JP2006222514A - Method of inspecting solid-state imaging device, inspection program, and electronic camera - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an inspection method capable of easily detecting fixed pattern noise of a solid-state imaging device. <P>SOLUTION: The inspection method acquires an image signal generated by the solid-state imaging device through emission of an illumination light for inspection to respectively generate a plurality of partial images with respect to a whole image indicated by the image signal by the same image size. Then the inspection method applies processing such as inversion or rotation to the partial images so as to unchange the image size and thereafter applies processing such as averaging to the partial images to produce a reference image, and respectively generates inspection images equivalent to differences between the reference image and the plurality of partial images. Assuming that there exists no fixed pattern noise at all, the reference image is a uniform image and all the inspection images have uniformly a black level. That is, normal pixels reach a black level in the inspection images, and the signal level is greatly different from those of the noise parts. Thus, the fixed pattern noise can easily be detected by using the inspection images. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a solid-state image sensor inspection method, a solid-state image sensor inspection program, and an electronic camera. In particular, the present invention relates to a technique for easily and accurately detecting fixed pattern noise in a solid-state image sensor inspection process.

In the inspection process of solid-state imaging devices, in addition to items representing device characteristics such as sensitivity, saturation signal, and color properties, items representing image quality include defects such as S / N ratio, point defects, line defects, spots, unevenness, etc. Check for fixed pattern noise.
Sensitivity, saturation signal, and the like can be easily quantified and easily inspected by measuring an image signal output after irradiating the solid-state image sensor with uniform light of a specified brightness. Since the point defect is a defect of one pixel unit, it can be easily detected by determining the threshold level (see, for example, Patent Document 1). Line defects are divided into vertical line defects and horizontal line defects, but can be easily detected because the generation direction is limited to the column direction or the row direction.

  As a conventional technique for detecting fixed pattern noise, Patent Document 2 and the like are known. FIG. 13 is an explanatory diagram showing an example of a conventional technique for detecting fixed pattern noise. As a detection method, first, after irradiating the solid-state image sensor with uniform light, an image signal is acquired from the solid-state image sensor. Next, an image based on this image signal is divided into a plurality of regions, an average value of signal levels is calculated for each image region, and fixed pattern noise is detected by comparing it with a threshold level.

  If the inspected solid-state imaging device is a perfect non-defective product, the average value of the signal level should be completely equal over the entire image area, but this is not the case when there is fixed pattern noise. FIG. 13A shows a schematic diagram in which one image obtained by irradiation with uniform light is divided into a number of image areas, and vertical and horizontal dotted lines in the figure are boundaries of the image areas. In this example, fixed pattern noise occurs in the three central image areas. FIG. 13B shows a signal level profile of each image region between the image regions P and P ′ in FIG. 13A, the horizontal axis indicates the horizontal position, and the vertical axis indicates the signal level. .

  As a cause of occurrence of the fixed pattern noise, there is an influence of non-uniformity and dust generation in the color filter process and the micro lens process formed on the outermost surface layer of the solid-state imaging device. As the color filter process, first, flattening is performed to eliminate a step due to the wiring layer of the solid-state imaging device formed on the semiconductor substrate (wafer). Thereafter, a color filter material is dropped and spin-coated. In the microlens process, after leveling the step of the color filter, the microlens material is dropped and spin-coated.

  For example, in the color filter step, after a pigment dispersion type color resist is dropped on the center of the wafer, the wafer is rotated at a high speed by a spinner so that the pigment resist is uniformly coated on the wafer surface. The entire surface of the wafer is not completely flattened in the flattening step, and a step remains, particularly in a scribe line. Due to this step difference, coating unevenness, that is, non-uniform film thickness of the color filter material or microlens material occurs.

In addition, foreign matter may be mixed during spin coating to become a nucleus, resulting in a region with a small coating amount. In this case, as shown in the schematic plan view of the one-chip solid-state imaging device shown in FIG. 14, for example, an elliptical unevenness whose major axis faces the wafer center occurs. As described above, the amount of light incident on the solid-state imaging device varies partially due to the non-uniformity of the film thickness. This causes a partial sensitivity variation of the solid-state imaging device, and generates fixed pattern noise.
JP 11-233580 A Japanese Patent Laid-Open No. 5-60650

  Fixed pattern noise has a non-uniform generation shape and a variety of generation directions, and there are a large difference and a small difference in signal level with respect to normal pixels. In particular, fixed pattern noise with a small difference in signal level with respect to normal pixels cannot always be detected easily. Therefore, there has been a demand for a technique for easily detecting such irregularly generated fixed pattern noise.

An object of the present invention is to provide an inspection method and an inspection program capable of easily detecting fixed pattern noise in a solid-state imaging device.
Another object of the present invention is to provide a technique for improving the image quality of an electronic camera based on fixed pattern noise detected by an inspection method suitable for the above purpose.

  The inspection method for a solid-state imaging device according to claim 1 includes the following steps. Those steps include a step of acquiring an image signal generated by the solid-state imaging device by irradiation of illumination light for inspection, and an entire image that is an image indicated by the acquired image signal, and a plurality of partial images with respect to the entire image are obtained. An extraction step for generating each image with the same image size, an inversion step for performing at least one of upside-down inversion, left-right inversion, and rotation so that the image size does not change for at least one partial image, and an inversion step Thereafter, a step of combining a plurality of partial images to generate a reference image having the same image size as the partial image, a step of generating an inspection image corresponding to a difference from the reference image for each of the plurality of partial images, And determining whether or not the solid-state imaging device is a non-defective product based on the image.

According to a second aspect of the present invention, in the inspection method for a solid-state imaging device according to the first aspect, "when all partial images are overlapped, an image region that is a partial image region of the partial image and is closest to the center of the entire image. In the inversion step, at least one of upside-down inversion, left-right inversion, and rotation is performed so that the positions are aligned.
According to a third aspect of the present invention, in the method for inspecting a solid-state imaging device according to the second aspect, the "in the extraction step, a part of the outer periphery of each partial image is a part of the outer periphery of the entire image, and each An even number of partial images are generated so that there are partial images that are point-symmetric with respect to the center of the entire image.

According to a fourth aspect of the present invention, in the inspection method for a solid-state image pickup device according to the third aspect, the “in the extraction step, the whole image is divided into four so that the center of the whole image is a corner of each partial image. It generates a partial image ”.
The invention of claim 5 is an inspection program for causing a computer to execute the following steps in order to determine whether or not the solid-state imaging device is a non-defective product. Those steps are a step of acquiring an image signal generated by the solid-state imaging device by irradiation of illumination light for inspection, and an entire image which is an image indicated by the acquired image signal, and each of a plurality of partial images with respect to the entire image. After the extraction step that generates the same image size, the inversion step that performs at least one of upside down, left and right inversion, and rotation so that the image size does not change for at least one partial image, and after the inversion step Combining a plurality of partial images to generate a reference image having the same image size as the partial images, generating a test image corresponding to a difference from the reference image for each of the plurality of partial images, and a test image And determining whether or not the solid-state imaging device is a non-defective product.

  According to a sixth aspect of the present invention, there is provided an electronic camera comprising: a solid-state imaging device; a signal processing unit that generates image data based on an image signal output from the solid-state imaging device; and an image processing unit that performs image processing on the image data. It is. According to the present invention, the fixed image noise estimated from the inspection image generated for the solid-state image sensor by the solid-state image sensor inspection method according to any one of claims 1 to 4 , That the process is reduced ”.

  In the present invention, after generating a plurality of partial images with the same image size for an image obtained by irradiating the illumination light for inspection on the solid-state imaging device, the partial images are inverted or rotated so that the image size does not change. To do. Then, after the reference images are generated by combining the inverted or rotated partial images, inspection images corresponding to differences from the reference image are generated for each of the partial images. If there is no fixed pattern noise at all, when a reference image is generated by averaging a plurality of partial images, the reference image is a uniform image, and all the inspection images are uniformly at the black level. That is, in the inspection image, the normal pixels are almost black level, and the signal level is greatly different only by the noise part. Therefore, fixed pattern noise can be easily detected by using an inspection image.

<Description of Principle of the Present Invention>
Prior to the description of the embodiments of the present invention, the principle of the present invention will be described.
FIG. 1 is a schematic plan view of a semiconductor wafer 12 on which a plurality of (in this example, 98) solid-state imaging device chips 10 are formed. As shown in the figure, on the surface of the wafer 12, the chips 10 are partitioned in a lattice shape by streets 14 (grooves for dicing). In the figure, the lines with arrows at both ends are imaginary lines passing through the center of the wafer, and are described to show the positional relationship between some chips 10 and the center of the wafer 12. Further, the four ellipse areas indicated by dotted lines and the portions denoted by E and F will be described later. The street 14 is a large step on the surface of the wafer 12. Therefore, during spin coating, radial coating unevenness may occur starting from the corner closest to the center of the wafer 12 in each chip 10.

  2A, 2B, 2C, and 2D are explanatory diagrams showing an example of radial coating unevenness, and chips denoted by reference numerals A, B, C, and D in FIG. 1, respectively. 10 is a schematic plan view focusing on FIG. In the following description, the chips 10 at positions A, B, C, and D are referred to as a chip 10A, a chip 10B, a chip 10C, and a chip 10D, respectively. The chips 10 </ b> A and 10 </ b> C are point-symmetric with respect to the center of the wafer 12, and the chips 10 </ b> B and 10 </ b> D are also point-symmetric with respect to the center of the wafer 12.

  In FIG. 2, the circles attached to any of the four corners of the chips 10 </ b> A to 10 </ b> D are described for convenience in order to show the positional relationship, and are attached to any of the four corners of the chips 10 </ b> A to 10 </ b> D in FIG. 1. Corresponds to the round mark. As shown in FIG. 2, each of the chips 10 </ b> A to 10 </ b> D has a pixel region 20 indicated by a dotted line and a plurality of pads 22. As shown in FIG. 2A, in the chip 10A, application unevenness occurs on the lower right side of the pixel region 20 starting from the lower right corner, and this application unevenness causes fixed pattern noise. Further, as shown in FIGS. 2B, 2C, and 2D, in the chips 10B, 10C, and 10D, coating unevenness occurs on the lower left side, the upper left side, and the upper right side, respectively, and these are fixed pattern noises. Give rise to

  Therefore, in the chips 10A, 10B, 10C, and 10D, fixed pattern noise is generated at different positions in the image. In FIG. 2, the chips 10A, 10B, 10C, and 10D on the outer peripheral side of the wafer 12 are shown. For example, as shown in the chip 10 denoted by E and F in FIG. It is considered that fixed pattern noise having a different shape is generated at a position different from 10A, 10B, 10C, and 10D.

  It is considered that the fixed pattern noise caused by the spin coating is generated at which position in the image and in what shape is different for each chip 10 and depends on which position on the wafer 12 is impositioned. Therefore, if all the chips 10 imprinted on one wafer 12 are individually inspected, the position of the pixel region where the fixed pattern noise occurs and the shape of the noise are the number of chips 10 on one wafer 12. There will be a pattern. This variety of patterns has made it difficult to detect fixed pattern noise easily and efficiently.

  However, it is considered that the fixed pattern noise caused by spin coating is generated point-symmetrically with respect to the center of the wafer 12 when viewed from the whole wafer 12. Specifically, in FIG. 2, coating unevenness in each of the chips 10 </ b> A to 10 </ b> D mainly occurs in a region between the corner closest to the center of the wafer 12 and the center of the chip, and is closest to the center of the wafer 12. This is common in that it occurs radially from the corner to the center of the chip.

  The present inventor focusing on the commonality and the point symmetry on the original wafer 12 applies a simple image processing to the image for inspection, thereby generating fixed pattern noise that is randomly generated in each chip 10. Developed a method for converting to a common and simple pattern. Specifically, after generating a plurality of partial images having the same image size with respect to an image (entire image) obtained from the chip 10 of the solid-state imaging device with uniform illumination light, the image is inverted or rotated to each partial image. Apply. Here, “the image size is the same” means that the shape of the outer edge of the image is the same in all the partial images, and if the image is rectangular, for example, the number of pixels in the horizontal direction and the pixels in the vertical direction. It means that the numbers are the same.

  In the inversion or rotation performed at this time, when the partial images are overlapped, it is desirable to align the “image region closest to the center of the entire image” in each partial image at a predetermined position. In order to facilitate this processing, each of the partial images is point-symmetric with respect to the center of the entire image so that a part of the outer periphery of each partial image becomes a part of the outer periphery of the entire image. It is preferable to generate an even number of partial images so that there are partial images at positions.

  3, 4, 5, and 6 are explanatory diagrams illustrating an example of the above-described image processing, and illustrate image processing for the chips 10 </ b> A, 10 </ b> B, 10 </ b> C, and 10 </ b> D, respectively. 3 to 6, the uppermost figure (a) is a schematic plan view of each of the chips 10A, 10B, 10C, and 10D. Circles are attached to the four corners of the rectangular pixel region 20, A triangle mark is attached to the center of the pixel region 20.

  Hereinafter, the image processing for the chip 10A will be described with reference to FIG. First, an image obtained from the chip 10A with uniform illumination light (an image corresponding to an image signal of all pixels in the pixel region 20 and hereinafter referred to as an entire image AW) is equally divided into four vertically and horizontally. Thereby, the partial images generated for the upper left, upper right, lower left, and lower right regions of the entire image AW are referred to as A1, A2, A3, and A4, respectively. FIG. 3B shows the partial images A1, A2, A3, and A4 with a triangle mark at the center of the whole image AW and circle marks at the four corners, respectively.

  Next, the corner corresponding to the center of the whole image AW in each partial image A1, A2, A3, A4 is aligned at a predetermined position (in this example, the lower right of the image) by inversion or rotation. Specifically, the partial image A2 is horizontally reversed, the partial image A3 is vertically reversed, and the partial image A4 is rotated 180 ° (horizontally reversed and vertically reversed). The partial images A2, A3, and A4 after the above processing are referred to as partial images A2 ', A3', and A4 ', respectively. The partial image A1 may be left as it is because the corner corresponding to the center of the entire image AW is located at the lower right of the image. FIG. 3C shows partial images A2 ', A3', A4 'and partial image A1 generated by the above processing. In the present specification, hereinafter, the processes of upside down, left and right inversion, and rotation by a predetermined angle performed on a partial image are collectively referred to as “inversion processing”. The inversion processing corresponds to the inversion step described in the claims, and has a concept including rotation.

  Next, the partial images A1, A2 ', A3', A4 'are averaged (or added), and an image generated thereby is defined as a reference image A5. The image size of the reference image A5 is the same as that of the partial images A1 to A4. The reason for making the image sizes the same when generating the partial images A1 to A4 in the previous procedure is that if the image sizes are different, they cannot be averaged or added. FIG. 3D shows a reference image A5.

  For the chips 10B, 10C, and 10D, the reference image is generated by the same process as described above. Specifically, for the chip 10B, an image obtained by uniform illumination light irradiation (hereinafter referred to as an entire image BW) is equally divided into four parts, as shown in FIG. B2, B3, and B4 are generated. Next, the partial image B2 is reversed left and right to be a partial image B2 ', the partial image B3 is vertically reversed to be a partial image B3', and the partial image B4 is rotated 180 degrees to be a partial image B4 '. Next, the partial images B1, B2 ', B3', B4 'are averaged to generate a reference image B5 as shown in FIG.

  For the chip 10C, an image (hereinafter referred to as an entire image CW) obtained by irradiation with uniform illumination light is equally divided into four, and partial images C1, C2, C3, C4 as shown in FIG. 5B. Is generated. Next, as shown in FIG. 5 (c), the partial image C2 is reversed left and right to become a partial image C2 ′, the partial image C3 is vertically reversed to be a partial image C3 ′, and the partial image C4 is rotated by 180 °. Let it be a partial image C4 ′. Next, the partial images C1, C2 ', C3', and C4 'are averaged to generate a reference image C5 as shown in FIG.

  For the chip 10D, an image (hereinafter referred to as an entire image DW) obtained by irradiation with uniform illumination light is equally divided into four, and partial images D1, D2, D3, D4 as shown in FIG. 6B. Is generated. Next, as shown in FIG. 6C, the partial image D2 is reversed left and right to become a partial image D2 ′, the partial image D3 is vertically reversed to be a partial image D3 ′, and the partial image D4 is rotated by 180 °. Let it be a partial image D4 ′. Next, the partial images D1, D2 ', D3', and D4 'are averaged to generate a reference image D5 as shown in FIG.

  As described above, the common reference images A5, B5, D5, and D5 are generated by performing the common processing on the chips 10A to 10D in which the fixed pattern noise appears differently. As is clear from comparison between the reference images A5, B5, D5, and D5, all have the same image pattern. That is, it is possible to convert fixed pattern noise that is randomly generated when viewed for each chip 10 into a common and simple pattern by performing common processing based on symmetry on each chip 10 as described above. The person elucidated.

  Note that the commonality here does not mean that it is common to all the chips 10 in the wafer 12, but depending on the position of the chip 10 in the wafer 12, some chips 10 have the same pattern in the reference image. It means that it can be converted into noise. It is considered that at least the chips 10 having a symmetrical positional relationship with respect to the center of the wafer 12 have the same fixed pattern noise in the reference image.

  In addition, the fixed pattern noise that appears only in a part of the image in the original whole images AW, BW, CW, and DW is enlarged to the whole image in the reference images A5, B5, C5, and D5 (in this example, it is enlarged four times). ) Therefore, based on the reference image generated in this way, detection of fixed pattern noise becomes easier than in the past. The above is the description of the principle of the present invention, and the embodiments will be described below.

<First Embodiment>
The first embodiment embodies the present invention as a method for inspecting a solid-state imaging device. As in the examples described with reference to FIGS. 3 to 6, the entire image obtained by uniform light irradiation is equally 4. After dividing and generating four partial images, a reference image is generated. Here, in the reference image obtained by averaging the four partial images, the difference between the signal level of the pixel in which the fixed pattern noise appears and the signal level of the normal pixel (that is, the noise intensity) is 4 minutes. 1 Therefore, it is desirable to emphasize noise as a process before detecting noise. Therefore, in the first embodiment, whether or not the chip 10 of the solid-state imaging device is a non-defective product is determined based on the inspection image after generating an inspection image in which noise is emphasized more than the reference image.

  FIG. 7 is an explanatory diagram of an inspection image. In the first embodiment, four inspection images are generated for each chip 10. Specifically, for three of the four partial images that are subjected to inversion processing, a difference image between each partial image after the inversion processing and the reference image is used as an inspection image. One of the four partial images that is not subjected to inversion processing uses a difference image between the partial image and the reference image as an inspection image.

  For example, the inspection image for the chip 10A is as shown in FIG. In FIG. 7A, inspection images A1-A5 are obtained by subtracting the signal level of each pixel in the reference image A5 from the signal level of each pixel of the partial image A1. In the inspection images A2′-A5, A3′-A5, and A4′-A5, the signal level of each pixel in the reference image A5 is subtracted from the signal level of each pixel in the partial images A2 ′, A3 ′, and A4 ′. It is a thing.

  Inspection images are generated for the other chips 10 in the same manner as described above. FIG. 7B shows inspection images B1-B5, B2'-B5, B3'-B5, and B4'-B5 for the chip 10B. FIG. 7C shows inspection images C1-C5, C2'-C5, C3'-C5, and C4'-C5 for the chip 10C. FIG. 7D shows inspection images D1-D5, D2'-D5, D3'-D5, and D4'-D5 for the chip 10D.

  The inspection image is the difference between the partial image after inversion processing or the partial image that has not been subjected to inversion processing, and the reference image, so that the normal pixel has a black level and the signal level differs greatly only in the noise portion. It becomes. Therefore, only the noise component is extracted from the inspection image. As a result, in the present invention, it is possible to easily determine a normal image region or a noise component, compared with a case where a normal pixel represents a predetermined luminance level by the illumination light for inspection.

FIG. 8 is a flowchart illustrating a method for inspecting a solid-state imaging device according to the first embodiment. The inspection apparatus (not shown) used here is installed with a program (corresponding to claim 5) in which the following steps S1 to S7 are encoded. The flow of the inspection process will be described below according to the step numbers shown in the figure.
In step S1, a plurality of solid-state imaging device chips 10 are placed on the stage of the inspection apparatus in the state of the wafer 12, then one uninspected chip is selected, and the probe needle of the inspection apparatus is brought into contact with the pad of the chip 10 . Next, the image signal for inspection is generated on the chip 10 of the solid-state image sensor by a known method such as darkening the inspection room and discharging the charges in each pixel and then irradiating the chip 10 with uniform illumination light. Then, the inspection apparatus acquires this image signal (image signal of the entire image).

Next, in step S2, the whole image is equally divided into four to generate image data of four partial images (corresponding to the extraction step recited in the claims).
Next, in step S3, the corner corresponding to the center of the whole image in the four partial images is aligned at a predetermined position (for example, the lower right of the image) by reversal processing. The reversal processing here has already been described with reference to FIGS. 3 to 6 as the principle of the present invention, and thus description thereof will be omitted.

Next, in step S4, the four partial images are averaged to generate a reference image.
Next, in step S5, four inspection images are generated based on the difference between the three partial images after the reversal process and the reference image, and the difference between the partial image not subjected to the reversal process and the reference image (described above). (See FIG. 7).
Next, in step S6, it is determined whether the chip 10 is a non-defective product based on the inspection image. If there is no fixed pattern noise in a certain chip 10, the reference image for the chip 10 is a uniform image, and all four inspection images should be uniformly at the black level. Therefore, for example, an inspection image is divided into a plurality of image areas, an average value of signal levels is calculated for each image area, and the fixed pattern noise is detected by comparing it with a threshold level. For example, for any of the four inspection images, it is determined as non-defective only if the image area determined to have fixed pattern noise is less than a predetermined number, and otherwise determined as non-defective. If the product is not a non-defective product, for example, the chip 10 may be marked on the wafer 12 (in this embodiment, a mark is not marked to prevent dust generation).

Next, in step S7, it is determined whether or not there is an uninspected chip 10 in the wafer 12. If not, the inspection for the wafer 12 is terminated. If there is an uninspected chip 10, the process returns to step S1 to inspect the uninspected chip 10. The above is the description of the inspection process.
As described above, in the first embodiment, the entire image obtained by the uniform light irradiation is divided into four, and the partial image is subjected to the inversion process, and then the reference image unique to each chip 10 is generated. Thereby, the fixed pattern noise generated only in a part of the original whole image is enlarged four times in the inspection image, so that it is easy to determine the presence or absence of the fixed pattern noise.

  Also, in the inspection image, normal pixels are almost black level, and the signal level is greatly different only in the noise portion. Therefore, an image area where fixed pattern noise appears and a normal image area can be easily identified by a threshold value. That is, the accuracy of detection of fixed pattern noise can be improved. As a result, the manufacturing cost can be reduced by performing sorting at the wafer stage and not sending the non-defective chip 10 to the mounting process. Furthermore, since the number of pixels of the inspection image is reduced to a quarter of the entire image, an effect of shortening the inspection time can be obtained.

  In the ion implantation process for forming the impurity diffusion region that is a main part of the solid-state imaging device, the concentration of implanted impurities is uneven due to uneven scanning of the ion beam or slight variations in the ion beam diameter during scanning. Sometimes. This injection unevenness becomes a sensitivity unevenness of the solid-state imaging device, causes fixed pattern noise, and deteriorates the image quality. Fixed pattern noise due to implantation unevenness occurs when similar noise is generated for all chips on one wafer and when the entire wafer is a regular pattern. There may be as many as. According to the first embodiment, it is possible to easily and reliably detect fixed pattern noise due to uneven injection without using various conventional image processing filters or performing complicated pattern matching that can be assumed. it can.

<Second Embodiment>
The second embodiment also embodies the present invention as a method for inspecting a solid-state imaging device, like the first embodiment. In the first embodiment, the entire area of the entire image is used for inspection. However, when it is known in advance that fixed pattern noise is generated in a specific area in the image, the inspection may be performed while paying attention to only the specific area. In the second embodiment, only the specific area is used for the inspection. Note that the flow of the inspection process of the solid-state imaging device of the second embodiment is the same as that of the first embodiment except for the difference in the method of generating partial images, and thus the flowchart is omitted.

  FIG. 9 is an explanatory diagram illustrating a method for generating a partial image and a reference image according to the second embodiment. Hereinafter, the inspection method for the solid-state imaging device of the second embodiment will be described with reference to FIG. FIG. 9A is a plan view of the chip 10 </ b> A, in which circles are added to the four corners of the pixel area 20 and a triangle mark is added to the center of the pixel area 20. In the second embodiment, as an example, in order to generate a partial image, the entire image is equally divided into 9 areas of 3 × 3 in length.

  Here, assuming that fixed pattern noise is generated near any one of the four corners of each chip 10, a region corresponding to the four corners of the entire image AW is selected from the nine regions, and the partial images XA1, XA3, and XA7 are selected. , XA9 is generated. FIG. 9B shows partial images XA1, XA3, XA7, and XA9 with circles of portions corresponding to the four corners of the entire image AW. In order to make the subsequent inversion processing easy to understand, in FIG. 9B, for each partial image XA1, XA3, XA7, XA9, a square mark is attached to the region closest to the center of the entire image AW ( The same applies to FIG. 9A).

  Next, inversion processing is performed to align the image area closest to the center of the entire image AW in each of the partial images XA1, XA3, XA7, and XA9 at a predetermined position (in this example, the lower right of the image). Specifically, the partial image XA3 is reversed left and right to be a partial image XA3 ', the partial image XA7 is vertically reversed to be a partial image XA7', and the partial image XA9 is rotated 180 degrees to be a partial image XA9 '. The partial image XA1 may be left as it is because the region closest to the center of the entire image AW is located at the lower right of the image. FIG. 9C shows these partial images XA3 ', XA7', XA9 'and partial image XA1.

  After the inversion process, the partial images XA1, XA3 ', XA7', and XA9 'are averaged to generate a reference image XA10 as shown in FIG. 9D. Thereafter, four inspection images XA1-XA10, XA3′-XA10, XA7′-XA10, XA9′-XA10 ( (Not shown). Then, using the four inspection images, it is determined whether or not it is a non-defective product as in the first embodiment.

  The above is the description of the method for inspecting the solid-state imaging device of the second embodiment, and the same effect as that of the first embodiment can be obtained in the second embodiment. Furthermore, in the second embodiment, since the amount of image data used for processing is small, inspection can be performed in a short time. Note that the entire image may be equally divided into a large number instead of 3 × 3 in the vertical direction, and four partial images corresponding to the four corners of the entire image may be used. In this case, the amount of image data used for processing is further reduced, and inspection can be performed in a shorter time.

<Third Embodiment>
Similarly to the first and second embodiments, the third embodiment also embodies the present invention as an inspection method for a solid-state imaging device. In the second embodiment, attention is paid to the four corner regions of the entire image. In the third embodiment, attention is paid to the central portion of one side of the entire image. Further, the partial images used for the inspection are adjacent to each other in the whole image in the first embodiment and separated in the whole image in the second embodiment, but may be overlapped in the whole image. In the third embodiment, an example of such overlapping will be described. Note that the flow of the inspection process of the solid-state imaging device of the third embodiment is the same as that of the first embodiment except for the difference in the generation method of the partial image, and thus the flowchart is omitted.

  FIG. 10 is an explanatory diagram illustrating a partial image and reference image generation method according to the third embodiment. Hereinafter, a method for inspecting a solid-state imaging device according to the third embodiment will be described with reference to FIG. FIG. 10A is a plan view of the chip 10A, in which a star is attached to the center of the four sides of the pixel region 20. FIG. Four regions indicated by dotted-line squares in the pixel region 20 of FIG. 10A are regions that become the partial images YA1, YA2, YA3, and YA4 of the third embodiment.

  The partial images YA1, YA2, YA3, and YA4 are all square, and all have a part of the outer periphery of the entire image AW as one side. The centers of the four sides of the entire image AW coincide with the centers of one side in the partial images YA1, YA2, YA3, and YA4, respectively. The partial images YA1 and YA3 are adjacent to each other. A part of the partial image YA2 overlaps a part of the partial image YA1 and a part of the partial image YA3, and a part of the partial image YA4 overlaps a part of the partial image YA1 and a part of the partial image YA3. Accordingly, the partial images YA1 and YA3 have a point-symmetrical positional relationship with respect to the center of the entire image, and the partial images YA2 and YA4 have a positional relationship of point-symmetrical with respect to the center of the entire image. FIG. 10B shows partial images YA1, YA2, YA3, and YA4 with a star mark corresponding to each central portion of the four sides of the entire image. In FIG. 10B, in order to make it easy to understand the positional relationship compared with that after the reversal processing at the center of the side opposite to the side marked with an asterisk in each partial image YA1, YA2, YA3, YA4. A square mark is attached.

  Then, after generating the four partial images, inversion processing is performed, and in each partial image YA1, YA2, YA3, YA4, the image area closest to the center of the entire image is set at a predetermined position (in this example, the upper center of the image). Align. Specifically, the partial image YA2 is rotated 90 ° counterclockwise to become a partial image YA2 ′, the partial image YA3 is inverted upside down to become a partial image YA3 ′, and the partial image YA4 is rotated 90 degrees clockwise (to the right). Rotate to obtain a partial image YA4 ′. The inversion process is not performed on the partial image YA1. FIG. 10C shows the partial images YA2 ', YA3', YA4 ', and the partial image YA1.

  After the inversion process, the partial images YA1, YA2 ', YA3', and YA4 'are averaged to generate a reference image YA5 as shown in FIG. Thereafter, four inspection images YA1-YA5, YA2′-YA5, YA3′-YA5, YA4′-YA5 ( (Not shown). Then, using these four inspection images, it is determined whether or not it is a non-defective product as in the first embodiment.

  The above is the description of the method for inspecting the solid-state imaging device of the third embodiment. In the third embodiment, the same effect as in the first and second embodiments can be obtained. In the third embodiment, the example in which the partial images YA1, YA2, YA3, and YA4 are adjacent to each other or overlap each other is described. However, these four images may be reduced and separated. In that case, it is desirable to reduce each of the partial images YA1, YA2, YA3, and YA4 so that a part of the outer periphery of the entire image AW is one side and the square shape is maintained.

Hereinafter, supplementary items of the first to third embodiments will be described.
In the first to third embodiments, the example in which the reference image is generated by averaging all the partial images in step S4 of FIG. 8 has been described. The present invention is not limited to such an embodiment. For example, each partial image may be added to generate a reference image having the same size as the partial image.
Since it is desirable to select only non-defective products at an early stage, it is desirable to inspect the solid-state imaging device in the state of the wafer 12 as in the first to third embodiments, but after dicing or mounting as a solid-state imaging device. You may inspect.

  The technique of the second embodiment is that the chip 10 is likely to cause uneven coating radially from any one of the four corners in that it efficiently inspects without using an image area where the possibility of occurrence of fixed pattern noise is low. It is considered effective for the chip 10 on the outer peripheral side rather than the center side of the wafer 12. The reason will be described below. In the arrangement of FIG. 1, the location where the distance from the wafer center is the shortest on the outer periphery of each chip 10 is one of the four corners, but this tendency is particularly significant in the chips (10A, etc.) on the outer peripheral side of the wafer 12. strong. That is, when the chip 10 is viewed as a rectangle, the corner closest to the wafer center has a considerably shorter distance from the wafer center than the center of each side. Therefore, the chip 10 having a high probability of occurrence of uneven coating radially from any one of the four corners is considered to be the chip 10 on the outer peripheral side of the wafer 12.

  On the contrary, since the technique of the third embodiment is effective for the chip 10 in which noise is generated at the center of one side of the entire image, it is within the four dotted ellipses in FIG. 1 rather than the outer peripheral side of the wafer 12. It is considered effective for the chip 10 on the center side as in the area of. Considering that the resist immediately after being dropped onto the wafer center has some spread before high-speed rotation by a spinner, the distance between the corner and the center of each side is the distance from the resist immediately after dropping. This is because there is not much change. Then, by predicting the appearance of fixed pattern noise from the position on the wafer 12, individual patterns are registered in advance for each position on the wafer 12, and pattern matching with the registered pattern is performed. Noise may be detected.

FIG. 11 is a flowchart for inspecting a solid-state imaging device in consideration of position information on the wafer 12. The inspection device (not shown) used here is installed with a program that encodes the processes of the following steps S21 to S28. The flow of the inspection process will be described below according to the step numbers shown in the figure.
In step S21, the image signal of the entire image is acquired by the inspection apparatus, as in step S1 of the first embodiment.

Next, in step S <b> 22, position information indicating where the chip 10 having acquired the entire image is located on the wafer 12 is acquired. The position information may be expressed by, for example, an orientation flat or an XY coordinate based on the wafer center.
Next, in step S23, a plurality of partial images are generated from the entire image by a method corresponding to the position on the wafer 12 (registered in advance). For example, for the chips 10 near the outer periphery of the wafer 12 such as the chips 10A, 10B, 10C, and 10D in FIG. 1, a partial image is generated in the same manner as in the second embodiment (also in the method of the first embodiment). Good). Further, for example, for a chip close to the center of the wafer 12 such as a chip 10 at a position included in four dotted ellipses in FIG. 1, a partial image is generated with the same pattern as in the third embodiment.

Next, in step S24, inversion processing is performed to align the image area closest to the center of the entire image in each partial image generated in step S23 at a predetermined position.
Next, in step S25, the partial images after the processing in step S24 are averaged to generate a reference image.
Next, in step S26, an inspection image is generated for each partial image based on the difference between each partial image after the reversal process (the partial image that is not subjected to the reversal process) and the reference image.

Next, in step S27, pattern inspection according to the position on the wafer 12 (registered in advance) is performed on the inspection image to detect the appearance and degree of fixed pattern noise, and whether the chip 10 is a good product. Determine whether or not.
Next, in step S28, it is determined whether or not there is an uninspected chip 10 in the wafer 12. If not, the inspection for the wafer 12 is terminated. If there is an uninspected chip 10, the process returns to step S1 to inspect the uninspected chip 10. The above is the description of the flow of the inspection process in FIG. As described above, if the pattern matching method for the partial image, the generation method, and the inspection image is changed according to the position of the chip 10 on the wafer 12, the inspection accuracy can be improved.

<Fourth Embodiment>
In the fourth embodiment, the present invention is applied to an electronic camera. Specifically, after generating an inspection image for the chip of the solid-state imaging device by the above-described inspection method, based on the inspection image, generate an image processing program that erases fixed pattern noise unique to the solid-state imaging device, The image processing program and the solid-state image sensor are mounted on an electronic camera.

  Here, as an example, the solid-state imaging device has a color filter array of three primary colors. In the inspection step, for example, it is desirable to irradiate the chip of the solid-state imaging device 120 with uniform white light in the state of the wafer to generate the entire image for each of the three primary colors. That is, the entire image is generated based on the image signal of the pixel that selectively receives the red component light, and the same applies to the green component and the blue component. Then, after generating a partial image, a reference image, and an inspection image for each of the three primary colors, an image processing program for erasing fixed pattern noise specific to the solid-state imaging device is created based on the inspection image.

  FIG. 12 is a block diagram of the electronic camera 108 in the fourth embodiment. As shown in the figure, the electronic camera 108 includes a photographing lens 112, a shutter 116, a solid-state imaging device 120, an analog signal processing unit 124, an A / D conversion unit 128, a timing generator 132, and a shutter driving unit 136. A focus control unit 140, an operation unit 144, an MPU (Micro Processor Unit) 148, a system bus 152, an image processing unit 156, a memory 160, a recording / reading unit 164, and an exchangeable recording medium 168. A monitor control unit 184 and a liquid crystal monitor 188.

  The focus control unit 140 adjusts the focus position of the photographic lens 112 in accordance with a command from the MPU 148. The shutter control unit 136 controls the travel of the front curtain and rear curtain of the shutter 116 in accordance with an instruction from the MPU 148. The timing generator 132 drives the solid-state image sensor 120 in accordance with a command from the MPU 148. The solid-state image sensor 120 accumulates signal charges corresponding to the amount of received light and outputs an analog image signal. The analog signal processing unit 124 performs clamp processing, sensitivity correction processing, A / D conversion, and the like on the image signal from the solid-state imaging device 120 to generate digital image data.

  The image processing unit 156 performs image processing on the image data in accordance with an instruction from the MPU 148. The memory 160 temporarily stores image data before data conversion and processing into a predetermined format. The MPU 148 performs system control of the electronic camera 108 via the system bus 152. The operation unit 144 includes an exposure condition setting button group, a release button, and the like. The signal processing unit described in the claims corresponds to the functions of the analog signal processing unit 124, the A / D conversion unit 128, the timing generator 132, and the MPU 148 that controls these to generate image data.

  The main feature of the electronic camera 108 of the fourth embodiment is that the above-described image processing program for erasing fixed pattern noise is installed in the MPU 148, and the other configuration is the same as that of the conventional electronic camera. Since this image processing program is created in accordance with the solid-state image sensor 120, the image quality is improved. Create a program that erases fixed pattern noise in accordance with the solid-state imaging device based on the inspection image, and install it in the electronic camera, as in this embodiment, and the solids that have been below the standard of good products until now Even when the image sensor is used in an electronic camera, sufficient image quality can be obtained. In that case, the manufacturing yield is improved.

  As described above in detail, the present invention is greatly applicable to a solid-state image sensor inspection method, a solid-state image sensor inspection program, and an electronic camera.

It is a plane schematic diagram of the semiconductor wafer in which the chip | tip of the some solid-state image sensor was formed. It is explanatory drawing which shows an example of radial coating nonuniformity, and is a plane schematic diagram paying attention to four chip | tip which attached | subjected the code | symbol of A, B, D, D in FIG. It is explanatory drawing which shows an example of the image process which is the principle of this invention, and respond | corresponds to the chip | tip 10A. It is explanatory drawing which shows an example of the image processing which is the principle of this invention, and respond | corresponds to the chip | tip 10B. It is explanatory drawing which shows an example of the image processing which is the principle of this invention, and respond | corresponds to chip | tip 10D. It is explanatory drawing which shows an example of the image processing which is the principle of this invention, and respond | corresponds to chip | tip 10D. It is explanatory drawing of the test | inspection image used in order to determine whether a solid-state image sensor is good or bad. It is a flowchart of the inspection method of the solid-state image sensor of a 1st embodiment. It is explanatory drawing which shows the production | generation method of the partial image and reference image in 2nd Embodiment. It is explanatory drawing which shows the production | generation method of the partial image and reference image in 3rd Embodiment. It is a flowchart in the case of test | inspecting a solid-state image sensor in consideration of the positional information on a wafer. It is a block diagram which shows schematic structure of the electronic camera in 4th Embodiment. It is explanatory drawing which shows an example of the prior art which detects fixed pattern noise. It is explanatory drawing which shows an example of the fixed pattern noise which generate | occur | produces because a foreign material mixes during spin coating and becomes a nucleus and the area | region where the coating amount is thin arises.

Explanation of symbols

10, 10A, 10B, 10D, 10D Chip 12 Wafer 14 Street 20 Pixel area 22 Pad 108 Electronic camera 112 Shooting lens 116 Shutter 120 Solid-state image sensor 124 Analog signal processing unit 128 A / D conversion unit 132 Timing generator 136 Shutter drive unit 140 Focus control unit 144 Operation unit 148 MPU
152 System Bus 156 Image Processing Unit 160 Memory 164 Record Reading Unit 168 Recording Medium 184 Monitor Control Unit 188 Liquid Crystal Monitor

Claims (6)

Obtaining an image signal generated by the solid-state imaging device by irradiation with illumination light for inspection;
An extraction step of generating a plurality of partial images with respect to the whole image with the same image size using the whole image which is an image indicated by the image signal;
An inversion step of performing at least one of upside down, left and right inversion, and rotation so that the image size does not change for at least one of the partial images;
After the inversion step, a plurality of the partial images are combined to generate a reference image having the same image size as the partial images;
Generating, for each of the plurality of partial images, an inspection image corresponding to a difference from the reference image;
And a step of determining whether or not the solid-state imaging device is a non-defective product based on the inspection image. In the inspection method of the solid-state image sensor according to claim 1,
When all the partial images are overlaid, in the inversion step, upside down, left and right inversion, so that the position of the image area that is a part of the partial image and closest to the center of the whole image is aligned. A method for inspecting a solid-state imaging device, comprising performing at least one of rotation processing. In the inspection method of the solid-state image sensor according to claim 2,
In the extraction step, a position that is point-symmetric with respect to the center of the whole image so that a part of the outer circumference of each partial image becomes a part of the outer circumference of the whole image An even number of the partial images are generated such that the partial image exists in the solid-state imaging device. In the inspection method of the solid-state image sensor according to claim 3,
In the extraction step, four partial images are generated by dividing the whole image into four so that the center of the whole image is a corner of each of the partial images. Method. An inspection program for determining whether a solid-state image sensor is a non-defective product,
Obtaining an image signal generated by the solid-state imaging device by irradiation with illumination light for inspection;
An extraction step of generating a plurality of partial images with respect to the whole image with the same image size using the whole image which is an image indicated by the image signal;
An inversion step of performing at least one of upside down, left and right inversion, and rotation so that the image size does not change for at least one of the partial images;
After the inversion step, a plurality of the partial images are combined to generate a reference image having the same image size as the partial images;
Generating, for each of the plurality of partial images, an inspection image corresponding to a difference from the reference image;
An inspection program for causing a computer to execute a step of determining whether or not the solid-state imaging device is a non-defective product based on the inspection image. An electronic camera comprising a solid-state imaging device, a signal processing unit that generates image data based on an image signal output from the solid-state imaging device, and an image processing unit that performs image processing on the image data,
The image processing is processing for reducing fixed pattern noise estimated from the inspection image generated for the solid-state imaging device by the solid-state imaging device inspection method according to any one of claims 1 to 4. An electronic camera characterized by being.

Images