TWI691715B - Automatic optical detection mechanism for detecting silicon wafer defects and method thereof - Google Patents

Abstract

An automatic optical detection mechanism for detecting silicon wafer defects and method thereof are provided. The detection mechanism includes a base, an illumination unit, an image capturing unit, and an image identification unit. A target silicon wafer is placed on the base. The illumination unit provides the light needed for illuminating the silicon wafer, and the image capturing unit captures a target image from the surface of the silicon wafer. The image identification unit receives the target image and identifies the abnormal parts in the image as defects. Therefore, the present invention automatically detects the defects of the silicon wafer.

Description

Automatic optical detection mechanism and method for detecting silicon wafer defects

The invention relates to a technology for detecting silicon wafer defects, in particular to an automatic optical detection mechanism and method for detecting silicon wafer defects.

Silicon wafers are round wafers made by cutting silicon rods, and after the silicon rods are cut into silicon wafers, there will be a lot of residue on the surface. Even through the polishing and cleaning processes, they will still be in Defects such as scratches or water stains exist on the surface of the silicon wafer. Therefore, the surface of the silicon wafer must be inspected for defects to ensure the shipment quality of the silicon wafer.

Regardless of whether it is a silicon wafer manufacturing plant or a recycled silicon wafer plant, the current method of detecting defects on the surface of silicon wafers is still manual detection, mainly to detect defects on the surface of silicon wafers through the experience of inspectors, but using The method of manually detecting the defects on the surface of silicon wafers is easy to misjudge. The defects of the silicon wafers are still found due to the fact that no defects are found manually, resulting in uneven quality.

The aforementioned problem of manually detecting the surface defects of the silicon crystal surface, especially when the surface of the silicon wafer is scratched, even if the inspector can find the defects of the scratch, but only the energy can get the length of the scratch, and the depth of the scratch cannot be judged For example, if the defects are removed by grinding and polishing, the removal amount is also estimated by experience and it is difficult to be precise. It may happen that the depth of the grinding and polishing is not enough, so that the defect removal cannot be completed at one time, so the efficiency of the silicon wafer process The problem of effective promotion. In addition to this, it is also possible that the thickness of the silicon wafer is reduced due to the excessive depth of grinding and polishing, which wastes silicon wafer material.

In order to solve the above-mentioned problems, the present invention provides an automatic optical inspection mechanism and method for detecting defects on a silicon wafer, which mainly uses automatic optical inspection technology to detect defects on the surface of the silicon wafer, which is different from manual inspection and has better accuracy.

An embodiment of the present invention provides an automatic optical inspection mechanism for detecting silicon wafer defects, which includes a base, an illumination unit, an image capture unit, and an image determination unit. The base can be used to detect defects The silicon wafer is set, the light unit corresponds to the base and provides the illumination of the silicon wafer, the image capture unit and the light unit are set on the same side of the silicon wafer, to capture the surface of the silicon wafer under the illumination of the light unit An image to be tested; the image determining unit is electrically connected to the image capturing unit, and after the image determining unit receives the image to be tested, it determines whether the image to be tested has image differences and is defective.

Among them, a horizontal displacement unit is further connected to the base, and the base is controlled to be displaced in the horizontal direction.

Among them, a steering unit is further included, which is connected to the base and the horizontal displacement unit, so that the silicon wafer on the base can be tilted to different angles for the image capturing unit to capture images of different angles of the silicon wafer to be tested.

Among them, the light brightness of the light unit when illuminating is greater than 10000 lumens.

Among them, a defect image database is further included. The defect image database is electrically connected to the image judgment unit, and stores a plurality of image judgment units corresponding to the defect images of the silicon wafer for the image judgment unit to judge the silicon wafer defects.

Wherein, a defect removing unit is further included to remove the defect.

An embodiment of the present invention provides an automatic optical inspection method for detecting defects of a silicon wafer. The step is to dispose the silicon wafer to be detected on the base, and the illumination unit provides illumination of the silicon wafer above the base , So that the surface of the silicon wafer has enough brightness to highlight defects, and the surface of the silicon wafer is illuminated by the image capture unit to capture the image to be tested on the surface of the silicon wafer above the base, and the image judgment unit receives the pending The image is measured, and the image judgment unit judges that there is an image abnormality in the image to be tested as a defect.

Among them, the position of the image capturing unit above the base is fixed, and the base can be displaced in a horizontal direction along a first axis and a second axis, the first axis is perpendicular to the second axis, and the base drives the standby The measured silicon wafer is horizontally displaced along the first axis and the second axis, and after the partial image of the silicon wafer to be measured is segmentally captured by the image capturing unit, the image determination unit integrates and restores the image to be tested .

Wherein, the pedestal has a rotation axis parallel to the horizontal direction, and the pedestal can be tilted to different angles with the rotation axis as the axis. When the defect has a depth, the silicon wafer is adjusted by the image capturing unit The angle of the captured image is used to capture a plurality of defect images of the defect with depth at different angles, and the image determination unit determines the actual depth of the defect based on the plurality of defect images.

Wherein, the image judgment unit is electrically connected with the defect image database to store the image of the image to be tested which is judged to be defective in the image judgment unit, so that the image judgment unit can judge the silicon wafer defects.

Wherein, when a defect image is generated, the defect image database can store the defect image by classification, and the image judgment unit automatically matches and compares the defect type according to the classification of the defect image.

Wherein, a defect removal unit is further used to remove the defect according to at least one of grinding, polishing or etching according to the defect type.

In this way, when there is a defect on the surface of the silicon wafer, the defect can be detected by the automatic optical inspection mechanism of the present invention. After the image judgment unit receives the image to be tested, it is determined whether there is an image difference between the image to be tested, and then it can be accurate To determine whether there is a defect on the surface of the silicon wafer, in order to solve the problem that the defect still exists due to the manual discovery of the defect when the silicon wafer is shipped out, to ensure that the quality of the silicon wafer is free of defects when shipped out of the factory.

In addition, when the surface of the silicon wafer is scratched, the silicon wafer to be detected can be tilted to different angles with the pedestal, and the angle of the image captured by the capturing unit can be adjusted, and then a plurality of different angles are captured The defect image determines the actual depth of the defect. Moreover, after the actual depth of the scratch is known, the defect removal unit can remove the defects at a time according to the known depth, so as to achieve the efficiency of the silicon wafer process can be effectively improved.

Furthermore, the defect image database stores the defect images determined as defective by the image determination unit, and classifies and stores the obtained defect images. When the obtained defect images are sufficient and complete, the image determination unit can Similar to the method of artificial intelligence, according to the classification of the defect image, the automatic match-merge is used to compare the defect types to achieve the effect of accurately determining the defect type and suggesting a better removal method.

In order to facilitate the description of the central idea of the present invention in the column of the above summary of the invention, it is expressed in a specific embodiment. In the embodiments, various objects are drawn according to the proportion, size, deformation or displacement suitable for description, rather than drawn according to the proportion of actual elements, which will be described first.

Please refer to FIGS. 1 to 4, the present invention provides an automatic optical inspection mechanism and method for detecting silicon wafer defects. The automatic optical inspection mechanism 100 includes a base 10, an illumination unit 20, and an image capture unit 30. And an image determination unit 40, in this embodiment further includes a defect image database 50 and a defect removal unit 60, wherein:

The base 10 is provided for the silicon wafer 200 to be inspected for defects. In this embodiment, the base 10 is connected to a horizontal displacement unit 11. The horizontal displacement unit 11 is provided on a base 12 in this embodiment, and is along a first axis X and a first axis in the horizontal direction In the two-axis Y displacement, the horizontal displacement unit 11 controls the base 10 to be displaced in the horizontal direction, wherein the first axis X and the second axis Y are perpendicular to the horizontal direction (as shown in FIG. 3 ). In this embodiment, a steering unit 13 is included. The steering unit 13 is connected to the base 10 and the horizontal displacement unit 11. In addition to the horizontal displacement unit 11, the base 10 can be horizontal along the first axis X and the second axis Y In addition to the displacement, the silicon wafer 200 on the base 10 can be tilted to different angles by the turning unit 13.

The illumination unit 20 is provided corresponding to the base 10 to provide illumination of the silicon wafer 200 on the base 10. In this embodiment, the brightness of the light L of the illumination unit 20 when illuminating is greater than 10,000 lumens, preferably 50,000 lumens. In the preferred embodiment, the light unit 20 may be a high power LED. In this embodiment, the number of the illuminating units 20 is two. The two illuminating units 20 are disposed on opposite sides of the base 10 and illuminate the silicon wafer 200 on the base 10 at an angle of 45 degrees.

The image capturing unit 30 is disposed on the same side of the silicon wafer 200 as the light illuminating unit 20. The surface of the silicon wafer 200 can be captured by the image capturing unit 30 under the illumination of the light illuminating unit 20 31 (the dotted line C indicates the image capturing direction). The image capturing unit 30 may be a CCD camera, a CMOS camera, or a similar function image capturing device. In this embodiment, a CCD camera is the preferred embodiment. Since the steering unit 13 is provided in this embodiment, when the silicon wafer 200 on the base 10 is tilted to different angles with the steering unit 13, the image capturing unit 30 can capture images of the silicon wafer 200 at different angles.

The image determining unit 40 is electrically connected to the image capturing unit 30, and after the image capturing unit 30 captures the image to be tested 31, the image determining unit 40 receives the image to be tested 31, and the image determining unit 40 It is determined whether the image to be tested 31 is defective due to image difference. The image determination unit 40 is executed by the central processing unit of the computer device.

The defect image database 50 is connected to the image determination unit 40 and stores a plurality of image determination units 40 corresponding to the defect images of the silicon wafer 200. The defect image refers to the image 31 to be tested is determined by the image determination unit 40 Images of defective parts. These defective images are stored in the defect image database 50 for the image determination unit 40 to determine the defect of the silicon wafer 200.

The defect removal unit 60 is used to remove defects on the silicon wafer 200, and may be at least one of a grinding machine, a polishing machine, and an etching machine or a combination of two or more (not shown). Among them, the amount of removal when the grinding machine grinds the surface of the silicon wafer 200 is large, the removal amount when the polishing machine grinds the surface of the silicon wafer 200 is small, and the etching machine is the surface of the silicon wafer 200 Perform trace removal.

With regard to the above-mentioned automatic optical inspection mechanism 100 for detecting silicon wafer defects, the detection method is described in this embodiment. When detecting the defects of the defective silicon wafer 200, the silicon wafer 200 to be detected is placed on the base 10, and the illumination unit 20 provides the illumination of the silicon wafer 200 above the base 10, so that the surface of the silicon wafer 200 has sufficient brightness to highlight defects, and the surface of the silicon wafer 200 is under the illumination of the illumination unit 20 , The image capturing unit 30 captures the image to be tested 31 on the surface of the silicon wafer 200 above the base 10, and after the image determining unit 40 receives the image to be tested 31, it is determined that there is an image abnormality in the image to be tested 31 as a defect . In this embodiment, the light unit 20 illuminates the surface of the silicon wafer 200 with high-intensity light L of 50,000 lumens.

In this embodiment, the size of the silicon wafer 200 is larger than the range of images that the image capture unit 30 can capture at one time. Therefore, the position of the image capture unit 30 in this embodiment above the base 10 is fixed, so When the defect detection of the silicon wafer 200 is performed, the method of this embodiment is that the silicon wafer 200 is displaced along a path P along the first axis X and the second axis Y along the base 10 on the horizontal displacement unit 11 ( As shown in FIG. 3), during the displacement of the silicon wafer 200 along the path P, the partial image of the silicon wafer 200 is segmentally captured by the image capturing unit 30, and then integrated and restored to standby by the image determining unit 40测image31. After being judged by the image judgment unit 40, the image 31 to be tested includes a defect image D1 and a defect image D2, wherein the defect image D1 is a scratch 201 on the silicon wafer 200, and the defect image D2 is the surface of the silicon wafer 200 Water stains.

The defect image D1 has a certain depth. In the detection method of this embodiment, the base 10 is connected to the steering unit 13, and the steering unit 13 has a rotation axis A parallel to the horizontal direction. The steering unit 13 can be tilted to different angles with the rotation axis A as the axis. As shown in FIG. 4, the scratch 201 on the silicon wafer 200 is inclined laterally in the depth direction. If the image capturing unit 30 is directly facing to capture the scratch 201, the true depth cannot be measured. In the detection method of this embodiment, the steering unit 13 is tilted to adjust the angle at which the silicon wafer 200 is captured by the image capture unit 30 to capture multiple scrapes at different angles (as shown in FIG. 4) After the defect image D1 of the flaw 201 is received, and the image determination unit 40 of the plurality of defect images D1 is received, the image determination unit 40 performs calculation according to the plurality of defect images D1 to determine the actual depth of the scratch 201.

As mentioned above, when the actual depth of the scratch 201 is known, in this embodiment, the defect removing unit 60 can remove the scratch 201 once. As mentioned above, the defect removal unit 60 may be a grinder, a polisher, or an etching machine. If the depth of the scratch 201 is deep, grinding, polishing, and etching may be performed according to the depth of the scratch 201, respectively. The scratch 201 is removed; if the depth of the scratch 201 is shallow, it can still be polished or etched, or the scratch 201 can be removed at once. If it is the water stain of the defect image D2, it can be removed by etching at a time.

In the detection method of this embodiment, the image to be tested 31 is determined by the image determination unit 40 as a defective defective image. The defective image can be stored through the defective image database 50, and the defective images are stored in categories. For example, the aforementioned defective image D1 is classified as a scratched image according to its image, and the defective image D2 is classified as a water stained image according to its image. When the image capture unit 30 captures the next image 31 to be tested, the image determination unit 40 can classify the defect image, and the automatic match combination compares the defect type to which it belongs. If it is similar to the defect image D1, it can be directly determined It is a scratch, and if it is similar to the defect image D2, it can be directly determined as a water stain. At this time, the defect removal unit 60 can perform a removal operation on the defect in a corresponding defect removal manner.

From the above description, it is not difficult to find the characteristics of the present invention, which are:

1. When there is a defect on the surface of the silicon wafer 200, the automatic optical inspection mechanism 100 of the present invention can automatically detect the defect of the silicon wafer 200, that is, the image judgment unit 40 receives the image 31 to be tested to determine whether there is an image difference. Determine the presence or absence of defects on the surface of the silicon wafer 200 fairly accurately. Compared with the conventional manual inspection, it can solve the problem that the silicon wafer 200 still has defects due to no defects found manually. No defects and consistent quality.

2. When the defects on the surface of the silicon wafer 200 are scratches, the silicon wafer 200 can be tilted with the base 10 to detect the depth, so as to capture a plurality of defect images at different angles to determine the actual depth of the defect. After the actual depth of the defect is known, the defect removal unit 60 can remove the defect at a time according to the known depth, so as to achieve the efficiency of the silicon wafer process can be effectively improved.

3. In the present invention, defective images (such as defective images D1 and D2) that are determined to be defective by the image determining unit 40 can be stored and classified by the defective image database 50, and the defects obtained as the number of inspections increases When the images are sufficient and complete, the image determination unit 40 can automatically match the defect types according to the classification of the defective images in a manner similar to artificial intelligence, so as to accurately determine the defect type. The effect of the best removal method.

The above-mentioned embodiments are only used to illustrate the present invention, not to limit the scope of the present invention. Any modifications or changes that do not violate the spirit of the present invention are within the scope of the invention to be protected.

100: automatic optical detection mechanism 200: silicon wafer 201: Scratch 10: Dock 11: Horizontal displacement unit 12: Base 13: steering unit 20: Illumination unit 30: Image capture unit 31: Image to be tested 40: Image judgment unit 50: Defect image database 60: Defect removal unit X: first axis Y: Second axis L: light C: dotted line P: path D1: Defective image D2: Defective image A: Rotation axis

FIG. 1 is a schematic diagram of the automatic optical detection mechanism of the present invention. 2 is a block diagram of the automatic optical detection mechanism of the present invention. FIG. 3 is a schematic view of the horizontal displacement of the base of the present invention to capture images in stages by the image capturing unit. FIG. 4 is a schematic diagram of the automatic optical inspection mechanism of the present invention capturing defect images when the silicon wafer is tilted to different angles.

100: automatic optical detection mechanism

200: silicon wafer

10: Dock

11: Horizontal displacement unit

12: Base

13: steering unit

20: Illumination unit

30: Image capture unit

X: first axis

L: light

C: dotted line

A: Rotation axis

Claims (9)

An automatic optical inspection mechanism for detecting defects of a silicon wafer includes: a base for setting a silicon wafer to be detected for defects; an illumination unit corresponding to the base and providing illumination of the silicon wafer; An image capturing unit, which is arranged on the same side of the silicon wafer as the illumination unit, to capture an image to be measured on the surface of the silicon wafer under the illumination of the illumination unit; an image determination unit, which Electrically connect the image capturing unit to receive the image to be tested, and determine whether the image to be tested is defective due to image differences; a horizontal displacement unit connected to the base to control the horizontal displacement of the base ; And a steering unit, which is connected to the base and the horizontal displacement unit, so that the silicon wafer on the base can be tilted to different angles for the image capture unit to capture the difference between the silicon wafer to be tested Angle image. The automatic optical inspection mechanism for detecting defects on a silicon wafer according to claim 1, wherein the light intensity of the illumination unit when illuminating is greater than 10,000 lumens. The automatic optical inspection mechanism for detecting silicon wafer defects according to claim 1, further comprising a defect image database electrically connected to the image judgment unit and storing a plurality of image judgment units corresponding to silicon The defect image of the wafer is used by the image judgment unit to judge the defect of the silicon wafer. The automatic optical inspection mechanism for detecting defects on a silicon wafer according to claim 1, further comprising a defect removal unit to remove the defects. An automatic optical inspection method for detecting defects on a silicon wafer includes the following steps: setting a silicon wafer to be detected on a susceptor; providing a illuminating unit to illuminate the silicon wafer above the susceptor To make the surface of the silicon wafer There is enough brightness to highlight the defect; the surface of the silicon wafer is under the illumination of the light unit, and an image capturing unit is used to capture one of the images of the silicon wafer surface to be tested above the base; and An image determination unit electrically connected to the image capture unit receives the image to be tested and uses the image determination unit to determine that there is an image abnormality in the image to be tested as a defect; wherein the image capture unit is above the base The position of the base is fixed, and the base can be displaced along a first axis and a second axis in the horizontal direction, the first axis is perpendicular to the second axis, and the silicon wafer to be tested is driven by the base Horizontal displacement along the first axis and the second axis, and partial image of the silicon wafer to be tested is segmentally captured by the image capturing unit, integrated by the image determining unit and restored to the test Image; wherein, the pedestal has a rotation axis parallel to the horizontal direction, the pedestal can be tilted to different angles with the rotation axis as the axis, and the silicon wafer is adjusted when the defect has a depth The angle of the image captured by the image capturing unit is used to capture a plurality of defect images of the defect with depth at different angles, and the image determining unit determines the actual depth of the defect based on the plurality of defect images. The automatic optical inspection method for detecting defects of a silicon wafer according to claim 5, wherein the light brightness of the illumination unit when illuminating is greater than 10,000 lumens. The automatic optical inspection method for detecting silicon wafer defects according to claim 5, wherein a defect image database is electrically connected to the image judgment unit to store the image to be tested and judged to be defective in the image judgment unit The defect image is used by the image determination unit to determine the silicon wafer defect. The automatic optical inspection method for detecting silicon wafer defects according to claim 7, wherein, when one of the aforementioned defect images is generated, the defect image database classifies and stores the defect images, and the image judgment unit can be based on the defect image Classification and automatic matchmaking compare the defect types to which they belong. The automatic optical inspection method for detecting silicon wafer defects according to claim 8, wherein a defect removal unit is further polished and polished according to the defect type Or at least one of etching removes the defect.

Images