TWI840217B - Wafer positioning system and method using Hough transform - Google Patents

Abstract

The present invention discloses a wafer positioning system and method using Hough transform, which includes a rotating mechanism, a control module, an image capture device and an information processing unit. The rotating mechanism includes a wafer carrier. The control module is used to control the wafer on the rotating platform to perform rotational positioning. The image capture device is arranged near the rotating mechanism to capture the image of the rotated and positioned wafer to form a wafer positioning image. The information processing unit includes an image recognition module, and uses an image positioning algorithm to calculate the tilt angle of the wafer positioning image after image processing to obtain the corresponding tilt angle value, and judge whether the tilt angle value exceeds the preset angle threshold. If the judgment result is yes, the corresponding compensation angle signal is output, and the compensation angle signal is transmitted to the control module, so that the control module drives the wafer on the rotating platform to make a correct rotational positioning according to the compensation angle signal, so that the wafer can obtain a more accurate angle positioning effect during rotational positioning, thereby solving the existing system defects such as the long detection time of the traditional wafer positioning system, thereby effectively shortening the recognition time and effectively improving the process efficiency of the wafer.

Description

Wafer positioning system and method using Hough transform

The present invention relates to a wafer positioning system and method using Hough transform, and in particular to a wafer positioning technology that can achieve a more accurate angle positioning effect when the wafer is rotated and positioned to solve the defects of the traditional wafer positioning system such as long detection time.

According to the semiconductor manufacturing process, automated equipment is used to load wafers into various inspection stations (such as reaction chambers, processing chambers, etc.) for corresponding processing steps. Traditional automated equipment uses robotic arms to repeatedly perform wafer handling tasks. Generally speaking, it is very important to accurately place the wafer at the optimal position of the process inspection station. The reason is that the accuracy of the center point position and azimuth angle of the wafer in the process inspection station does have a great impact on the process yield, so accurate positioning and identification can effectively improve the process efficiency of the wafer.

Most of the known wafer positioning technologies are to identify the notch on the edge of the wafer by human, manual, or visual methods; or to align the flat edge with an orientation indicator; however, this technology has a large error rate, slow positioning speed, and cannot accurately rotate the wafer to a predetermined angle according to the needs of the subsequent processing procedures, thus affecting the yield and efficiency of wafer processing; another known wafer positioning technology is to install a penetrating sensor on the edge of the wafer to detect whether there is a notch on the edge, and determine the wafer orientation based on the position of the notch? This penetrating sensing method is indeed affected by parameters such as wafer thickness, substrate material, and surface treatment, resulting in misjudgment.

Furthermore, as far as we know, the patent cases related to the present invention are as follows:

1. Invention Announcement No. I588932 "Wafer Positioning Identification Device and Method Thereof" comprises a wafer body, a plurality of marking pattern portions formed at different locations of the wafer body by non-destructive means, and a detection device. Each marking pattern has a different shape from the others. The detection device detects the marking pattern by sending a detection signal to the marking pattern, and the marking pattern returns a positioning signal to interpret the correct position and corresponding angle of the wafer body. Although the patent has a wafer positioning effect, it uses the position detection of the marking pattern and the light intensity reflection image to identify the wafer positioning, rather than performing tilt angle processing and calculation on the wafer positioning image, so that the tilt angle value cannot be obtained to determine whether the wafer angle is tilted, and thus the wafer positioning and identification effect cannot be achieved, resulting in the inability to effectively improve the wafer process efficiency.

2. Invention Announcement No. I662648 "Wafer Positioning Identification Device and Method Thereof" is used to identify and locate the correct position of a wafer when the wafer is rotating. The wafer positioning identification device comprises a base, which is a stable platform with at least one opening; a carrier, located on the base, for carrying the wafer; wherein the surface and bottom surface of the wafer have a first feature and a second feature respectively; a rotating mechanism, The wafer is combined with the base assembly, and when the rotating mechanism rises to support the wafer, a rotating force is provided to rotate the wafer; at least two light source modules are respectively arranged on the upper plane and the lower plane of the base; at least two image sensing modules are respectively arranged on the upper plane and the lower plane of the base; and a control circuit module is electrically connected to the rotating mechanism, the light source modules and the image sensing modules; wherein the control circuit module is used to control the light source modules and the image sensing modules located on the upper plane of the base to complete the recognition of the first feature, and the control circuit module simultaneously controls the light source modules and the image sensing modules located on the lower plane of the base to complete the recognition of the second feature through the opening; wherein, according to the recognition results of the first feature and the second feature, the control circuit module controls the rotating mechanism to stop rotating at the correct position to complete The positioning action of the wafer; although the patent has the effect of wafer positioning; however, it uses the position detection of the first and second features and the light intensity reflection image to identify the wafer positioning, rather than performing tilt angle processing and calculation on the wafer positioning image, so that the tilt angle value cannot be obtained to determine whether the wafer angle is tilted, and thus the precise positioning and identification effect of the wafer cannot be achieved, resulting in the inability to effectively improve the process efficiency of the wafer.

In view of this, the known wafer positioning technology and the patents are indeed not perfect in terms of technology and functionality, and there is still a need for further improvement; therefore, after continuous efforts and research and development, the inventors of the present invention finally developed a set of the present invention that is different from the above-mentioned known ball catching robot technology.

The main purpose of the present invention is to provide a wafer positioning system and method using the Hough transform, which is mainly to obtain a more accurate angle positioning effect when the wafer is rotated and positioned, so as to solve the existing system deficiencies such as the long detection time of the traditional wafer positioning system, so as to effectively shorten the positioning and identification time, thereby effectively improving the process efficiency of the wafer. The technical means used to achieve the main purpose of the present invention include a rotating mechanism, a control module, an image capture device and an information processing unit. The rotating mechanism includes a carrier for the wafer. The control module is used to control the rotational positioning of the wafer on the rotating platform. The image capture device is arranged near the rotating mechanism to capture the image of the rotated and positioned wafer and form an image as a wafer positioning image. The information processing unit includes an image recognition module, and uses an image positioning algorithm to calculate the tilt angle of the wafer positioning image after image processing to obtain the corresponding tilt angle value, and judge whether the tilt angle value exceeds the preset angle threshold. If the judgment result is yes, the corresponding compensation angle signal is output and the compensation angle signal is transmitted to the control module, so that the control module drives the wafer on the rotating platform to make a correct rotation positioning according to the compensation angle signal.

1: Viewing station

10: Rotating mechanism

11: Rotating platform

20: Control module

30: Image capture device

40: Information processing unit

41: Image recognition module

50: Wafer

51,51": Positioning mark structure

60:Signal transmission module

L: Straight line

Figure 1 is a schematic diagram of the implementation of the specific application architecture of the present invention.

Figure 2 is a schematic diagram of a functional block implementation of a circuit architecture of the present invention.

Figure 3 is a schematic diagram of the functional block implementation of another circuit architecture of the present invention.

Figure 4 is a schematic diagram of the present invention using a positioning mark structure to perform tilt angle calculation.

Figure 5 (a) is a schematic diagram of a binary wafer positioning image after removing the background of the wafer positioning image of the notch; (b) is a schematic diagram of the positioning mark feature generated by masking the wafer positioning image with a mask pattern; (c) is a schematic diagram of the implementation of generating the positioning mark angle compensation by Hough transforming the positioning mark feature.

Figure 6 (a) is a schematic diagram of a binary wafer positioning image after removing the background of a flat edge wafer positioning image; (b) is a schematic diagram of the implementation of angle compensation for positioning marks generated by Hough transforming the wafer positioning image.

In order to allow the Honorable Review Committee to further understand the overall technical features of the present invention and the technical means for achieving the purpose of the present invention, a specific embodiment is provided with accompanying drawings for detailed description:

Please refer to FIGS. 1 to 3 for an embodiment to achieve the main purpose of the present invention, which includes a rotating mechanism 10, a control module 20, an image capture device 30 and an information processing unit 40. The rotating mechanism 10 includes a rotating platform 11, and the rotating platform 11 is used to carry a wafer 50 with a positioning mark structure 51 (such as a flat edge or a notch) for rotational positioning. The control module 20 is used to control the wafer 50 on the rotating platform 11 for rotational positioning. The image capture device 30 is disposed near the rotating mechanism 10 to capture an image of the wafer 50 that has been rotated and positioned and to form an image of the wafer positioning. The information processing unit 40 includes an image recognition module 41 for performing image processing and image recognition on the wafer positioning image, and performs tilt angle processing calculation on the wafer positioning image after image processing using an image positioning algorithm to obtain a corresponding tilt angle value, and judges whether the tilt angle value exceeds a preset angle threshold. If the judgment result is yes, a corresponding compensation angle signal is output, and the compensation angle signal is transmitted to the control module 20, so that the control module 20 drives the wafer 50 on the rotating platform 11 to perform a corrected rotation positioning according to the compensation angle signal.

Specifically, in a specific embodiment of the present invention, a rotating mechanism 10, a control module 20 and an image capture device 30 are respectively provided at each inspection station 1. The image capture device 30 and the information processing unit 40 are connected via a wired or wireless signal transmission module 60, so that the information processing unit 40 can receive the wafer positioning image transmitted by the image capture device 30; in addition, the control module 20 and the image capture device 30 can also be provided on the rotating mechanism 10, so that the rotating mechanism 10, the control module 20 and the image capture device 30 can be transported by a robot arm and arrive at the next inspection station 1.

In the above-mentioned embodiment, the image positioning algorithm obtains the slope by Hough transform and trigonometric function operation, and obtains the origin of the positioning mark structure 51 and the tilt angle value after rotation positioning from the slope.

As mentioned above, the above Hough transform can be expressed by the following formula:

y=m x+b.

The representative expression of the trigonometric function is:

r = x cosθ + y sinθ Wherein, b is the intercept of the straight line, m is the slope of the straight line, r is the distance from the origin to the straight line, and θ is the angle between r and the x-axis. Specifically, as shown in FIG. 4 , the x-axis can also represent the original position edge of the positioning mark structure 51, and the straight line L can represent the edge of the positioning mark structure 51" after rotation and positioning.

In an application embodiment of the present invention, when the image recognition module determines that the wafer positioning image is tilted in a plane, the edge detection algorithm is used to find the boundary of the wafer positioning image, and the tilt angle value is calculated by Hough transformation. The wafer positioning image is then rotated and corrected by affine transformation, and a compensation angle signal is calculated based on the rotation correction result.

In another application embodiment of the present invention, when the image recognition module determines that the wafer positioning image is tilted on the Z axis, the edge detection algorithm is used to find the boundary of the wafer positioning image, and a perspective transformation is used to project the points on the viewing plane of the wafer positioning image to the real plane, and then the tilt angle value is calculated by Hough transformation. Finally, an affine transformation is used to rotate the wafer positioning image, and a compensation angle signal is calculated based on the rotation correction result.

The present invention discloses a wafer 50 positioning system using Hough transform, which uses Hough transform and inverse trigonometric functions to obtain a slope to find the angle difference between the wafer 50 positioning edge (such as a flat edge or a notch) and the device origin. When the Hough transform is used, the image tilt angle difference can be divided into two situations. One is plane tilt. In this case, the camera device and the positioning device are parallel. The image taken only needs to be rotated to correct the tilt angle difference. The other is Z-axis tilt. In this case, there is a certain angle between the camera device and the positioning device. The image taken needs to be perspective transformed first, and then rotated and other operations can be performed to complete the correction. General wafer 50 equipment does not have this problem. The key to image tilt correction is to automatically detect the image tilt direction and tilt angle based on the image positioning target features. For plane tilt, first use the edge (contour) detection algorithm to find the boundary of the image, then use the Radon transform method (projection-based method), Hough transform method, linear regression method, etc. to find the tilt angle, and then use affine transformation to rotate. The affine transformation of the image must be performed in a homogeneous coordinate system. Homogeneous coordinates can represent rotation, scaling, and translation with matrices of the same size, and their combination is linear and reversible. For the Z-axis tilt, the edge (contour) detection algorithm is first used to find the boundary of the image, and then the perspective transformation is used to project the points on the viewing plane to the real plane, and then the affine transformation is used for rotation. The functions include the control interface, the deceleration mechanism interface and the visual lens interface. The viewing station 1 is set up to include a platform covered with an external camera lens and a communication signal transmission device. The communication interface uses a CAN Bus transmission rate of 20kbps and a 4-way RS485 transmission rate of 9600bps. The camera platform includes an image recognition module, an information processing unit 40 and an image capture device 30. The system is initially established by capturing images from the input image. When the wafer 50 to be tested enters the inspection and monitoring area of the machine inspection station 1, the image capture device 30 captures the wafer 50 as a wafer positioning image. The information processing unit 40 processes and calculates the tilt angle of the wafer positioning image through algorithm technology to obtain the corresponding compensation angle information.

Generally speaking, Hough transform is a feature extraction technique that is widely used in image analysis, computer vision, and digital image processing. Hough transform is used to identify features in objects, such as lines. The algorithm flow is as follows: given an object and the type of shape to be identified, the algorithm will perform voting in the parameter space to determine the shape of the object, which is determined by the local maximum in the accumulator space. Any straight line can be represented by a slope and an intercept. In the patent, a straight line is parameterized using the slope and intercept, but this will lead to an unbounded transform space because the slope may be infinite.

In a plane rectangular coordinate system (x-y), a straight line can be expressed by the equation:

y = m ₀ x + b , where the intercept on the y-axis is b, and m is the slope of the line. When the line is perpendicular to the x-axis, the slope is infinite, so the parameters representing the line are:

r = x cosθ + y sinθ , where r is the distance from the origin to the line, and θ is the angle between r and the x-axis. The use of parameter space solves the problem of the original parameter space divergence, and then the parameters of each line segment can be compared. Some people call the θ-r plane the Hough space of two-dimensional lines. This representation method makes the Hough transform very similar to the two-dimensional Radon transform. Therefore, each line in the image can be associated with a pair of parameters. This parameter plane is sometimes called the Hough space, which is used for a set of two-dimensional lines.

So for the image, for each pixel and its neighbors, the Hough transform algorithm is used to determine if there is enough evidence for a straight line at that pixel. If so, it calculates the parameters for the line, then looks up the accumulator bin that the parameters fall into, and increases the value of that bin (i.e., the vote value). By looking up the bin with the highest value, usually by finding the local maximum in the accumulator space, the most likely lines can be extracted and their geometric definitions read out. The final result of the linear Hough transform is a two-dimensional array (matrix) similar to an accumulator, one dimension of which is the quantized angle and the other dimension is the quantized distance. The value of each element of the matrix is equal to the sum of the points or pixels that are on the line represented by the quantized parameter. So the element with the highest value represents the most represented straight line in the input image. Therefore, given a lot of points, the problem of judging whether these points are collinear becomes the problem of judging whether a bunch of curves on a plane (each point on the plane represents a curve) are collinear after the Hough transform.

The Hough transform algorithm for detecting straight lines uses a two-dimensional matrix called an accumulator to detect whether there are straight lines in the image. It can be described by the equation r=x\cos θ+y\sin θ}r=x\cos θ+y\sin θ}. The dimension of the accumulator matrix is equal to the total number of unknown parameters. For example, to find whether there is a straight line, there are two variables in its parameter space, r and θ, so the dimension of the accumulator matrix is 2. And this two-dimensional accumulator matrix, one dimension represents the quantized r, and the other dimension represents the quantized θ. Therefore, the value of each element of the matrix is the sum of the number of straight lines that can be represented by the element. Therefore, the maximum value of the matrix element means that the confidence that the straight line represented by the element appears in the picture is the greatest.

For each pixel (pixel) (x, y) and its neighboring points, the algorithm will use some evidence to determine whether there is a straight line passing through the pixel (x, y) and its neighboring points. If so, the algorithm will increase the element in the accumulator matrix corresponding to the parameter (r, \θ) of the straight line by 1. Finally, by selecting some local maximum values in the accumulator matrix that are greater than the threshold, it is possible to find the straight line that really exists in the image. In other cases, not using the critical point threshold and using other techniques can make the algorithm perform better. However, the Hough transform can only obtain the parameters of the line, but not the length of the line, so the line must be matched to the line on the image in the next step after the Hough transform. The source of Hough transform errors may be the imperfection of the image (noise, missing pixels), which causes the edge detector to detect incorrect boundaries.

Specifically, FIG5(a) shows a binary wafer positioning image after removing the back of the wafer positioning image of the notch, FIG5(b) shows a positioning mark feature generated by masking the wafer positioning image with a Mask graphic, and FIG5(c) shows a positioning mark angle compensation implementation by Hough transforming the positioning mark feature. In addition, FIG6(a) shows a binary wafer positioning image after removing the back of the wafer positioning image of the flat edge; FIG6(b) shows a positioning mark angle compensation implementation by Hough transforming the wafer positioning image.

After the description of the above specific embodiments, the present invention can indeed achieve a more accurate angle positioning effect when the wafer is rotated and positioned, thereby solving the existing system defects such as the long detection time of the traditional wafer positioning system, so it can effectively shorten the positioning recognition time, thereby effectively improving the process efficiency of the wafer.

The above is only a feasible implementation example of the present invention and is not intended to limit the scope of the present invention. Any equivalent implementation based on the content, features and spirit described in the following claims shall be included in the scope of the present invention. The structural features of the present invention specifically defined in the claims are not seen in similar articles and are practical and progressive. They have met the requirements for invention patents. We hereby submit an application in accordance with the law and sincerely request the Jun Bureau to grant the patent in accordance with the law to protect the legitimate rights and interests of the present applicant.

1: Viewing station

10: Rotating mechanism

11: Rotating platform

20: Control module

30: Image capture device

40: Information processing unit

50: Wafer

51: Positioning mark structure

60:Signal transmission module

Claims (9)

A wafer positioning system using Hough transform, comprising: A rotating mechanism, comprising a rotating platform, the rotating platform is used to carry a wafer with a positioning mark structure for rotational positioning; A control module, which is used to control the rotational positioning of the wafer on the rotating platform; An image capture device, which is arranged near the rotating mechanism, and is used to capture the image of the wafer that has been rotated and positioned to form a wafer positioning image; and An information processing unit includes an image recognition module for performing image processing and image recognition on the wafer positioning image, and performing tilt angle processing and calculation on the wafer positioning image after the image processing using an image positioning algorithm to obtain a corresponding tilt angle value, and judge whether the tilt angle value exceeds a preset angle threshold. If the judgment result is yes, a corresponding compensation angle signal is output, and the compensation angle signal is transmitted to the control module, so that the control module drives the wafer on the rotating platform to perform corrected rotation positioning according to the compensation angle signal. A wafer positioning system using Hough transform as described in claim 1, wherein the image positioning algorithm obtains a slope by using Hough transform and trigonometric function operations, and obtains the origin of the positioning mark structure and the tilt angle value after rotation positioning from the slope. A wafer positioning system using Hough transform as described in claim 2, wherein the Hough transform is represented by the formula y=m x+b; the trigonometric function is represented by: r = x cosθ + y sinθ ; where b is the intercept of the line, m is the slope of the line, r is the distance from the origin to the line, and θ is the angle between r and the x-axis. A wafer positioning system using Hough transform as described in claim 1, wherein the positioning mark structure is selected from one of a flat edge and a notch. A wafer positioning method using Hough transform, comprising: A rotating mechanism, a control module, an image capture device and an information processing unit are provided; wherein the rotating mechanism includes a rotating platform, the rotating platform is used to carry a wafer with a positioning mark structure for rotation positioning, and the information processing unit includes an image recognition module; The control module controls the wafer on the rotating platform to perform rotational positioning; The image capture device is arranged near the rotating mechanism to capture the image of the wafer that has been rotated and positioned to form a wafer positioning image; and The image recognition module processes the wafer positioning image captured in real time, and uses an image positioning algorithm to process and calculate the tilt angle of the wafer positioning image after the image processing to obtain the corresponding tilt angle value, and judges whether the tilt angle value exceeds a preset angle threshold. If the judgment result is yes, the corresponding compensation angle signal is output, and the compensation angle signal is transmitted to the control module, so that the control module drives the wafer on the rotating platform to make a correct rotational positioning according to the compensation angle signal. The wafer positioning method using Hough transform as described in claim 5, wherein the image positioning algorithm obtains the slope by using Hough transform and trigonometric function calculation, and obtains the origin of the positioning mark structure and the tilt angle value after rotation positioning from the slope. The wafer positioning method using Hough transform as described in claim 6, wherein the representative formula of the Hough transform is: y=m x+b; the representative formula of the trigonometric function is: r = x cosθ + y sinθ ; where b is the intercept of the line, m is the slope of the line, r is the distance from the origin to the line, and θ is the angle between r and the x-axis. The wafer positioning method using Hough transform as described in claim 5, wherein when the image recognition module determines that the wafer positioning image is tilted in a plane, an edge detection algorithm is used to find the boundary of the wafer positioning image, and the tilt angle value is calculated using the Hough transform, and then an affine transformation is used to perform rotation correction on the wafer positioning image, and the compensation angle signal is calculated based on the rotation correction result. The wafer positioning method using Hough transform as described in claim 5, wherein when the image recognition module determines that the wafer positioning image is tilted on the Z axis, an edge detection algorithm is used to find the boundary of the wafer positioning image, and a perspective transformation is used to project the points on the viewing plane of the wafer positioning image onto the real plane, and then the Hough transform is used to calculate the tilt angle value, and finally an affine transformation is used to perform rotation correction on the wafer positioning image, and the compensation angle signal is calculated based on the rotation correction result.

Images