JPH0682724B2 - Wafer defect inspection system - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION Object of the Invention (Industrial field of use) The present invention relates to a wafer defect inspection apparatus for extracting a defective portion of a semiconductor wafer or the like, and more particularly, to accurately detect a defective portion of a wafer online. The present invention relates to a wafer defect inspection device.

(Prior Art) In general, a semiconductor base material is sliced to manufacture a wafer having a predetermined shape. However, a crystalline defect is generated on a sliced surface of a wafer manufactured by this manufacturing due to various factors,
Further, this defect is represented as a linear defect having a characteristic form having a width of 1 μm and a length of 10 to 30 μm in the horizontal or vertical direction with respect to the wafer reference direction. The large number of such defects causes a significant deterioration in product quality.

Therefore, conventionally, various defect inspection means are used to detect defects in the wafer. One of the defect inspection means is to perform a chemical treatment on the wafer itself to visualize a defective portion, and then visually observe it with a microscope. If it is judged as dust, it is not counted but only a specific defect is detected. The quality of the wafer is determined by counting and considering past experience and the like.

Another defect inspection means scans the inspection surface of the wafer by using an image pickup device by laser reflection, compares the electric signals obtained by this scanning with a predetermined threshold value, and includes the specified rubber. The quality of the wafer is determined by determining the defects, reading the defects, and counting the defects.

(Problems to be Solved by the Invention) However, the former defect inspection means by visual observation is unsuitable for online operation, causes a reduction in the efficiency of inspection work, and is highly accurate unless it has sufficient experience. I can't carry out such inspections.

On the other hand, the latter defect inspection means using the binarization method cannot judge highly accurately in the same manner because it determines that the defect is a specific defect including dust.

The present invention has been made in view of the above circumstances, and provides a wafer defect inspection apparatus that can improve the efficiency of inspection work by going online and can reliably separate dust and specific defects to realize highly accurate defect inspection. The purpose is to provide.

[Structure of the Invention] (Means for Solving Problems) According to the wafer defect inspection apparatus of the present invention, the inspection surface of the wafer is scanned by the image input means, and the wafer image obtained by this scanning is used to identify a defect candidate. The grayscale image inputted from the image inputting means by this defect candidate specifying means is binarized by a predetermined first threshold value, and the isolated point image is removed to make the connected image a defect candidate. Identify,
The defect candidate specified here is input to the characteristic parameter measuring means, a plurality of characteristic parameters are obtained by using a predetermined arithmetic expression, and further the second threshold value set in advance for each characteristic parameter in the defect part determining means. Whether or not the defect candidate is a defect is determined based on the magnitude relation between the value and the characteristic parameter obtained by the characteristic parameter measuring means.

(Operation) Therefore, by using the above means, after the defect candidate is specified by the defect candidate specifying means, a plurality of characteristic parameters are obtained for each defect candidate using a predetermined arithmetic expression,
Since a defect candidate is determined to be a defect from the plurality of feature parameters and the threshold value set for each feature parameter and is output, it is possible to quickly and online by using no artificial intuition or experience. The quality of the wafer can be determined with high accuracy.

(Embodiment) An embodiment of the present invention will be described below with reference to FIGS. 1 to 4. FIG. 1 is a functional block diagram showing a main part of the device of the present invention, FIG. 2 is a structural example of the device of the present invention, and FIG.
FIG. 4 is an operation flow chart of the device of the present invention, and FIG. 4 is an explanatory diagram for specifying a defect candidate. First, as shown in FIG. 1, the apparatus of the present invention functionally scans the inspection surface of a wafer to capture a wafer image, image input means A, and a grayscale image input by the image input means A to a predetermined level. Defect candidate specifying means B for binarizing with a threshold value and specifying defect candidates, for example, for each defect part candidate obtained by this defect candidate specifying means B, for example, area S, vertical / horizontal length Ly, Lx, aspect ratio Lx / Ly, circumscribed rectangle area ratio
It comprises a characteristic parameter measuring means C for obtaining characteristic parameters such as S / Lx and Ly, and a defect portion judging means D for judging whether it is dust or a specific defect using these plural characteristic parameters.

Next, the configuration of the device of the present invention will be described in more detail with reference to FIG. In the figure, reference numeral 11 is a control computer, which sends necessary control signals to various parts of the device through an input / output bus 12, a direct memory bus DMA and a control line (not shown) to perform predetermined control. And has a function of executing image processing based on a predetermined program. Usually this control computer
11 is generally called a central processing unit CPU. A line printer 14, a CRT display unit 15, etc. are connected to the input / output bus 12 via an input / output interface 13, for example. Reference numeral 16 is a main storage unit that stores a program for image processing, necessary fixed constants and other arithmetic expressions, and 17 is temporary data during image processing that is processed by the control computer 11 and other components described later, as needed. It is an auxiliary storage unit that stores the information. Further, although not shown, it has a direct memory controller which directly reads data from, for example, the main storage unit 16, the image memory 22 and the auxiliary storage unit 17 via the direct memory bus DMA. It is designed to be stored (not shown). 18 is a multi-bus command from the control computer 11
It is an image processing sinter face having a function of transmitting it to each image processing element described later through or converting it into a predetermined signal. Between the multi-bus 19 and the image data bus 20 derived from the image processing sinter face 18, an image input device 21 as the image input means A and an image input device 21 constituting the defect candidate specifying means B are inputted. An image memory 22 for storing a wafer image, an intermediate result image, etc., an inter-image calculation unit 23, a histogram measurement unit 24, an image binarization unit 25, a logical filter 26 and a region numbering unit 27 are connected, and a characteristic parameter measuring means. The image memory 22 and the circumscribed rectangle measuring unit 28 which are configured as a part of C are connected.

The image input device 21 scans the inspection surface of the wafer and captures an optical image output from the inspection surface of the wafer as a grayscale image of the wafer with a line sensor, an X-ray imaging device, a television camera, or the like via a microscope, for example. input.

The inter-image calculation unit 23 performs addition / subtraction or a logical sum calculation between pixels at the same position in each of the preset reference image and the grayscale image from the image input device 12, and the image input device 21
To obtain a sensitivity correction value for each of the detection elements. The histogram measuring unit 24 obtains the frequency of shading of the shading image. The image binarization unit 25 binarizes the grayscale image while changing the binarization level in accordance with the change in the brightness of the input image to acquire the binary image. The logical filter 26 removes the isolated point image from the binary image by using the isolated point removal table, the x-direction connection table and the y-direction connection table.
One pixel in the direction and one pixel in the y direction are connected. The area numbering unit 27 assigns, for example, a gray value number to each area where the binary images are connected to identify a defect candidate.

The circumscribing rectangle measuring unit 28 obtains the minimum value coordinates x ₁ , y ₁ and the maximum value coordinates x ₂ , y ₂ for each area numbered defect candidate, and obtains the characteristic parameter by a predetermined arithmetic expression.

Next, regarding the operation of the apparatus configured as described above,
A description will be given with reference to the drawings and FIG. Control computer
A drive control unit 11 controls an image input device 21 through an image processing interface 18 based on a program stored in a main storage unit 16. By this control, the microscope and the television camera as the image input device 21 scan the wafer inspection surface, capture a microscopic image of a part of the wafer with the television camera, input it as a grayscale image of the wafer inspection surface, and input it to the image memory 22. The images are sequentially stored in a predetermined order, the field of view of the microscope is further moved, and a grayscale image of the entire wafer is acquired by the same operation (step S1). Thereafter, in step S2, the inter-image calculation unit 23 is instructed by the control computer 11 to add or subtract between the same pixels of the grayscale image of the wafer inspection surface tabulated in the image memory 22 and the preset reference image, or A logical product sum operation is performed to obtain a correction value, the correction value is added to the grayscale image of the wafer to perform sensitivity correction between the respective detection elements of the television camera, and similarly stored in the image memory 22. Subsequently, the histogram measurement unit 24 is drive-controlled by a command from the control computer 11, and here, the frequency of the grayscale of the grayscale image is measured, and h ₁ (x) = 1/4 {h (x -1) + 2h (x) + h (x + 1)} or h ₂ (x) = 1/8 {h (× + 2) + 2h (x-1) + 2h (x) + 2h (x + 1) + h (x + 2)} etc. Then, the gradation value x max of the peak of the smoothed histogram is obtained. Further, in the image binarization unit 25, for the grayscale value x max of the histogram peak, x max
The corrected grayscale image is binarized using + α (for example, α = 50 for 256 gradations) as a binarization threshold value.
In this way, the threshold of binarization is slightly changed according to the maximum gray value of the histogram, so that the change in the brightness of the input gray image is followed. Therefore, after the binarization process, as shown in FIG. 4, a large number of isolated point images, connected images, and the like are present.

Next, in step S3, the control computer 11 controls the operation of the logic filter unit 26 based on the program, and here, 2
For the value image, the isolated point image a and the connected image b in the x and y directions are found by, for example, the logical product of the adjacent binary images, and the isolated point image is removed (step S3).

Next, in step S4, the operation of the area numbering unit 27 is controlled based on the program. Here, the area number is added to each area image such as a connected image, and the defect candidate is specified and stored in the image memory 22.

Thereafter, the number of pixels in each area of the image numbered in step S5 is determined by the histogram measuring section 24, that is, the area S is obtained, and the maximum and minimum x and y coordinates are determined by the circumscribing rectangle measuring section 28 for each area. The vertical length Ly
And the horizontal length Lx. In addition, characteristic parameters such as the aspect ratio Lx / Ly and the circumscribed rectangular area ratio S / Lx · Ly are obtained.

When a plurality of characteristic parameters are obtained for each defect candidate as described above, it is determined whether these defect candidates are caused by dust or the like, or whether they belong to a specific defect. For this purpose, a preset threshold value is used for each characteristic parameter. That is, for example, 20 ≦
S ≦ 350, 8 ≦ Lx ≦ 50 and 1 ≦ Ly ≦ 12, or 1 ≦ Lx ≦ 1
2 and 8 ≤ Ly ≤ 50, 1.5 ≤ Lx / Ly ≤ 100 or 0.01 ≤ Lx / Ly ≤ 0.67, 0.4 ≤ S / Lx ・ Ly ≤ 1.0, check the magnitude relationship between the threshold and the characteristic parameter, and If it is within the threshold value, it is determined as a defect (step S6). Then, the number of areas determined to be defective in step S7 is counted, the distribution of the number of defects is output for each wafer, and displayed on the CRT display unit 15 as necessary,
The quality of the wafer is determined.

Therefore, according to the above embodiment, the threshold value is set according to the brightness of the grayscale image of the wafer, and the grayscale image corrected by this threshold value is binarized. Can be clearly distinguished from other images, and
It is possible to reliably remove unnecessary images. Further, since the area numbers are assigned and stored for each defect candidate, it is easy to examine the defect distribution for each part of the wafer. Further, since a plurality of characteristic parameters are obtained, and the magnitude relationship is determined for each based on the threshold value corresponding to each characteristic parameter, and the defect is determined based on the overall relationship, the accuracy of the determination result As a result, highly accurate inspection can be performed. Moreover, since no artificial judgment is required, it can greatly contribute to the efficiency of the inspection work.

In the above embodiment, the area S, the length and width Ly, Lx, the aspect ratio Lx / Ly, the circumscribing rectangle ratio S / Lx · Ly, etc. are used as the characteristic parameters, but the defect parameters are not used for all the defect parameters. For example, two or more characteristic parameters may be used to determine the defect in the candidate. Although the linear defect is described as an example, it goes without saying that various characteristic parameters are used depending on the form of the defect to be specified.
Further, in this embodiment, 50 is used as the correction constant α at the time of image binarization, but other positive integers, negative integers or zero can be used. In addition, the present invention can be modified in various ways without departing from the scope of the invention.

[Effects of the Invention] As described in detail above, according to the present invention, it is possible to improve the efficiency of the inspection work by making it online, and to reliably separate dust from specific defects and to count only specific defects. A wafer defect inspection device capable of inspecting a wafer with high accuracy can be provided.

[Brief description of drawings]

1 to 4 are shown for explaining an embodiment of a wafer defect inspection apparatus according to the present invention. FIG. 1 is a functional block diagram showing a main part of the apparatus of the present invention, and FIG. FIG. 3 is a diagram showing an example of the configuration of the apparatus, FIG. 3 is an operation flowchart for explaining a series of image processing, and FIG. 4 is an explanatory diagram for identifying a defect candidate. A ... Image input means, B ... Defect candidate specifying means, C ...
Characteristic parameter measuring means, D ... Defect determination means, 11 ...
… Control computer, 16 …… Main memory, 21 …… Image input device, 22 …… Image memory, 23 …… Inter-image calculation section, 24 ……
Histogram measurement unit, 25 ... Image binarization unit, 26 ... Logical filter unit, 27 ... Region numbering unit, 28 ... Circumscribed rectangle measurement unit.

Claims (4)

[Claims] 1. An image inputting means for scanning an inspection surface of a wafer and inputting a wafer image obtained by this scanning, and a grayscale image input by this image inputting means are binarized by a first threshold value. Defect candidate specifying means for specifying a defect candidate,
Feature parameter measuring means for obtaining a plurality of feature parameters by a predetermined arithmetic expression for each defect candidate identified by the defect candidate identifying means, and a predetermined second threshold value for each feature parameter are set. And a defective portion judging means for comparing with a second threshold value corresponding to each characteristic parameter obtained by the parameter measuring means, and judging whether or not the candidate of the defective portion is a defect from the magnitude relation thereof. A wafer defect inspection apparatus characterized by the above. 2. The defect candidate specifying means binarizes the grayscale image by using the first threshold value which changes according to the brightness of the grayscale image, and the binarizing means. Means for removing an isolated point image from the binarized image obtained by using a logical filter; and area numbering means for identifying a defect candidate by giving an area number to the connected image after the removal of the isolated point image The wafer defect inspection apparatus according to claim 1, which comprises: 3. The characteristic parameter measuring means, for each of the defect candidates numbered with the area, an area, an aspect ratio, an aspect ratio,
The wafer defect inspection apparatus according to claim 1, wherein the characteristic parameter is obtained by using at least two or more arithmetic expressions such as the circumscribed rectangular area ratio. 4. The defective portion determination means includes threshold value setting means for setting the second threshold value for each characteristic parameter,
And a means for comparing the respective characteristic parameters obtained by the characteristic parameter measuring means with a second threshold value set by the threshold value setting means, and judging the specific defect from the magnitude relation. A wafer defect inspection apparatus according to claim 1.