined crack in red.

  In addition, the polycrystalline silicon wafer 2 in which the crack is detected is sent to the storage drive unit 16 from the control unit 17 so that the pedestal 31 slides along the rail 32 and has a crack. Is placed on the extension of the wafer transfer unit 3 and is stored by the storage transfer mechanism 15. If a crack is also detected in the next transferred polycrystalline silicon wafer 2, the wafer storage unit 10 stores the polycrystalline silicon wafer 2 in the stage 30 without moving. If no crack is detected in the polycrystalline silicon wafer 2, an operation signal is sent from the control unit 17 to the storage drive unit 16, and the pedestal 31 slides along the rail 32 to store the polycrystalline semiconductor wafer without cracks. The stage 30 is placed on the extension of the wafer transfer unit 3 and is stored by the storage transfer mechanism 15. Thus, the polycrystalline silicon wafer 2 is sorted and stored depending on the presence or absence of internal cracks.

  According to the crack detection apparatus for a polycrystalline silicon wafer according to the present embodiment, the pattern due to the grain boundary is canceled by subtracting the wafer surface image data by reflected light and the infrared transmission image data by infrared transmission light, Since it is possible to determine whether a pattern such as a crystal grain boundary that cannot be canceled out due to a difference in crystal grain size is a crack, it is possible to accurately detect an internal crack existing in the polycrystalline silicon wafer.

  In addition, since it is not necessary to stop the polycrystalline silicon wafer by using a line sensor camera as an imaging means, the presence or absence of internal cracks is determined while the polycrystalline silicon wafer is being transferred, and polycrystalline silicon is determined by the presence or absence of internal cracks. Wafers can be stored separately. As described above, since the internal cracks of the polycrystalline silicon wafer can be discriminated by the continuous flow operation, the crack inspection time can be shortened.

  In addition, since the crack is discriminated based on the distribution state of the suspicious pixels, the brightness for detecting infrared rays may be lower than that of a conventional crack detection device, and the line sensor camera uses a normal visible light region as the light receiving range. It can be used. As a result, the cost performance is superior compared to the case of using an infrared camera.

  The present invention is not limited to the above embodiment, and can be modified as follows.

  (1) In the above embodiment, the first calculation unit 22 calculates the difference between the wafer surface image data and the infrared image data, and then the threshold calculation unit 23 calculates a threshold for detecting cracks. However, a process of emphasizing a crack may be performed between the difference calculation and the threshold calculation. For example, as shown in FIG. 14, when the first calculation unit 22 performs the difference calculation between the wafer surface image data and the infrared image data, the filter processing unit 33 emphasizes the difference value of the pixel having a large brightness difference. The image processing unit 8 may further include a second calculation unit 34 that performs a difference calculation between the filtered image processed by the filter processing unit 33 and the infrared image data.

  In the filter processing unit 33, when the difference calculation between the wafer surface image data and the infrared image data is performed, a pixel having a large brightness difference is processed as image data in which the difference value is emphasized as a bright part. By this treatment, the gap between the crack and the crystal grain boundary is emphasized and remains as a bright part.

  The second calculation unit 34 performs a difference calculation between the filtered image processed by the filter processing unit 33 and the infrared image data, so that the contrast ratio between the crack and the misalignment of the crystal grain boundary and the image data of the crystal grain is obtained. Is more emphasized. That is, only cracks remain as image data with enhanced brightness. If the threshold value calculation unit 23 calculates a threshold value based on the image data processed by the second calculation unit 34 and the binarization processing unit 24 performs binarization processing, the binarized image As the data, image data having a high probability of being a crack can be obtained.

  (2) In the above embodiment, the detection of cracks is detected by pattern matching, but it may be detected by the dispersion density of the labeled pixel colonies. The pixel colony is divided into small areas, and the number of suspected cracks for the number of pixels in the small area is calculated for each small area. Then, regarding the detection ratios in all the small regions, the variance is obtained and the standard deviation is calculated. Since the standard deviation value indicates variations in pixel colonies, cracks can be determined based on the standard deviation value.

  (3) In the above embodiment, the threshold value is calculated after the difference, but the brightness of the image data of either the wafer surface image data or the infrared transmission image data may be reversed before the difference, or the wafer White balance that matches the brightness of the surface image data and the infrared transmission image data may be taken. This makes it easier to calculate the threshold value.

It is the schematic of the crack detection apparatus of the polycrystalline silicon wafer which concerns on an Example. It is a block diagram of the image processing part in the crack detection apparatus of the polycrystalline silicon wafer which concerns on an Example. It is a flowchart which shows a series of sequences which detect the crack of the polycrystalline silicon wafer which concerns on an Example. It is explanatory drawing which images the image by the reflected light of the polycrystalline silicon wafer which concerns on an Example. It is explanatory drawing which images the image by the transmitted light of the polycrystalline silicon wafer which concerns on an Example. It is explanatory drawing which shows the picked-up image of the polycrystalline silicon wafer by the reflected light based on an Example. It is explanatory drawing which shows the captured image of the polycrystal silicon wafer by the infrared rays transmitted light which concerns on an Example. It is explanatory drawing which shows the image which piled up the captured image by the reflected light based on an Example, and the captured image by infrared rays transmitted light. It is explanatory drawing which calculates a threshold value in the image process part which concerns on an Example. It is explanatory drawing which detects a pixel colony in the image process part which concerns on an Example. It is a figure which shows the pattern which performs pattern matching in the image process part which concerns on an Example. It is a figure which shows the pattern which performs pattern matching in the image process part which concerns on an Example. It is explanatory drawing which performs pattern matching in the image process part which concerns on an Example. It is a block diagram of the image processing part in the crack detection apparatus of the polycrystalline silicon wafer which concerns on the other Example of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Polycrystalline silicon wafer detection apparatus 2 ... Polycrystalline silicon wafer 3 ... Wafer conveyance part 4 ... White light source 5 ... 1st line sensor camera 6 ... Infrared tube 7 ... 2nd line sensor camera 8 ... Image processing part 9 ... Host Computer 10 ... Wafer storage unit 11 ... transfer drive unit 15 ... storage transfer mechanism 16 ... storage drive unit 17 ... control unit 20 ... first memory unit 21 ... second memory unit 22 ... first calculation unit 23 ... threshold value calculation unit 24 ... Binary processing unit 25 ... Labeling unit 26 ... Crack discrimination unit 33 ... Filter processing unit 34 ... Second calculation unit PM1, PM2 ... Template image

Claims (6)

A crack detection device for a polycrystalline silicon wafer,
Light irradiation means for irradiating white light or monochromatic light on one surface of the polycrystalline silicon wafer;
First imaging means for imaging reflected light of the surface irradiated with the white light or monochromatic light of the polycrystalline silicon wafer;
Infrared irradiation means for irradiating infrared rays on the other surface of the polycrystalline silicon wafer;
Second imaging means for imaging the infrared transmitted light that is transmitted from the other surface of the polycrystalline silicon wafer to one surface;
A first computing unit for subtracting, for each pixel, wafer surface image data obtained from the first imaging means and wafer transmission image data obtained from the second imaging means;
A threshold value calculation unit that calculates a threshold value for determining a pixel suspected of being cracked from the difference value obtained by the first calculation unit;
A binarization processing unit for discriminating a pixel suspected of being cracked from the difference value obtained by the first calculation unit according to the threshold value calculated by the threshold value calculation unit;
A crack detection apparatus for a polycrystalline silicon wafer, comprising: a crack discriminating unit that discriminates a crack from a distribution state of pixels suspected of being cracked discriminated by the binarization processing unit. A crack detection device for a polycrystalline silicon wafer,
Light irradiation means for irradiating white light or monochromatic light on one surface of the polycrystalline silicon wafer;
First imaging means for imaging reflected light of the surface irradiated with the white light or monochromatic light of the polycrystalline silicon wafer;
Infrared irradiation means for irradiating infrared rays on the other surface of the polycrystalline silicon wafer;
Second imaging means for imaging the infrared transmitted light that is transmitted from the other surface of the polycrystalline silicon wafer to one surface;
A first computing unit for subtracting, for each pixel, wafer surface image data obtained from the first imaging means and wafer transmission image data obtained from the second imaging means;
A filter processing unit that emphasizes a difference value obtained by the first calculation unit for a pixel having a larger brightness difference when the wafer calculation image data and the wafer transmission image data are differentiated by the first calculation unit; ,
A second arithmetic unit for substituting the image data processed by the filter processing unit and the wafer transmission image data for each pixel;
A threshold value calculation unit that calculates a threshold value for determining a pixel suspected of being cracked from the difference value obtained by the second calculation unit;
A binarization processing unit that discriminates a pixel suspected of being cracked from the difference value obtained by the second calculation unit according to the threshold value calculated by the threshold value calculation unit;
A crack detection apparatus for a polycrystalline silicon wafer, comprising: a crack discriminating unit that discriminates a crack from a distribution state of pixels suspected of being cracked discriminated by the binarization processing unit. In the crack detection apparatus of the polycrystalline silicon wafer according to claim 1 or 2,
The crack detection unit for a polycrystalline silicon wafer, wherein the crack determination unit determines a crack by pattern matching with a predetermined template. In the crack detection apparatus of the polycrystalline silicon wafer according to claim 1 or 2,
The crack detection unit for a polycrystalline silicon wafer, wherein the crack determination unit determines a crack based on a dispersion density of pixels suspected of being cracked. In the crack detection apparatus of the polycrystalline silicon wafer according to any one of claims 1 to 4,
A crack detection apparatus for a polycrystalline silicon wafer, comprising: a wafer storage unit that stores the polycrystalline silicon wafer by sorting the polycrystalline silicon wafer based on the presence or absence of a crack based on determination by a crack determination unit. In the crack detection apparatus of the polycrystalline silicon wafer as described in any one of Claim 1 to 5,
A crack detection apparatus for a polycrystalline silicon wafer, wherein the first imaging means and the second imaging means are line sensor cameras.

Images