JPH09196859A - Surface defect inspection device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To detect surface defect of a silicon wafer and judge its sort. SOLUTION: When a white illumination light 1 is spectrally separated by a prism 2 and cast on a surface to be inspected 5 upon convergence made by a condenser lens 4, a plurality of beams of light having different wavelengths are cast on the surface 5 at different angles. A two-dimensional color sensor 7 photographs the surface 5 through a photographing lens 6 and outputs color image signals ar, ag, ab of R, G, B. In case where the surface includes a defect having a gradually inclining angle, only the light of red wavelength projected at an angle nearest the perpendicular is reflected in the direction of the photographing lens 6, so that reddish photo is obtained. In case where the surface includes a defect having a large inclination angle, the violet rays projected at an angle nearest the horizontal are photographed. That is, each defect is photographed in different color depending upon the sort of the defect. Thus the defect detection is conducted by an image processing part 8, and the sort of the defect is judged according to the color information.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a surface defect inspection apparatus, and more particularly to a surface defect inspection apparatus for detecting defects existing on the surface of a silicon wafer or the like.

[0002]

2. Description of the Related Art As a conventional technique, for example, Japanese Patent Laid-Open No. 62-62
There is a surface defect inspection apparatus described in JP-A-38348.

FIG. 5 is a block diagram showing a conventional surface defect inspection apparatus. The surface defect inspection apparatus shown in FIG. 5 includes a red light source 1 that illuminates a surface 5 to be inspected from an angle substantially perpendicular to the surface 5.
1, the blue light source 12 that illuminates from an oblique direction, and the surface 5 to be inspected are imaged through the image pickup lens 6 at a substantially vertical angle, and are internally dispersed into respective wavelengths of the red light source 11 and the blue light source 12. Output corresponding video signals ar and ab 2
The three-dimensional color sensor 7 and the image processing unit 8 that detects defects in the image signals ar and ab by image processing.

Next, the operation will be described.

On the surface 5 to be inspected, defects having surfaces having different inclination angles are mixed. That is, "dent"
And a defect having a surface with a large inclination angle such as “scratch”. here,
Since the red light source 11 is illuminated at a substantially vertical angle,
In the case of light of this wavelength, only reflected light from a defect having a surface with a gentle inclination angle is imaged. On the contrary, blue light source 1
Only defects with a large tilt angle surface are imaged at the two wavelengths. Therefore, the video signal ar includes only an image of a defect having a surface having a gentle inclination angle, and the video signal ab includes only an image of a defect having a surface having a large inclination angle. Therefore, this video signal ar,
By performing the defect detection process in the image processing unit 8 for each of the abs, a plurality of types of defects are simultaneously and selectively detected.

[0006]

The above-described conventional surface defect inspection apparatus uses a plurality of illumination light sources having different wavelengths in order to inspect a plurality of types of defects having different surface inclination angles simultaneously and selectively. Irradiate from different angles,
Spectral imaging is performed by a color camera, and defect detection processing is separately performed for each of a plurality of illumination wavelengths.

Therefore, it is necessary not only to adjust the illumination angle to the optimum one according to the type of defect, but also because the illumination angle is discrete, some defects may be overlooked. In addition, in order to structure the defect type selection capability, it is necessary to increase the number of light sources and the number of image pickup / detection means, which has a drawback that the configuration becomes complicated.

[0008]

The surface defect inspection apparatus of the present invention is different in that it separates the white illuminating light into continuous spectrums for each wavelength, and collects the illuminating light separated by the spectral means. Converging means for irradiating the same region of the surface to be inspected with light of different wavelengths at different angles, color imaging means for imaging the surface to be inspected and outputting a plurality of color video signals, and receiving the color video signal. And an image processing unit that detects a defect on the surface to be inspected and determines the type of the defect, the spectroscopic unit can use a spectral prism or a diffraction grating, and the condensing unit is an optical lens, or
A parabolic mirror can be used, and the color imaging means is R
Color video signals of (red), G (green) and B (purple) can be output.

Further, the image processing section outputs a plurality of image memories for storing digital image data for each of the plurality of color video signals, and added image data obtained by adding data of pixels at the same positions of the plurality of image memories. An adder, a binarization processing unit that outputs binary image data obtained by binarizing the added image data, and a set of labeled pixels of an image composed of the binary image data, which has a predetermined area or more. A labeling processing unit that detects a defect as a defect and outputs defect coordinate data that is the coordinate of the defect, and compares the size of the pixel data of the coordinates indicated by the defect coordinate data of each of the plurality of image memories to determine the defect. It can also be configured with a decoder that determines the type and outputs defect type data.

[0010]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Next, the present invention will be described in detail with reference to the drawings.

FIG. 1 is a block diagram showing a surface defect inspection apparatus according to a first embodiment of the present invention.

In the surface defect inspection apparatus shown in FIG. 1, a spectral prism 2 as a spectroscopic means for continuously spectrally splitting the white illumination light 1 for each wavelength and a multi-wavelength illumination light 3 split by the spectral prism 2 are in the same area. By condensing light on the surface 5 to be inspected at different angles by condensing the light onto the surface 5 to be inspected, and the surface 5 to be inspected through the imaging lens 6 in the vertical direction. Imaged, R (red), G
(Green), B (purple) color video signals ar, a
a two-dimensional color sensor 7 as a color image pickup means for outputting g, ab, an image processing unit 8 for receiving a color video signal ar, ag, ab and detecting a defect and selecting a defect type based on color information. It is configured to include.

FIG. 2 is a block diagram showing the inside of the image processing unit 8.

The image processing section 8 includes a color video signal ar, a
Three A / D converters 81 for A / D converting g and ab and outputting digital image data br, bg, and bb.
And image data br, bg, bb are stored respectively 3
One image memory 82, an adder 83 that adds the data of the same address of the three image memories 82 and outputs the added image data c, and a binary that binarizes the added image data c and outputs the binary image data d The conversion processing unit 84 and the labeling processing unit 85 that measures the coordinates and area of the defect based on the binary image data d and outputs the defect detection data e and the defect coordinate data f.
And the image data br, bg, bb at the address indicated by the defect coordinate data f are normalized with reference to one of these data (for example, image data br), and the normalized data g
The defect type data h based on the ratio of the normalization processing unit 86 that outputs r, gg, gb and the normalized data gr, gg, gb.
And a decoder 87 for outputting

Next, the operation will be described.

FIG. 3 is a schematic view showing a cross section of three types of defects mixed on the surface 5 to be inspected and having surfaces with different inclination angles. Of the defects shown in FIG. 3, the dent 9
1 is a defect having a surface having a gentle inclination angle, and the shallow scratch 92 is a defect having a surface having an intermediate inclination angle,
The deep scratch 93 is a defect having a surface with a large inclination angle.

In FIG. 1, since the white illumination light 1 includes each wavelength from red to purple, when it enters the spectroscopic prism 2, it is continuously dispersed into each wavelength and is multi-wavelength illumination light 3.
Is emitted as. Since this multi-wavelength illumination light 3 spreads in a fan shape at different angles according to each wavelength at the time of being emitted from the spectral prism 2, when it is condensed by the condensing lens 4 and irradiated on the surface 5 to be inspected, On the gymnastics surface 5 to be inspected,
Light of different wavelengths will be emitted at different angles. In the case of the surface defect detection apparatus shown in FIG. 1, the red wavelength is emitted at the angle closest to the vertical, and the purple wavelength is emitted at the angle closest to the horizontal. Here, the condenser lens 4 is a lens in which chromatic aberration is corrected, and light of each wavelength illuminates the same area on the surface 5 to be inspected.

Next, the two-dimensional color sensor 7 picks up an image of the surface 5 to be inspected through the image pickup lens 6 and outputs R, G, B color image signals ar, ag, ab. At this time, since the dent 91 shown in FIG. 3 has a surface with a gentle inclination angle, only the light of the red wavelength emitted from the dent 91 at an angle that is the most vertical is reflected in the imaging lens 6 direction. It will be imaged in red. vice versa,
Since the deep scratch 93 has a surface with a large inclination angle, it is imaged in purple that is irradiated at an angle that is the most horizontal, and the shallow scratch 92 has a surface with a moderate inclination angle, and therefore is irradiated at a medium angle. The captured image is captured in green. That is, each defect is imaged in a different color according to its type. The edge portion of the defect reflects light of each wavelength and thus is imaged in white.

Therefore, next, the image processing unit 8 detects defects and sorts the defect types based on the color information.

In FIG. 2, color video signals ar, a
Digital image data br, bg, and bb obtained by A / D converting g and ab by the three A / D converting units 81 are stored in the three image memories 82. Next, the adder 83
Then, the data of the same address in the three image memories 82 are added, and the added image data c is binarized by the binarization processing unit 84. The labeling processing unit 85 labels each set of “1” pixels (or “0” pixels) connected to each other in the image composed of the binary image data C, and Among them, those having a certain area or more are recognized as defects, and when a set of such defective pixels is detected, the defect detection data e is output. Further, the labeling processing unit 85 outputs the center of gravity of the set of defective pixels as defective coordinate data f.

The normalization processing unit 86 normalizes the image data br, bg, bb at the address indicated by the defect coordinate data f so as to be represented by a ratio based on one of the data, for example, the image data br. . The decoder 87 compares the normalized data gr, gg, gb and compares the defect type data h
Is output. That is, when the normalized data gr has a value of 1, and the normalized data gg and gb are both smaller than 1, the two-dimensional color sensor 7 receives the most red light, so that the defect indicated by the defect coordinate data f indicates the surface. The defect type data h, which is a dent having a gentle inclination, is output. Similarly, the normalized data g out of the normalized data gr, gg, gb
Since the two-dimensional color sensor 7 receives the largest amount of green light when g is the largest, it is assumed that the defect is a shallow scratch with a moderate surface inclination, and the two-dimensional color sensor 7 is the one where the normalized data gb is the largest. Receives the largest amount of violet light, the defect type data h indicating that the defect is a deep scratch having a large surface inclination is output.

FIG. 4 is a block diagram of a surface defect inspection apparatus according to a second embodiment of the present invention.

The surface defect inspection apparatus shown in FIG. 4 has the same structure and operation as the defect inspection apparatus shown in FIG. 1 except that the diffraction grating 10 is used as the spectroscopic means.

In the present invention, a parabolic mirror can be used as the light collecting means in addition to the optical lens. A color line sensor other than the two-dimensional color camera may be used as the image capturing means, and the surface to be inspected may be moved at a constant speed with respect to the color line sensor to capture an image.

The defect coordinate data is not limited to the center of gravity of the pixel set, and for example, the upper left point of the labeled pixel set can be used.

[0026]

The surface defect inspection apparatus of the present invention irradiates from different angles using illumination light sources of different wavelengths, and instead of separately performing imaging and defect detection processing for each of the plurality of illumination wavelengths, white By continuously splitting light, light of different wavelengths is irradiated onto the surface to be inspected at different continuous angles, and the color image signals of a plurality of colors obtained by imaging with the color imaging means are used to detect the surface of the object to be inspected. Since the defect is detected and the defect type is determined, it is not necessary to adjust the illumination angle according to the defect type, and the defect is not overlooked. Further, there is an effect that the ability to discriminate defect types can be improved with a simple configuration.

[Brief description of drawings]

FIG. 1 is a configuration diagram of a surface defect inspection apparatus according to a first embodiment of the present invention.

FIG. 2 is a block diagram showing the inside of an image processing unit 8 in FIG.

3 is a cross-sectional view of an object to be inspected 5 schematically showing a plurality of types of defects detected by the surface defect inspection apparatus shown in FIG.

FIG. 4 is a configuration diagram of a surface defect inspection apparatus according to a second embodiment of the present invention.

FIG. 5 is a configuration diagram of a conventional surface defect inspection apparatus.

[Explanation of symbols]

 1 White Illumination Light 2 Spectral Prism 3 Multi-wavelength Illumination Light 4 Condenser Lens 5 Surface to be Inspected 6 Imaging Lens 7 Two-dimensional Color Sensor 8 Image Processing Unit 81 A / D Converter 82 Image Memory 83 Adder 84 Binarization Processing Part 85 Labeling processing part 86 Normalization processing part 87 Decoder 91 Depression 92 Shallow scratch 93 Deep scratch 10 Diffraction grating 11 Red light source 12 Blue light source ar, ag, ab Color video signal br, bg, bb Image data c Addition image data d 2 Value image data e Defect detection data f Defect coordinate data gr, gg, gb Normalized data h Defect type data

Claims (7)

[Claims] 1. A spectroscopic unit that continuously disperses white illumination light for each wavelength, and light of different wavelengths at different angles by converging the illumination light that has been dispersed by the spectroscopic unit on the surface to be inspected. Condensing means for irradiating the same area, color imaging means for imaging the surface to be inspected and outputting a plurality of color video signals, and detecting defects on the surface to be inspected by receiving the color video signals and An apparatus for inspecting a surface defect, comprising: an image processing unit for determining the type of defect. 2. The surface defect inspection apparatus according to claim 1, wherein the spectroscopic means comprises a spectroscopic prism. 3. The surface defect inspection apparatus according to claim 1, wherein the spectroscopic means is a diffraction grating. 4. The surface defect inspection apparatus according to claim 1, wherein the condensing means is an optical lens. 5. The surface defect inspection apparatus according to claim 1, wherein the light collecting means is a parabolic mirror. 6. The image processing unit outputs a plurality of image memories for storing digital image data for each of a plurality of color video signals, and added image data obtained by adding data of pixels at the same positions of the plurality of image memories. An adder, a binarization processing unit that outputs binary image data obtained by binarizing the added image data, and a set of labeled pixels of an image composed of the binary image data, which has a predetermined area or more. A labeling processing unit that detects a defect as a defect and outputs defect coordinate data that is the coordinate of the defect, and compares the size of the pixel data of the coordinates indicated by the defect coordinate data of each of the plurality of image memories to determine the defect. 6. The surface defect inspection apparatus according to claim 1, further comprising a decoder that determines a type and outputs defect type data. 7. The surface defect inspection apparatus according to claim 1, wherein the color image pickup means outputs R (red), G (green) and B (purple) color image signals.