JPH01143938A - Defect detecting method for regular pattern - Google Patents

Abstract

PURPOSE:To securely detect a defect in a regular pattern at a high speed by finding a permissible range for defect decision by using a standard deviation in brightness level of each unit pattern. CONSTITUTION:Many unit patterns which are chip parts are arrayed on a semiconductor wafer 1 regularly according to a prescribed rule. This semiconductor wafer 1 is irradiated with light emitted by a lighting device 2. An image signal outputted by an image pickup device 3 arranged above the semiconductor wafer 1 is processed by an image processor 5. Then the brightness level in at least each unit pattern area is measured to find the standard deviation in the brightness level of at least each unit pattern which is viewed from the whole regular pattern. Then this standard deviation is used to find the permissible range for defect decision. Further, a frequency distribution is generated from standard deviations to decide that a unit pattern having a standard deviation exceeding the permissible range is defective.

Description

Detailed Description of the Invention [Purpose of the Invention (Industrial Application Field) The present invention relates to a method for detecting defects in a regular pattern suitable for detecting defects in a regular pattern in a chip portion formed on a semiconductor wafer, for example. Regarding the method.

(Prior Art) Semiconductor wafers on which a regular pattern is formed by arranging a large number of chip portions as unit patterns according to a predetermined rule are inspected to see if the regular pattern is accurately formed. This inspection used to be carried out exclusively visually, but recently this inspection has come to be carried out automatically. An example of this is processing performed by fast Fourier transform, that is, FFT processing.

This FFT processing irradiates the semiconductor wafer with light, receives the reflected light, and analyzes the components of this reflected light into frequency components corresponding to the regular pattern. Since frequency components other than the frequency components appear, the presence or absence of defects is determined from these frequency components.

However, in the above-mentioned FFT processing, the frequency components corresponding to the regular pattern must be removed in order to obtain the frequency components corresponding to the defects, so the processing takes time. Furthermore, in order to execute this process, constraints must be given to the process so that frequency components corresponding to defects are not removed. Another problem is that the FFT processing device is expensive.

(Problems to be Solved by the Invention) As described above, FFT processing is susceptible to the influence of regular patterns in order to detect defects, and the processing takes time.

SUMMARY OF THE INVENTION Therefore, an object of the present invention is to provide a regular pattern defect detection method that can reliably and quickly detect defects within a regular pattern.

[Structure of the Invention] (Means and Effects for Solving the Problems) The present invention measures each brightness level in at least each unit pattern area and establishes a standard for the brightness level of at least each unit pattern when viewed from the entire regular pattern. Determine the deviation, then use this standard deviation to determine the defect determination tolerance range, and then determine whether there is a defect in the unit pattern of the standard deviation that falls outside the defect determination tolerance range in the frequency distribution created from the standard deviation. This is a method for detecting defects in regular patterns.

(Example) Hereinafter, an example of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram of a regular pattern defect detection apparatus for semiconductor wafers to which a regular pattern defect detection method is applied. In the figure, reference numeral 1 denotes a semiconductor wafer, and on this semiconductor wafer 1, a large number of unit patterns 1a, which are chip portions as shown in FIG. 2, are regularly arranged in accordance with a predetermined rule. The semiconductor wafer 1 is then irradiated with light emitted from an illumination device 2.

Further, above the semiconductor wafer 1, an imaging device 3 consisting of an ITV (industrial television) camera and a CCD (solid-state imaging device) is arranged. The image signal output from the imaging device 3 is digitized by an A/D (analog/digital) converter 4 and sent to an image processing device 5.

This image processing device 5 has a function of detecting whether or not there is a defect in the regular pattern of the semiconductor wafer 1, and its specific configuration is as follows. That is, the input digital image signal is stored in the image memory 6 as image data. The main control section 7 is the third
It has a function of controlling the operation of the image processing device 5 according to the program of the defect detection flowchart shown in the figure.

Now, as shown in FIG. 4, the area setting section 8 divides the regular pattern into unit patterns 1a to create detection areas for each unit pattern, and a detection area pattern Q for creating these areas is set. . Brightness level? The ipl determining unit 9 determines each brightness level in each area divided by the detection area pattern Q, that is, x1. x2. x3. −・−x
This is to find n. The standard deviation calculation unit 10 has a function of calculating and finding the standard deviation σn using the brightness levels of each area divided by the detection area pattern Q. First, the average value x − (x 1 + x
2 + x 3, +-+ x n) / n-(
Σxi)/n...(1)...(2
) is now required. Furthermore, this standard deviation calculation unit 10 has a function of determining the average σ of each of the determined standard deviations σn, and also determining the standard deviation σ of this average 7.

The defect determination tolerance range calculating section 11 has a function of calculating a defect determination tolerance range S from the average value σ obtained by the standard deviation calculation section 10 and its standard deviation σr, and this defect determination tolerance range S is as follows. is required. In other words, σ-α conflict σr≦S≦σ+α fistσr. Note that α is a constant.

The frequency distribution calculation unit 12 has a function of calculating the frequency distribution of the standard deviation σn of each area divided by the detection area pattern Q, and the defect determination unit 13 further uses the frequency distribution of the standard deviation σn calculated by the frequency distribution calculation unit 12. It has a function of detecting a standard deviation σn that deviates from the defect determination tolerance range S in the frequency distribution and determining that a defect exists in the area of this standard deviation σn. Note that this defect determining section 13 has a function of notifying an external notification means G when determining that there is a defect.

Next, the operation of the apparatus configured as described above will be explained with reference to the defect detection flowchart shown in FIG. The surface of the semiconductor wafer 1 is in a state where light is irradiated from the illumination device 2, and in this state, the imaging device 3 images the semiconductor wafer 1 and outputs the image signal. This image signal is converted into a digital image signal by the A/D converter 4 and sent to the image processing device 5. This image processing device 5
A digital image signal is taken in and stored in the image memory 6 as image data. Here, if the captured image has a regular pattern as shown in FIG. 2, the image data of this image is stored in the image memory 6.

In addition, in FIG. 2, Fl and F2 are defects. Next, the process moves to step e2, and the area setting unit 8 sets the image data to [.

A detection area pattern Q is set and divided into areas each having a unit pattern la as shown in FIG.

Once each area is divided in this way, the process moves to step e3, where the brightness level measuring unit 9 determines the brightness levels xl, x2, . x3. ...Find xn. These brightness levels are sent to the standard deviation calculation unit 10 according to a command from the main control unit 7, and the standard deviation calculation unit 1
0 is the standard deviation σ for each region in the next step e4
Calculate and find n. In other words, this standard deviation calculation unit 10 first calculates and finds the average X of each luminance level from the above equation (1), and then calculates each standard deviation σn of each area by calculating the above equation (2). , and further find the average σ of these standard deviations σ and the standard deviation σ corresponding to this average σ.

Next, proceeding to step e5, the main control section 7 sends the determined average σ of the brightness level and its standard deviation σr to the defect determination tolerance range calculation section 11. This defect determination tolerance range calculation unit 11
The defect judgment allowable range S σ−α・σr≦S≦σ+α−σr is determined from the average 7 and the standard deviation σr.

On the other hand, the frequency distribution calculation section 12 calculates the frequency distribution of standard deviations from each standard deviation σn as shown in FIG.
Pass it to 3. In step c7, the defect determination unit 13 determines whether there is a standard deviation σn that is outside the defect determination tolerance range S in the standard deviation frequency distribution, and if there is, defects F1 and F2 are present in the area of this standard deviation σn. It is determined that there is. Then, the defect determination unit 13 notifies external reporting means of this fact in step e9.

In this way, in the above embodiment, each luminance level in each region is determined by 11j, the standard deviation σn of the luminance level for each region viewed from the entire regular pattern is determined, and then this standard deviation σn is used to calculate the luminance level. The defect determination tolerance range S is determined, and then in the frequency distribution created from the standard deviation σn, it is determined that there is a defect in a unit pattern with a standard deviation σn that falls outside the defect determination tolerance range S, which affects the rule beta. Only the defects Fl and F2 can be reliably detected without being detected, and the processing can be performed at high speed because only the standard deviation is determined. Therefore, such a processing device can be easily configured and inexpensive.

Note that the present invention is not limited to the above-mentioned embodiment, and may be modified without departing from the spirit thereof. For example, as shown in No. 64, the detection area pattern Q may be enlarged and ζ to a size that accommodates, for example, four unit patterns 1a, compared to the above-described one embodiment. This case is particularly effective when the size of the unit pattern 1a changes.

[Effects of the Invention] As described in detail above, according to the present invention, it is possible to provide a method for detecting defects in regular patterns that can reliably and quickly detect defects in regular patterns.

[Brief explanation of the drawing]

1 to 5 are diagrams for explaining the regular pattern defect detection method according to the present invention, in which FIG. 1 is a block diagram of a regular pattern defect inspection apparatus to which the present method is applied, and FIG.
3 is a schematic diagram of a regular pattern, FIG. 3 is a defect detection flowchart, FIG. 4 is a schematic diagram showing a detection area, FIG. 5 is a frequency distribution diagram of standard deviation, and FIG. 6 is a diagram showing a modification. DESCRIPTION OF SYMBOLS 1... Semiconductor wafer, 2... Illumination device, 3... Imaging device, 5... Image processing device, 6... Image memory,
7... Main control section, 8... Area setting section, 9... Brightness level measuring section, 10... Standard deviation calculation section, 11...
Defect determination tolerance range calculation unit, 12... frequency distribution calculation unit,
13... Defect determination section. Applicant's agent Patent attorney Takehiko Suzue

Claims (1)

[Claims] In a defect detection method for a regular pattern in which a large number of unit patterns are arranged according to a predetermined rule, each brightness level in at least each of the unit pattern areas is measured to determine the brightness level of at least each of the unit patterns as seen from the entire regular pattern. Find the standard deviation of , then use this standard deviation to find a defect judgment tolerance range, and then find the defect in the unit pattern of the standard deviation that falls outside the defect judgment tolerance range in the frequency distribution created from the standard deviation. A regular pattern defect detection method characterized by determining that.