KR100774826B1 - Method for detecting the defect of wafer - Google Patents

Abstract

A method of detecting a wafer defect is provided to improve a yield by detecting accurately the defect by comparing a fixed reference image with an image of a die to be detected. An image of a normal die having no defect is set as a fixed reference image(S201). The fixed reference image is compared with an image of a die to be detected(S203), and then the image of the die to be detected is compared with a realtime reference image of a die formed followed by the die to be detected(S205). If there are differences between the fixed reference image and the image of the die to be detected and between the image of the die to be detected and the realtime reference image, it is determined that there is a defect in the die to be detected(S207,S209).

Description

Method for detecting the defect of wafer

1 is a plan view showing a portion of a wafer in which a defect occurs in a die;

2 is a plan view showing a case where a defect occurs in a die of an edge portion starting from a wafer and a case where no defect occurs;

3 is a plan view showing a portion of a wafer having a shot unit of 2 * 2 during exposure;

4 is a flowchart showing a wafer defect detection method of the present invention;

FIG. 5 is a partial plan view showing a case where a defect occurs in a die of an edge portion in which a shot unit of a wafer first starts at 2 * 2; FIG.

* Description of the symbols for the main parts of the drawings *

W: Wafer 110: Shot Area

d: defect

The present invention relates to a wafer defect detection method, and more particularly, to a wafer defect detection method for detecting whether a defect is detected by sensing an image of a wafer die.

In general, a semiconductor device includes a Fab process for forming an electrical circuit including electrical elements on a silicon wafer used as a semiconductor substrate, and an EDS (electrical) for inspecting electrical characteristics of the semiconductor devices formed in the fab process. die sorting) and a package assembly process for encapsulating and individualizing the semiconductor devices with an epoxy resin.

The fab process includes a deposition process for forming a film on a semiconductor substrate, a chemical mechanical polishing process for planarizing the film, a photolithography process for forming a photoresist pattern on the film, and the photoresist pattern. An etching process for forming the film into a pattern having electrical characteristics, an ion implantation process for implanting specific ions into a predetermined region of the semiconductor substrate, a cleaning process for removing impurities on the semiconductor substrate, and Inspection processes for inspecting defects of the formed film or pattern, and the like.

The inspection process is performed to detect defects in a film or pattern formed on the semiconductor substrate. The defects lower the operating characteristics of the semiconductor device and reduce the production efficiency for improving the competitiveness. The defects may have various forms such as scratches, particles, unremoved portions of the material film formed on the surface of the semiconductor substrate, and defects not detected through the inspection process may be the cause of defects in the semiconductor device manufactured from the semiconductor substrate. Works.

1 is a plan view showing a portion of a wafer in which a defect has occurred in a die. In Fig. 1, A, B, C, D, and E each represent one die, and a defect d is formed in the C die.

In the conventional defect detection method, a double detection is performed by comparing images of the die and the die in real time to recognize the defect as a defect.

That is, in the case of detecting a defect of the die B in FIG. 1, the image of the die A and the image of the die B are compared to determine whether there is a difference, and then the image of the die B and the image of the die C are compared to determine whether there is a difference. After judging, it is judged as a defect only when it is determined that there is a difference in both times. If this is arranged as a judgment formula, (A-B) and (B-C) are summarized into two judgment formulas, and as a result, since there is a difference only in image comparison between the B die and the C die, it is not judged as a defect.

In the case of detecting a defect of the C die, it is determined from the judgment equations of (B-C) and (C-D) that there is a difference in the image both times, and it is determined that the C die is defective.

However, this method makes an error in the determination depending on whether a defect occurs in the die at the beginning of the edge when the wafer is composed of 2 * 2 or 2 * 1 dies, which will be described with reference to FIG. 2. do.

2 is a plan view showing a case where a defect occurs in a die of an edge portion starting at the wafer and a case where no defect occurs. In Figure 2 A and A1 die refers to the beginning die of the edge portion, there is a defect (d) in the B die, the A1 die and the C1 die. Since dies A and A1 are the first dies, the decision equations are (A-B) and (A-C) for die A, and (A1-B1) and (A1-C1) for die A1.

In case of A, B, C die defects, there is a difference in case of A die only in contrast result of judgment formula (AB) and (AC). In the case of the die B, both the determination results of (AB) and (BC) are different in the image, and thus double detection is recognized as a defect.

Decisions on the dies A1, B1, and C1 determine that the defect is a double detection as a result of the contrast between the judgment formulas (A1-B1) and (A1-C1) in the case of die A1, B1, and C1. ) And (B1-C1) are double detection because there is a difference in the image, there is a problem that the error is determined to be defective even though there is no actual defect.

This error occurs because the defect detection principle is a 2 * 2 or 2 * 1 shot unit in the double detection method.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problems. When detecting defects of a wafer, a wafer shot unit is 2 * 2 by setting an image of a normal die as a fixed reference image and comparing the fixed reference image with an image of a detection target die. Another object of the present invention is to provide a wafer defect detection method capable of accurately detecting defects even in a 2 * 1 die, thereby improving yield.

The wafer leveling method of the present invention for realizing the above object comprises the steps of: setting an image of a normal die without defects among the dies of the wafer as a fixed reference image; The fixed reference image and the image of the detection target die to detect a defect are compared to determine whether there is a difference, and the real-time reference image of the die formed successively after the image of the detection target die and the detection target die is determined. Comparing and determining whether there is a difference; And determining that there is a defect in the detection target die when there is a difference between the fixed reference image and the image of the detection target die and the difference between the image of the detection target die and the real-time reference image. It is done by

In this case, the fixed reference image may be set to an image of a normal die by prescanning a wafer to be inspected.

In addition, the wafer may be configured of an exposure shot unit consisting of 2 * 2 or 2 * 1 dies.

Hereinafter, the configuration and operation of the preferred embodiment of the present invention will be described in detail with reference to the accompanying drawings. Here, in adding reference numerals to the elements of each drawing, it should be noted that the same elements are denoted by the same reference numerals as much as possible even if they are shown in different drawings.

3 is a plan view illustrating a wafer having a shot unit of 2 * 2 during exposure, and FIG. 4 is a flowchart illustrating a wafer defect detection method of the present invention.

In the wafer W, a shot region which is a unit of an exposure process is set, and a plurality of dies formed of semiconductor chips are formed in one shot region. In FIG. 3, the number of dies formed in one shot region 110 is 2 * 2, and four dies are formed, and A, B, C, D, and E are continuously formed dies ( Die) is assigned a serial number arbitrarily.

In order to check whether or not a defect is formed in each die of the wafer W, the present invention first sets an image of a normal die without defect among the die of the wafer W as a fixed reference image ( S201).

In this case, the image of the die is sensed using an image sensor (not shown), and the sensed image is stored in a computer to become a reference image for detecting a defect of the die.

In this case, the fixed reference image may be composed of a random extraction of what is determined to be a normal die image by pre-scanning the wafer W before the defect detection of the wafer W is performed.

When the fixed reference image is set in the control unit as a reference image, defect detection proceeds by comparing the die with a die to detect whether there is an actual defect.

That is, the fixed reference image and the image of the detection target die to detect a defect are compared to first determine whether there is a difference (S203), and subsequently formed next to the image of the detection target die and the detection target die. It is determined whether there is a difference by comparing the real-time reference image which is the image of the die (S205). The order of these two steps may be reversed.

The judgment of this stage is as follows.

(1) Fixed reference image-Image of die to be detected

(2) Image of detection target die-Real-time reference image

In this case, for example, in the case of a defective die, the defective part is recognized as white and the remaining normal part is recognized as dark. Therefore, when comparing images of a die that is defective and a die that is not defective, it is recognized that there is a difference in the defect portion White.

As a result of determining whether there is a difference between the images, when there is a difference between the fixed reference image and the image of the detection target die, and there is also a difference between the image of the detection target die and the real-time reference image, it is determined that the detection target die is defective (S207, S209).

That is, as a result of comparing the fixed reference image, the image of the detection target die, and the real-time reference image, respectively, the determinations (1) and (2) do not have any difference in the image, and thus no detection is made or the determination (1) , (2) If detection occurs in one of the two judgment results, it becomes single detection, and the die to be detected is judged to be free of defects. Only when the die to be detected is determined to be defective.

In FIG. 3, the decision equation of the present invention is applied to the die A and the die B having the defect (d) as follows. Herein, the fixed reference image is represented by R, and A, B, and C denote images of dies A, B, and C, which are the same below.

In the case of die A, (1) R-A, (2) A-B judgment equations are applied, and detection is made only in case of (2), so it is determined that there is no defect.

In the case of the B die, (1) R-B and (2) the judgment formulas of B-C are applied, and in the case of (1) and (2), both detection is performed and double detection.

FIG. 5 is a partial plan view illustrating a case where a defect occurs in a die of an edge portion at which a shot unit of a wafer first starts at 2 * 2. The defect d is formed in both A die and C die. In this case, a decision error is made in the B die when the conventional defect determination method is applied.

Applying the judgment formula of the present invention is as follows.

In the case of die A, (1) R-A, (2) A-B judgment formulas are applied, and both (1) and (2) are different, resulting in double detection, so it is determined that the die A is defective.

In the case of die B, (1) R-B and (2) B-C are applied, and it is recognized that there is a difference only in case (2).

In the case of die C, since the double-detection is performed by applying the determination formulas of (1) R-C and (2) C-D, it is determined that the die is defective.

Therefore, in the present invention, it is possible to make a normal determination without making an error even in a 2 * 2 die in the present invention.

The defect determination method as described above is applicable not only to the case where the shot unit is 2 * 2 or 2 * 1, but also to the wafer of all shot units.

In addition, the present invention is not only applicable to a pattern error or a particle on the wafer to be inspected, but also to a defect detection even when a defect is repeatedly generated for each shot of the wafer due to a pattern error or particle in the reticle. Normal judgment is possible without making an error.

As described above, the present invention has been described using embodiments, but these embodiments are merely exemplary, and those skilled in the art may make various modifications and changes without departing from the spirit of the present invention. I can understand that.

As described in detail above, according to the wafer defect detection method according to the present invention, regardless of the wafer shot unit, even when the inspection target is a die of the wafer edge portion, it is possible to judge normally without committing an error in determining the yield. There is an advantage to improve.

Claims (3)

Setting an image of a normal die without defects among the dies of the wafer as a fixed reference image; The fixed reference image and the image of the detection target die to detect a defect are compared to determine whether there is a difference, and the real-time reference image of the die formed successively after the image of the detection target die and the detection target die is determined. Comparing and determining whether there is a difference; Determining whether there is a defect in the detection target die when there is a difference between the fixed reference image and the image of the detection target die and the difference between the image of the detection target die and the real-time reference image as a result of the determination of the difference between the images; Wafer defect detection method comprising a. The method of claim 1, The fixed reference image is a wafer defect detection method, characterized in that the pre-scan of the wafer to be inspected is set to an image of a normal die. The method according to claim 1 or 2, The wafer is a wafer defect detection method, characterized in that the exposure shot (shot) unit consists of 2 * 2 or 2 * 1 die.

Images