JP2647051B2 - Appearance inspection device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a wafer (semiconductor substrate)
More particularly, the present invention relates to a visual inspection apparatus for detecting a defective portion on a wafer generated in a semiconductor integrated circuit manufacturing process.

[0002]

2. Description of the Related Art Conventionally, as a method of performing a visual inspection of a wafer, there are a method of using a visual inspection device for automatically performing a visual inspection by a pattern matching method and a method of performing a visual inspection by a manual visual inspection. Were present.

FIG. 6 is a block diagram showing a configuration of an example of an appearance inspection apparatus for automatically performing an appearance inspection by pattern matching. This appearance inspection apparatus is disclosed in
It corresponds to the "automatic visual inspection device" according to Japanese Patent Application Laid-Open No. 107946 (the invention according to the publication discloses an appropriate threshold setting method in the automatic visual inspection device).

In this appearance inspection apparatus, as shown in FIG. 6, a sample stage 602 is mounted on an upper surface of an XY stage 601, and a wafer 603 (a wafer to be inspected (sample)) is set on the sample stage 602. You.

[0005] A light source 604 is provided for illuminating the surface of the wafer 603, and a condenser lens 605 is arranged on the optical path of the light source 604.

An objective lens 606 is provided above the wafer 603.
Are arranged, and a half mirror 607 is arranged above the objective lens 606 and on the emission optical path of the condenser lens 605.

[0007] Further, an image pickup means 608 is arranged at a focus position of the objective lens 606. Imaging means 608
Is for photoelectrically converting the reflected light from the wafer 603,
One-dimensional line sensor or two-dimensional ITV (Indu
trial TeleVision. Industrial television)
It is configured using a camera. The imaging unit 608 captures an image of the inspection object (wafer 603).

The difference detection circuit 611 includes two chips (for example, an n-th chip and an (n + 1) -th chip in FIG. 7).
At the same position (for example, the position of (i, j) in FIG. 7).
An image signal (corresponding to an “inspection area” in the present invention described later) (a signal obtained by converting an image captured by the imaging unit 608 by the signal detection circuit 609) is compared to calculate a difference value indicating a difference between the two patterns. (In this comparison, data in the image storage unit 610 is used).

[0009] The defect determining unit 612 determines a defect based on the difference value and the threshold value (the threshold value in the threshold value register 614). That is, when the difference value is equal to or more than a certain value, it is determined that “a defect exists”.

In the invention according to the above-mentioned Japanese Patent Application Laid-Open No. 4-107946, a large number of thresholds are set at regular intervals by the operation of the difference image storage unit 613 and the main control unit 615 to detect the number of defects. The main point is to obtain the optimum threshold value by statistical processing based on the distribution of the difference value.

By the way, in the appearance inspection, the presence or absence of a defect is determined for various inspection items as shown in the following.

Inspection items (exposure defect inspection items) relating to the presence or absence of defects caused by exposure defects (exposure defects in the photolithography process) such as erroneous cutting or joining of patterns, dimensional abnormalities, shape defects, and unexposed.

Inspection items related to presence / absence of defects caused by coating defects such as abnormal film thickness of photoresist, coating unevenness, and uncoated resist (coating defective inspection item)

Inspection items related to the presence / absence of defects due to development defects such as over-development, under-development, and partial undevelopment (development defect inspection items)

Inspection items related to the presence or absence of defects due to the environment such as scratches and dust on the wafer surface (environmental defect inspection items)

Of the various inspection items described above, an exposure defect inspection item relating to erroneous cutting or joining of a pattern or dimensional abnormality cannot be inspected unless a method called pattern matching is used. However, for the other inspection items (1) to (4), even without using a pattern matching method, the surface of the wafer is irradiated with spotlight light, and the presence or absence of a defect is determined by a visual inspection using scattered light generated thereby. It is known that it is possible to do so.

The "visual inspection method" which is the second method in the conventional visual inspection method is applied to such inspection items.

In the visual inspection by visual inspection, the judgment accompanying the sensory inspection in which the scattered light is sensed by a human is performed as described above, and the same operator can continue to inspect many chips to be inspected for a long time. Was needed.

[0019]

The above-described conventional appearance inspection apparatus (appearance inspection apparatus using a pattern matching technique) has the following problems.

Since a defective portion is detected by using a pattern matching technique, the optical system needs to have the ability to resolve all the patterns in the chip. For this reason, in view of the current miniaturization of the semiconductor integrated circuit (the minimum line width of the pattern in the chip is about 0.5 μm, which reaches a half-micron area), the pattern (sub-micron pattern) in the chip is For resolution, an optical system with a rather high magnification objective must be used. As a result, the area that can be compared and inspected at one time is limited to the size of the field of view of the optical system and becomes a very small range, and it takes a long time of 1 to 2 hours to inspect the entire wafer. That is, since the defect of the pattern portion is identified by determining the microscopic image information using a method called pattern matching,
A high-magnification and high-precision optical system is required, and a long time is required for inspection.

This type of visual inspection apparatus currently in practical use is very expensive (because it requires an optical system with high magnification and high accuracy as described above).

In view of the above-described problems in the conventional visual inspection apparatus, it is actually very difficult to introduce a visual inspection apparatus in a mass production site and automate the visual inspection of all wafers. Was.

Therefore, the above-described visual inspection is performed by visual inspection, and at least an automatic visual inspection using pattern matching (appearance inspection by a conventional visual inspection device) and a visual inspection are used in combination (for example, pattern inspection). For inspection items that can only be performed by matching, an automatic appearance inspection is performed by an appearance inspection device, and for other inspection items, a visual inspection is performed.)

However, in the visual inspection by visual inspection, it is necessary to make a judgment accompanying the sensory inspection and to continue the inspection for a long time by the same operator, so that a product based on a judgment error (a semiconductor manufactured through the visual inspection step) is required. Product) and health problems such as a decrease in worker's eyesight.

As described above, from the viewpoint of stabilizing products in the semiconductor manufacturing process and improving the working environment of workers,
“Automation of visual inspection process” has been one of the important issues in the field of semiconductor manufacturing.

In view of the above, an object of the present invention is to automatically perform a visual inspection of an inspection item (an inspection item other than an exposure failure inspection item) which has been conventionally performed by a visual inspection in a short time. It is an object of the present invention to provide a visual inspection device capable of performing the above. Thereby, the object of stabilizing the quality of the product and solving the health problem of the worker can be achieved. Further, it is possible to realize a faster and cheaper appearance inspection than the appearance inspection using a conventional appearance inspection apparatus using a pattern matching technique.

[0027]

An appearance inspection apparatus according to the present invention comprises an XY stage on which a wafer is installed and driven for each unit drive amount based on the size of an inspection area, and each chip in the wafer on the XY stage. A spotlight for irradiating each inspection area with light from an obliquely upward direction, and a light that is disposed above the XY stage and detects the intensity of scattered light in each inspection area with respect to the light irradiated by the spotlight. A light intensity detection sensor for measuring intensity data, and an automatic variable for narrowing a blind width so as to set a light receiving range of the light intensity detection sensor based on the size of each inspection area on each chip in a wafer on the XY stage. When controlling the blind and light intensity data, control the setting of the unit drive amount of the XY stage so as to match the size of the inspection area. Control for controlling the blind width of the automatic variable blind, and judging good / defective of the chip to be inspected based on the light intensity data measured by the light intensity detection sensor and a predetermined threshold value. Based on statistical processing of light intensity data for a plurality of inspection areas at the same position on each chip in the threshold value detection wafer, based on the statistical controller. A statistical processing computer for setting a threshold value to be used.

[0028]

In the appearance inspection apparatus of the present invention, the XY stage (XY stage on which the wafer is placed) is driven for each unit drive amount based on the size of the inspection area, and the spotlight is driven by the XY stage.
Light is applied to each inspection area of each chip in the wafer on the stage from an obliquely upward direction, and a light intensity detection sensor disposed above the XY stage is used to detect light emitted by the spotlight in each inspection area. The intensity of the scattered light is detected, the light intensity data is measured, and the automatic variable blind is set such that the light receiving range of the light intensity detection sensor is set based on the size of each inspection area of each chip on the wafer on the XY stage. The width is reduced, and the controller controls the setting of the unit drive amount of the XY stage and the control of the blind width of the automatic variable blind so as to adapt to the size of the inspection area when measuring the light intensity data. Determining whether the chip to be inspected is good or bad based on the light intensity data measured by the detection sensor and a predetermined threshold value; Used by the control computer based on statistical processing of light intensity data on a plurality of inspection areas at the same position on each chip in the wafer for threshold value detection by the control controller to determine whether the chip to be inspected is good or defective. Set the threshold to be set.

[0029]

Next, the present invention will be described in detail with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of a first embodiment of the appearance inspection apparatus of the present invention.

The visual inspection apparatus of this embodiment includes a wafer 1
An XY stage 2, a control controller 3, a light intensity detection sensor 4, an automatic variable blind 5 (aperture mechanism),
It includes a spotlight 6 and a computer 7 for statistical processing.

FIG. 2 is a diagram showing each chip on the wafer 1 on the XY stage 2 and each inspection area on the chip.

FIG. 4 is a flowchart showing the processing of the appearance inspection apparatus of this embodiment. This processing includes a threshold value setting processing step 40, a light intensity data measurement step 41, and a defect automatic determination step 42. The threshold value setting processing step 40 includes a light intensity data acquisition step 40.
1, a step 402 for obtaining a visual inspection selection result, a step 403 for extracting non-defective data, and a step 40 for calculating an average value or the like.
4, a defective data extraction step 405, a defective data division step 406, a minimum / maximum value extraction step 407, and a threshold value setting step 408.

FIG. 5 is a diagram for explaining the operation of the statistical processing computer 7 in the threshold setting processing (processing realized by the threshold setting processing step 40 in FIG. 4).

Next, the operation of the thus-configured appearance inspection apparatus of this embodiment will be described.

First, assuming that a threshold value has been set in advance (see threshold value setting processing step 40), an operation when an appearance inspection is actually performed will be described. The details of the threshold value setting processing step 40 will be described later.

The inspector sets the wafer 1 (the wafer to be inspected) on the XY stage 2.

As shown in FIG. 2, a plurality of chips (chips,...) Are formed on the wafer 1. Each chip is divided into approximately 10 (in the example of FIG. 2, 9 chips) inspection area and handled in appearance inspection.

The size (size) of the inspection area is set so as to be appropriate for detecting scattered light, that is, so that an area having a uniform luminance distribution is one inspection area. For example, in a memory chip, a cell part and a peripheral part are used, and in a microcomputer, a ROM (Read Only) is used.
Memory), RAM (Random Access)
It can be easily divided into areas in which the shape of wiring such as a ss memory part and a data bus part is repeated and the luminance distribution is uniform.

It should be noted that the above "about 10" means 8 to
Twelve ranges (9 in FIG. 2) are empirically determined as appropriate numerical values. However, this numerical value is merely an example, and other numerical values may be set in accordance with various environments of the visual inspection so that the inspection area is an appropriate area for detecting scattered light (an area having uniform luminance distribution). It is needless to say that can be set.

The XY stage 2 includes a control controller 3
Controls the unit drive amount (X direction and Y direction (in the orthogonal coordinates) based on the size of the inspection area set in the chip formed on the wafer 1 (hereinafter, referred to as the chip to be inspected) which is the inspection object. Horizontal and vertical)
(A unit drive amount).

Above the XY stage 2, a light intensity detection sensor 4 is disposed. The optical system of the light receiving section of the light intensity detection sensor 4 includes an automatic variable blind 5. The aperture of the blind width of the automatic variable blind 5 can be changed under the control of the controller 3. This blind width is adapted to the inspection area set on the chip to be inspected.

The spotlight 6 is disposed at a position where light can be irradiated from obliquely above the surface of the wafer 1. The spotlight 6 irradiates each inspection area with light obliquely from above.

The light intensity detection sensor 4 includes a spotlight 6
The intensity of scattered light in each inspection area with respect to the light irradiated by the method is detected, and data (light intensity data) indicating the intensity of the scattered light is measured.

The control controller 3 presupposes the presence of the wafer 1, the XY stage 2, the light intensity detection sensor 4, the automatic variable blind 5 and the spotlight 6 as described above, and adjusts the XY stage according to the size of the inspection area. 2 is performed step-by-step control (control for driving each unit drive amount) and control for narrowing the blind width of the automatic variable blind 5 is performed, and measurement is performed for each inspection area in the inspection target chip in the wafer 1 which is the inspection target. The obtained light intensity data is obtained (step 41).

Further, the controller 3 compares the light intensity data for each inspection area with a threshold value (threshold value according to the type of the chip to be inspected and the position of the inspection area on the chip) to determine the upper limit. The chip including the inspection area where the light intensity data equal to or higher than the threshold value or the light intensity data equal to or lower than the lower threshold value is measured is determined to be "defective" (the chip which is not so is determined to be "good") (step 42). .

Since patterns and films of various materials, shapes and dimensions are formed on the semiconductor integrated circuit in the inspection area, when the inspection area is irradiated with light, the shape or the shape of the pattern is varied. Scattered light is generated.

When a defect occurs in the shape of the pattern due to a cause in a process, a scratch, dust, or the like,
The intensity of the scattered light in the defective inspection area changes as compared with the intensity of the scattered light in the normal inspection area. The controller 3 makes an automatic determination of the presence or absence of a defect in each inspection area by utilizing such properties.

Secondly, the operation at the time of "threshold value setting process" mainly involving the statistical processing computer 7 will be described.

The statistical processing computer 7 determines the threshold values (“type of chip to be inspected” and “position of the inspection area on the chip”) used in the automatic judgment step 42 by the control controller 3. Threshold) is determined by a statistical method.

In the appearance inspection apparatus of this embodiment, the computer 7 for statistical processing includes a wafer 1 (hereinafter, referred to as a threshold value) for detecting a threshold.
The threshold value is determined based on statistical processing on the light intensity data of the scattered light of the plurality of inspection areas at the same position in each chip in the threshold value detection wafer.

Hereinafter, such a statistical method will be specifically described (step 4 in the threshold setting process in FIG. 4).
0).

First, the light intensity data is measured for each inspection area of each chip in the threshold value detection wafer in the same manner as in the light intensity data measurement step 41 for the wafer 1 to be inspected. Will be The statistical processing computer 7 acquires the measured light intensity data (step 401).

In addition, a visual inspection similar to the visual inspection in the prior art is performed, and a selection is made as to whether each of the inspection areas is a good product or a defective product. The statistical processing computer 7 acquires the selection result (step 402).

FIG. 5 is a diagram showing an example of the distribution of the light intensity data thus obtained. This distribution shows the scattered light intensity for the inspection area at the same position in each chip. Here, the judgment of normal (good) / bad is made by visual inspection.

On the basis of the light intensity data and the results of the selection by visual inspection, the computer 7 for statistical processing
Performs the following processing.

First, from the light intensity data group for the inspection area group at the same position in each chip in the threshold value detection wafer, the light intensity data for the inspection area determined to be defective in the visual inspection is removed. Only the light intensity data (non-defective data) for the determined inspection area is extracted (step 403).

Next, the average value, the upper limit value, and the lower limit value of the non-defective data set extracted in step 403 are calculated (step 404).

On the other hand, from the light intensity data group for the inspection area group at the same position in each chip in the wafer for threshold value detection, the light intensity data (defective product data) for the inspection area determined to be defective in the visual inspection is obtained. Extract (step 405).

Next, each defective product data extracted in step 405 belongs to a data group (hereinafter referred to as an excessively defective data set) located above the range of the non-defective data set (the side where the scattered light intensity is excessive). And data belonging to a data group (hereinafter referred to as an under-defective data set) located on the lower side (the side where the scattered light intensity is too low) (step 40).
6).

Further, the minimum value of the excessively defective data set (hereinafter, referred to as the minimum value in the excessively defective) and the maximum value of the undersized defective data set (hereinafter, referred to as the maximum value in the excessively small defect) are extracted (step 407).

Then, based on the average value, the upper limit value, and the lower limit value calculated in step 404, and the minimum value in the excessive failure and the maximum value in the small defect extracted in step 407, an optimum threshold value (upper limit of the upper limit) is determined. A threshold and a lower threshold (see FIG. 5) are set (step 408).

The setting of such a threshold value is set for each type and process of the object to be inspected (type of the chip to be inspected) and for each position of the inspection area on the chip (therefore, the type and the type of the object to be inspected) A different threshold value is set for each position of the inspection area on the chip).

In this case, the threshold value can be set not only as a light intensity value as shown in FIG. 5 but also as a ratio to an average value calculated from a non-defective data group. 4)). When such a threshold value is set, the threshold value (the threshold value set as a ratio to the average value) is changed from the threshold value (the threshold value set as a ratio to the average value) to the threshold value as the value of the light intensity in the automatic presence / absence determination step 42. Is performed, the light intensity data measured in the light intensity data measurement step 41 is compared with a threshold value.

Incidentally, it goes without saying that the above-described threshold value setting processing is performed by taking samples that are statistically sufficiently reliable (there may be a plurality of threshold value detection wafers). In addition, while performing the actual visual inspection, the visual inspection is appropriately performed in the course of the visual inspection, and the light intensity data is measured and collected for setting the threshold value, and the threshold value is set by statistical processing. Is also possible.

FIG. 3 is a block diagram showing the configuration of a second embodiment of the appearance inspection apparatus of the present invention. The appearance inspection apparatus according to the present embodiment corresponds to the appearance inspection apparatus according to the second aspect of the present invention.

The appearance inspection apparatus of this embodiment includes a wafer 1
It comprises an XY stage 2, a control controller 3, a light intensity detection sensor 4, an automatic variable blind 5, a spotlight 6, a statistical processing computer 7, and a spotlight angle variable mechanism 8. . The components denoted by reference numerals 1 to 7 are the same as the components denoted by the same reference numerals in the first embodiment.

The appearance inspection apparatus of this embodiment is different from the appearance inspection apparatus of the first embodiment (characteristic) in that the spotlight angle can be changed so that the irradiation angle of the spotlight 6 on the wafer 1 can be automatically changed. That is, a mechanism 8 is provided.

Next, the operation of the thus-configured appearance inspection apparatus of this embodiment will be described.

Since the degree of pattern integration differs depending on the type of the chip to be inspected, there is a property that the spatial distribution of the intensity of the scattered light generated with respect to the light emitted from the spotlight 6 is different.

Based on such properties, the spotlight angle varying mechanism 8 controls the light intensity received by the light intensity detection sensor 4 (scattering of the inspection area in the chip to be inspected) according to the type of the chip to be inspected. The irradiation angle of the spotlight 6 is set (changed) so that the intensity of light (light intensity) fluctuates most due to the occurrence of a defect (so that the distribution of non-defective products and defective products is separated as much as possible).

Further, the spotlight angle varying mechanism 8 comprises:
Similarly, when setting the threshold value, the irradiation angle of the spotlight 6 is set according to the type of chip in the wafer for threshold value detection.

At the time of the light intensity data measurement step 41 and the light intensity data acquisition step 401 in the threshold value setting processing step 40 (when measuring the light intensity data), the controller 3 controls the spotlight angle variable mechanism. 8 controls to set the irradiation angle as described above.

As a result, in the appearance inspection apparatus of the present embodiment, more appropriate measurement of light intensity data and setting of a threshold value can be performed, and higher accuracy than in the appearance inspection apparatus of the first embodiment. Automatic appearance inspection can be performed.

[0075]

As described above, the present invention does not use a method called pattern matching (appearance inspection method based on microscopic information) as in a conventional appearance inspection apparatus.
Conventionally, visual inspection is performed among the inspection items of the appearance inspection in the pre-process of the manufacturing of the semiconductor integrated circuit by adopting a method of performing the appearance inspection by capturing the macroscopic image information of the intensity of the scattered light for each inspection area. (Inspection items can be automatically inspected in a short time) (a higher-speed and lower-cost inspection can be realized as compared with a conventional visual inspection device using a pattern matching technique. Has the effect).

As described above, it is possible to automate the appearance inspection process which has been performed by the visual inspection in the conventional technology (in the conventional actual appearance inspection process, several to ten to ten workers are required per line. However, since this can be completely automated), the quality of the product (semiconductor product manufactured through the visual inspection process) can be stabilized, and the visual acuity of the operator due to the visual inspection is reduced. This has the effect of preventing the occurrence of health problems.

Further, according to the second aspect of the present invention, the irradiation angle of the spotlight can be changed and set so as to realize appropriate detection of occurrence of a defect according to the type of the chip to be inspected. The measurement of light intensity data and the setting of a threshold value become possible, and there is an effect that a more accurate automatic appearance inspection can be realized.

When the inspection area is divided into areas with uniform luminance distribution, the luminance of the area becomes uniform over the entire area. If there is a defective part, the luminance is naturally different from the luminance of the non-defective part.
There is an effect that there is a margin in setting the threshold value, and a more accurate automatic appearance inspection can be realized.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a first embodiment of a visual inspection device of the present invention.

FIG. 2 is a diagram showing a chip on a wafer in FIG. 1 and an inspection area on the chip.

FIG. 3 is a block diagram showing the configuration of a second embodiment of the visual inspection device of the present invention.

FIG. 4 is a flowchart showing a process of the visual inspection device shown in FIG. 1;

FIG. 5 is a diagram for explaining an operation at the time of threshold setting processing by the visual inspection device shown in FIG. 1 (a diagram showing an example of a distribution of scattered light intensity with respect to an inspection area at the same position on each chip); It is.

FIG. 6 is a block diagram showing a configuration of an example of a conventional visual inspection device.

FIG. 7 is a diagram for explaining the operation of the appearance inspection device shown in FIG. 6;

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Wafer 2 XY stage 3 Controller 4 Light intensity detection sensor 5 Automatic variable blind 6 Spotlight 7 Statistical processing computer 8 Spotlight angle variable mechanism 40 Threshold value setting processing step 41 Light intensity data measurement step 42 Automatic judgment of defect existence Step 401 Light intensity data acquisition step 402 Visual inspection selection result acquisition step 403 Non-defective data extraction step 404 Average value calculation step 405 Defective data extraction step 406 Defective data division step 407 Minimum / maximum value extraction step 408 Threshold value setting Steps

Claims (6)

(57) [Claims] An XY stage on which a wafer is installed and driven for each unit drive amount based on the size of an inspection area; and a light obliquely upward from each inspection area of each chip in the wafer on the XY stage. A light intensity detection sensor that is disposed above the XY stage, detects the intensity of scattered light in each inspection area with respect to the light emitted by the spot light, and measures light intensity data; An automatic variable blind for narrowing a blind width so as to set a light receiving range of the light intensity detection sensor based on a size of each inspection area on each chip in a wafer on an XY stage; Control of the setting of the unit drive amount of the XY stage and adjustment of the blind width of the automatic variable blind so as to conform to the size of the blind. A controller for determining whether the chip to be inspected is good or defective based on the light intensity data measured by the light intensity detection sensor and a predetermined threshold value. Based on a statistical processing of light intensity data for a plurality of inspection areas at the same position on each chip in the detection wafer, a threshold value used in the control controller for judging pass / fail of the chip to be inspected is determined. A visual inspection apparatus comprising: a statistical processing computer for setting. 2. A spotlight angle variable mechanism for setting an irradiation angle of a spotlight such that the light intensity received by the light intensity detection sensor varies most due to the occurrence of a defect according to the type of the chip to be inspected. 2. The visual inspection apparatus according to claim 1, further comprising: a controller for controlling setting of an irradiation angle by the spotlight angle variable mechanism when measuring data. 3. A light intensity data acquisition step of acquiring light intensity data for each inspection area of each chip in a threshold value detection wafer from a light intensity detection sensor; and each chip in the threshold value detection wafer. A screening result obtaining step of obtaining the result of the visual inspection that each inspection area is good or bad, and the light intensity data obtaining step based on the screening result obtained in the screening result obtaining step A non-defective data extraction step of extracting non-defective data from the light intensity data obtained in the step, an average value of the non-defective data group extracted in the non-defective data extraction step, an average value calculating step of calculating an upper limit and a lower limit, and the like. The light intensity data acquired in the light intensity data acquisition step based on the selection result acquired in the selection result acquisition step A defective product data extracting step of extracting defective product data from the defective product data group; a defective product data dividing step of dividing the defective product data group extracted in the defective product data extracting step into an excessively defective data set and an undersized defective data set; A minimum value / maximum value extraction step of extracting a minimum value within an excessive defect and a maximum value within an excessive defect from the excessive defect data set and the excessive defect data set divided by the non-defective data division step; and An upper limit threshold and a lower limit threshold based on the calculated average value, upper limit value, and lower limit value and the minimum value within the excessive defect and the maximum value within the small defect extracted in the minimum value / maximum value extraction step. And a threshold value setting step of setting a threshold value. The visual inspection device according to claim 1 or 2. 4. The statistical processing computer according to claim 3, further comprising a statistical processing computer for setting a threshold value expressed as a ratio to the average value calculated in the average value calculation step in the threshold value setting step. Appearance inspection device. 5. The inspection area has a size of one chip and one chip.
5. The visual inspection device according to claim 1, wherein the size is divided into about zero. 6. The visual inspection apparatus according to claim 1, wherein the inspection area is divided into areas having uniform luminance distribution.