JPS61245007A - Defect inspecting device - Google Patents

Abstract

PURPOSE:To execute quickly and with high accuracy an inspection of a defect by deciding in detail a defect candidate which has been extracted by the first means, by the second means. CONSTITUTION:An input image pattern is supplied to a defect roughly extracting part 4 through an image input device 2, and A/D converter 3, image patterns of the corresponding parts of two bodies to be inspected are compared, and a defect candidate is extracted roughly. When the rough extraction processing is repeated by a prescribed number of times, and no defect candidate is detected at all, the inspection is ended. On the contrary, if the defect candidate is detected even once, a position where the defect candidate has been generated is stored by a position detecting circuit 5. Subsequently, by a defect minutely extracting part 6, and a feature extracting and comparing circuit 7, only the defect candidate of the position stored before hand is brought to a minute decision processing. In this way, even if the minute decision processing is not executed to all the input image patterns, an inspection of a defect based on the input image pattern can be executed quickly and with a high accuracy.

Description

[Detailed Description of the Invention] [Technical Field] The present invention relates to an inspection technique and a technique that is particularly effective when applied to defect inspection using image patterns, such as defect inspection of semiconductor chips on which semiconductor integrated circuits are formed. It relates to effective technology for the use of.

[Background technology]

For example, in pattern inspection of a semiconductor chip on which a semiconductor integrated circuit is formed, an image pattern of the semiconductor chip is electrically read, and defects are determined and extracted from the read image pattern.

In this case, there are roughly two types of means for determining and extracting the defect.

One of them is to compare the input image pattern and the reference pattern in a one-to-one correspondence, and uniformly determine defects based on the degree of agreement between the two. In this case, the criteria for this judgment are uniform, and any discrepancy greater than or equal to - is considered to be a defect.

Another method uses a large number of dictionary patterns registered for each portion, and by referring to these dictionary patterns, the defect status of each input image pattern is determined in detail. In this case, the criteria for the determination vary depending on the position and type of the input image pattern, and the defects are divided into types and determined in detail.

In the former method, defects can be detected at high speed, but because the criteria for judgment is uniform, the inspection cannot help but be rough, and as a result, even good products are judged to be defective. The problem was that it often happened.

On the other hand, the latter method has the advantage of being able to determine defects very appropriately and accurately by referencing a large number of dictionary patterns, but it takes a long time to process the determination, and therefore requires a large number of inspected objects in the mass production process. There was a problem in that it was difficult to adapt it to the use of inspecting body defects.

As described above, the inventors have found that the conventional defect inspection apparatus of this type has a problem in that it is difficult to achieve both inspection speed and accuracy.

The pattern recognition technology that can be used in this type of defect inspection apparatus is described, for example, in "Nikkei Mechanical 1982.6° 21J pages 77-79, published by Nikkei McGraw-Hill.

[Purpose of the invention]

An object of the present invention is to provide a defect inspection technique that can quickly and accurately inspect defects based on an input image pattern.

The above and other objects and novel features of the present invention will become apparent from the description of this specification and the accompanying drawings.

[Summary of the invention]

A brief description of typical inventions disclosed in this application is as follows.

In other words, a first means that only roughly extracts defect candidates;
and a second means for determining defects in detail,
By using the second means to determine in detail only the defect candidates extracted by the first means, the object of the present invention is to quickly and accurately inspect defects.

〔Example〕

Hereinafter, typical embodiments of the present invention will be described with reference to the drawings.

In the drawings, the same reference numerals indicate the same or corresponding parts.

FIG. 1 shows an embodiment of a defect inspection apparatus according to the present invention.

The device shown in the figure is a defect inspection device that inspects defects in an object to be inspected based on its image pattern, and roughly extracts defect candidate parts from the input image pattern electrically read from the object to be inspected. The present invention includes a first defect determination means and a second defect extraction means for performing detailed defect determination only on the image pattern of the defect candidate portion roughly extracted by the first defect extraction means. It is characterized by

In FIG. 1, 1 is an object to be inspected, such as a semiconductor chip on which a semiconductor integrated circuit is formed, 2 is an image input device using a COD or an image pickup tube, and 3 is an image input device 2.
2 shows an A/D converter that digitizes the image signal 20 read by the computer. 21 indicates the digitized image signal.

Further, 4 indicates a rough defect extraction part. This defect extraction section 4 previously corresponds to the defect extraction section of the memory 1, and compares the imaged cover patterns of mutually corresponding parts of two inspected objects, and determines defect candidates based on the degree of mismatch between the two. Perform the rough drawing. Reference numeral 22 indicates a judgment output when the rough defect extraction unit 4 detects and extracts a defect candidate. When the judgment output 22 is output, the position detection circuit 5 detects the position of the rough extracted defect candidate. The detected position data is then input to the overall control unit (CPU) 9 via the detected position data selector 8 and stored therein. The overall control device 9 is constructed using a CPU (microcomputer). Reference numeral 27 indicates data 27 inputted and outputted from this overall control device 9 via the data selector 8.

Reference numeral 6 indicates a defect detail extraction section, and 7 indicates a feature extraction comparison circuit. The defect detail extraction section 6 and the feature extraction comparison circuit 7 correspond to the second defect extraction section, and refer to the dictionary pattern. Detailed defect extraction section 6, which performs detailed defect determination of the image pattern of the portion extracted by rough defect extraction section 4, includes a dictionary image memory, a signal normalization circuit, etc., as will be described in detail later. include. The feature extraction and comparison circuit 7 extracts the features of the input image pattern 24 and the dictionary-registered image pattern 25 and compares them with each other to perform detailed defect determination processing. The output 26 of the feature extraction and comparison circuit 26 is sent to the overall control device 9 via the data selector 8.

10 is a synchronization signal generator, which generates predetermined synchronization signals H, V
etc. are generated and given to each part. H indicates a horizontal synchronization signal, and ■ indicates a vertical synchronization signal. Images are read in synchronization with these synchronization signals H and V. This synchronizing signal generating circuit 10 operates based on commands 28 from the overall control device 9.

11 is an XY stage, on which the object to be inspected 1 such as a semiconductor wafer is placed. This XY stage 11
The image input device 2 reads only a part of the object 1 to be inspected. However, by moving the XY stage 11 in the XY directions, the entire object 1 to be inspected can be scanned and read.

FIG. 2 shows details of the rough defect extraction section 4. As shown in FIG.

As shown in the figure, the defect rough extraction section 4 has one image memory 40.
1. Digital addition circuit 402, absolute value conversion circuit 403,
Digital comparison circuit 404, threshold setting register 405
, a signal selection circuit 406, a logic inversion circuit 407, and the like.

In the figure, the digitized input image signal 21 is
The image pattern represented by that signal is stored in the image memory 4.
01 in the form of a bit pattern, and is input to one addition input (+) of the digital addition circuit 402. The previously inputted image signal is inputted to the other addition input (+) of this digital addition circuit 402 via a logic inversion circuit 4.07. As a result, the adder circuit 402 calculates and outputs the difference between the previous input image signal and the current image signal for each input port of one image signal. The absolute value of this difference is taken by an absolute value conversion circuit 403. The absolute value 21' of this difference is compared in magnitude with a reference threshold value in a digital comparison circuit 404, and the comparison result is outputted as the judgment output 22 (FIG. 1).

FIG. 3 shows details of the defect detail extraction section 5.

As shown in the figure, the defect detail extraction section 6 is configured by a digital comparison circuit 601, a threshold value table 602, a dictionary image memory 603, and the like.

In the figure, a digital comparison circuit 601 and a threshold table 602 normalize the input image signal 21 for comparison processing. The threshold table 602 is configured using, for example, a ROM, and generates two threshold values 61 and 62, an upper limit and a lower limit, in order to cut out a predetermined range of levels. Each level of the two thresholds 61, 62 is determined by the control device 9.
Each was selected based on instructions from

Next, the operation will be explained.

FIG. 4 shows the operation of the device described above.

In the figure, in steps 82 to s5, the operation as the first defect extraction means is performed.

That is, in steps 32 to s5, a process is performed in which image patterns of mutually corresponding parts of two objects to be inspected are simply compared with each other and defect candidates are roughly extracted based on the degree of coincidence between the two. Note that sl is a routine for initialization, and the actual processing is performed in a loop from 32 to s5.

Further, in slo, the operation as the second defect extraction means is performed. That is, in this slo, a process is executed to perform detailed defect determination of the image pattern by referring to a large number of dictionary patterns.

Here, if no defect candidate is detected even once during the loop execution from 82 to s5, s
The test ends without executing IO.

However, if a defect candidate is detected - times during the execution of the loop from s2 to s5, the branch routine 56t-s7 is executed, the flag F is set to 1, and the position where the defect candidate appeared is be remembered.

After the loop of S2 to S5 has been executed a predetermined number of times, the second defect extraction routine of S10 is executed by checking the flag in s8. At this time, in the second defect extraction routine, only the defect candidates at the positions stored in sl are subjected to detailed determination processing. This allows you to perform detailed determination processing only on some image patterns that may contain defects, without having to perform detailed determination processing on all input image patterns.
It is possible to obtain the same results as when detailed defect determination is performed on virtually all input images.

As described above, defect inspection based on the input image pattern can be performed quickly and with high precision.

Note that the first defect extraction and sl in steps 52 to s5.

The second defect extraction can be performed by arbitrarily selecting only one of them by the control program of the control device.

〔effect〕

(1) A first means for only roughly extracting defect candidates, and a second means for determining defects in detail;
By using the second means to determine in detail only the defect candidates extracted by the second means, it is possible to quickly and accurately inspect defects.

Although the invention made by the present inventor has been specifically explained above based on examples, it goes without saying that this invention is not limited to the above embodiments, and can be modified in various ways without departing from the gist thereof. do not have. ' Above, we have explained the case where the invention made by the present inventor is applied to the field of application which is the background of the invention, which is the pattern inspection technology of printed conductor chips. It can also be applied to pattern inspection technology for printed matter.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an embodiment of a defect inspection device according to the present invention, FIG. 2 is a block diagram showing a part of FIG. 1 in detail, and FIG. 3 is a block diagram showing other parts of FIG. 1 in detail. The block diagram shown. FIG. 4 is a flowchart showing the operation of the apparatus shown in FIG. DESCRIPTION OF SYMBOLS 1... Object to be inspected, 2... Image input device, 4... Defect extraction means output circuit as a first defect extraction means, 6... Defect detail extraction unit as a second defect extraction means , 7... Feature extraction comparison circuit, 9... Overall control device. Figure 1 Figure 2 Cause Figure 3

Claims (1)

[Claims] 1. A defect inspection device that inspects defects in an object to be inspected based on its image pattern, which roughly extracts defect candidate portions from an input image pattern electrically read from the object to be inspected. and a second defect extraction means that performs detailed defect determination only on the image pattern of the defect candidate portion roughly extracted by the first defect extraction means. Features of defect inspection equipment. 2. The first defect extraction means is configured to compare image patterns of mutually corresponding parts of the two objects to be inspected and roughly extract defect candidates based on the degree of coincidence between the two; Claims characterized in that the defect extracting means is configured to perform detailed defect determination of the portion of the image pattern roughly extracted by the first defect extracting means by referring to a dictionary pattern. The defect inspection device according to item 1.