JPS60245222A - Appearance inspection and equipment thereof - Google Patents

Abstract

PURPOSE:To reduce the total time of an appearance inspection operation by detecting a defective position including an erroneous detection as the first inspection and further by excluding the erroneous detection from the defective position as the second and the later inspections. CONSTITUTION:At the first time, a stage 11 is moved at high speed, a threshold voltage is set at low value and an overall high detection rate inspection is carried out. In the second inspection, the stage 11 is driven at low speed at a defective position. The results are that many erroneous detections are excluded. In the third inspection, the stage 11 is stopped or driven at low speed and the threshold voltage is continuously varied from high to low value whereby almost all the erroneous detections are excluded. At last, a final judgement is done by human viewing and the erroneous detection, if any, is excluded. In this way, the time of an appearance inspection operation can be reduced.

Description

DETAILED DESCRIPTION OF THE INVENTION [Technical Field] The present invention relates to foreign matter inspection technology, and particularly to a technology that is effective for use in visual inspection of masks, wafers, and the like.

[Background technology]

When inspecting the appearance of a photomask,
．． It is conceivable to compare mask patterns that are formed separately.

Therefore, an inspection device as shown in FIG. 1 can be considered.

In the figure, a high-brightness CRT is used, and a light beam 2 emitted from a scanning point on the high-brightness CRT is directed onto a photomask 3 by two lenses 4a and 4b kept at a distance equal to the chip spacing of the photomask 3. The image is formed and scanned over the photomask 3. Two optical sensors 5a and 5b are placed under the photomask 3 at positions that receive light from lenses 4a and 4b.

Note that the optical sensors 5a and 5b are linked to the distance between the lenses 4a and 4b. Lenses 4a and 4b are integrally fixed to the CRTl. The light beam 2 scans adjacent mask patterns 76a and 6b on the photomask 3 simultaneously.

Adjacent pellets (chips) as shown in Figure 2
are 7a and 7b, and this bellet 7a.

If the mask patterns on 7b are 6a and 6b, when the scanning point advances along the line shown by the dashed-dotted line 8, it will scan the normal mask pattern part, so it will enter the two optical sensors 5a and 5b. The resulting scene is the same, and the difference between the electrical signals based on each light amount becomes zero and does not reach the preset threshold voltage (detection level). No pattern anomalies are detected in the comparison. When the scanning point advances along the line shown by the dashed-dotted line 9, one scanning point is the defective area 1.
When the temperature reaches 50°, there is a difference in the amount of light entering the two optical sensors 5a and 5b, that is, the amount of light is not detected by the optical sensor 5a, so the output signals (electrical signals) of the optical sensors 5a and 5b
When a difference occurs and the voltage exceeds a preset threshold voltage, it is detected as a defect (pattern abnormality). In other words, the two chips 7a, 7 on this scanning line 9
In the comparison of b, when the scanning point reaches the defective part 10, one of the light rays 2 will be blocked by the defective part 10 on the chip 7a, and the other light ray 2 will pass through the corresponding part on the chip 7b, and the light amount will change. A difference is generated in the outputs of the original optical sensors 5a and 5b, and a defect is detected.

In this way, as the scanning points progress, corresponding locations on two adjacent chips are compared one after another, and if there is a defective location on the scanning line, it is detected. In this case, both compared corresponding locations are treated as defects, for example.

Incidentally, as mask patterns become finer, the size of external defects in mask patterns that must be detected also becomes smaller. For example, pattern miniaturization is 3 μm → 2 μm →
As the size of the defect decreases from 1.5μm → 1μm, the size of the defect also decreases from 1.5μm → 1μm → 0.8μm → 0.5μm.
becomes smaller as in Therefore, as patterns become finer, it is necessary to improve the detection sensitivity for detecting defects.

Therefore, in order to improve the detection sensitivity, it is possible to increase the lens magnification, lower the threshold voltage (detection level), etc.

However, when the detection sensitivity is improved in this way, even smaller defects can be detected, but so-called false detections, in which non-defects are detected, also increase. If there are 10 defects, there will be 5 to 10 times as many false detections as this, and in the end, the inspection throughput will be reduced because the household members will have to judge whether there are defects or false detections.

One way to keep this false detection low is to reduce the moving speed of the stage (not shown) on which the photomask is placed and reduce the influence of vibrations on that stage.
Even if this is done, the overall inspection will take longer and reduce throughput. In addition, as another method for suppressing the above-mentioned false detection, it may be possible to increase the threshold voltage, but the inventor has clarified that in this case, there are problems such as a decrease in defect detection accuracy. .

It has been clarified by the present inventors that the above-mentioned problems occur not only in the case of detecting external defects in photomasks, but also in detecting external defects in notches in Ueno et al.

A document that describes appearance inspection (defect inspection) is Electronic Materials 1982 Special Edition "Optical Inspection Machine 1JP243~
There is P247 (published by Kogyo Kenkyukai, November 1, 1982)
(published on the 5th).

[Purpose of the invention]

The purpose of the present invention is to reduce the time required for visual inspection of patterns on substrates (
(Improve throughput)
It is an object of the present invention to provide a visual inspection method and a visual inspection apparatus that can improve defect detection accuracy at the same time.

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

[Summary of the invention]

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

That is, the first invention of the present invention scans the substrate using a flying spot generating cathode ray tube to detect shape defects in a plurality of patterns formed on a substrate, and compares the patterns two by two. In the visual inspection method, the first inspection is to perform a full-surface inspection to detect defect positions, including false positives, and the second and subsequent inspections are to detect defect positions based on the previous inspection results. Furthermore, by eliminating false detections, the final judgment time for humans to determine whether or not a defect exists can be extremely shortened, reducing the total visual inspection work time, improving throughput, and at the same time improving defect detection accuracy. This is something that can be improved.

In a second aspect of the present invention, a substrate on which a plurality of patterns are formed is placed on a stage, and two patterns on the substrate are simultaneously scanned and compared using a flying spot generating cathode ray tube. In addition to detecting defects in the pattern shape, the inspection device includes a stage drive circuit that drives and controls the moving speed of the stage based on a control command, and a stage drive circuit that controls the movement speed of the stage based on a control command, and a stage drive circuit that detects defects in the pattern shape on the substrate. two photodetectors that respectively detect the images of the flying spots irradiated on the two photodetectors; a comparison circuit that compares the outputs of these photodetectors and sends out a deviation output; A defect detection and removal circuit that detects pattern defects by comparing them with a variable detection level and removes minute defects among the defects, and stores the defect position on the substrate based on the defect detection signal from this defect detection and removal circuit. For example, in the first inspection, the detection level of the defect detection and removal circuit is set low, and in the second inspection, in addition to this, the previous g The moving speed of the stage is slowed down or stopped to check again whether or not there is an erroneous detection of the defect position, and the erroneously detected position is excluded.Furthermore, in the third inspection, At the defect position that is the result of the second inspection, the moving speed of the stage is slowed down or stopped, and the detection level of the defect detection and removal circuit is varied from a high value to a low value. By removing minute defects that occur when the detection level is below a predetermined value, the rate of false positives included in the detected defect position is greatly reduced, and the final determination of whether or not the defect is caused by humans is greatly reduced. This makes it possible to extremely shorten the judgment work time, reduce the total visual inspection work time (improve throughput), and at the same time improve defect detection accuracy.

〔Example〕

FIG. 3 shows an embodiment of the visual inspection device according to the present invention,
This is a case where the method is applied to a photomask as a substrate to be inspected.

In the figure, the same reference numerals are used for the same parts or parts having the same functions as in FIG. 1. 11 is a stage on which the photomask 3 is placed; 12 is a stage drive circuit; this stage drive circuit 12 moves the stage 11 in a horizontal plane on the X axis,
It is movable in the Y-axis direction. Note that, according to the program, the CPU 3 sets the speed of the stage 11 to 37.5 cm/soc in the first inspection, and the speed of the stage 11 in the second inspection to the maximum speed up to the defect position according to the output of the memory 16, and 18.7 at defect location
In the third inspection, the speed of the stage 11 is set to the maximum speed up to the defect position stored in the memory 16 according to the second inspection result, and then stops or slows down at the defect position so that the speed is 5 crn/see. 5 cm/sec). 14 is a two-chip comparator circuit that inputs the outputs from the optical sensors 5a and 5b, calculates the deviation, and outputs this; 15
checks whether the output (deviation signal) of this comparison circuit 14 is higher than the threshold voltage (detection level),
This is a defect detection and removal circuit that determines a defect if the deviation signal is equal to or higher than a threshold voltage, detects this, and eliminates a permissible minute defect that occurs when the threshold voltage is equal to or lower than a predetermined value. Note that the threshold voltage, which is the detection level, is set to a low value (for example, 0.35 V) for the first and second tests, and is set to a low value (for example, 0.35 V) for the third test. In this test, the voltage can be varied from a high value (for example, 0.6V) to a low value (for example, 0.4V). 17 is a synchronization circuit for the stage and the flying spot, which controls the CRTI based on the output of the stage drive circuit 12 so that the scanning speed of the flying spot of the CRTI and the speed of the stage 11 can be synchronized; This prevents duplicate detection of defects at the same location on the same two chips. Note that the CRTI is controlled by the CPU 3.

Based on this configuration, a method for detecting defects in the appearance (shape) of the mask pattern of the photomask 3 will be described. In addition, multiple inspections can be automatically performed using the CPU1 of the Microconbeater.
The test is carried out according to the program No. 3, and the following explanation will be given by taking as an example the case where the test is carried out three times.

First, in the first inspection, the defect detection sensitivity is increased and a high detection rate inspection is performed over the entire surface of the photomask 3.

That is, in the first inspection, based on a control command from the CPU 13, the stage drive circuit 12 sets the moving speed of the stage 11 to, for example, 37.5 cm /see, and sets the threshold which is the defect detection level of the defect detection and removal circuit 15. Set the voltage to a low value (eg 0.35V). Then, as described above, the appearance of the photomask 3 is inspected using the flying spot two-chip comparison method. That is, the second
For example, as shown in the figure, mask patterns 6a, 6b#-
The formed adjacent pellets (chips) 7a and 7b are simultaneously scanned. When the flying spot (scanning point) moves from left to right along the line indicated by the dashed line 9, it initially scans a normal pattern area, so it enters the two optical sensors 5a and 5b. The resulting scene is the same, and when the comparison circuit 14 takes the output difference between the optical sensors 5a and 5b, the deviation output becomes zero, and the defect detection and removal circuit 15 does not reach the preset threshold voltage (detection level), so the defect is detected. No detection occurs. However, only when the scanning point advances and reaches the defective part 10, the amount of light is not detected by the optical sensor 5a, and a difference occurs between the outputs of the optical sensors 5a and 5b, and the difference signal is extracted by the comparator circuit 14.

When the deviation signal exceeds the threshold voltage, the defect detection and removal circuit 15 determines that there is a defect (pattern abnormality), detects both of the compared scan positions as defect positions, and stores the detected defect position. The defect inspection signal is supplied to the memory 16, and the C
It is stored that the address corresponding to the RTI scan is the defective position.

Next, when the scanning point advances along the line shown by the dashed line 8,
Since the normal pattern area is scanned simultaneously, the amount of light entering the two optical sensors 5a and 5b is the same (or zero),
Since the output difference between the optical sensors 5a and 5b becomes zero and does not reach the preset threshold voltage, no defect (pattern abnormality) is detected by comparing the two chips 7a and 7b on the scanning line 8.

As described above, as shown by the arrows in FIG. 4, the pellets (chips) on which the mask pattern is formed scan the photomask 3 arranged vertically and horizontally. At this time, as the scanning point progresses, corresponding positions on two adjacent chips are simultaneously scanned one after another to compare patterns, and if there is a defective position on the scanning line, that defective position is detected. When the defect detection and removal circuit 15 determines that a defect exists, both of the compared corresponding positions are processed as defects. Then, based on the defect detection signal from the defect detection and removal circuit 15, it is detected that defect positions (original defect positions or erroneously detected positions) are distributed within the photomask 3 as shown by the X marks in FIG.
is memorized. As described above, the defect positions indicated by the X marks include many false detections.

In the first inspection described above, the threshold voltage of the detection level in the defect detection and removal circuit 15 is lowered, so if there is even a slight difference between the outputs of the optical sensors 5a and 5b, it will be detected as a defect position, and the comparison will be made. Since both are considered to be defects even if only one of them is an original defect, the defect detection sensitivity increases although there are many false detections.

- In the second inspection, a step inspection (re-inspection) of only the defect location is performed instead of a full-scale inspection. Since the defect position is input to the CPU 13 from the memory 16, the CPU 13 operates the stage 11 at the highest speed until the defect position is reached, and at a lower speed (for example, 18, 75 cmls) at the defect position detected in the first inspection, according to the program. ) A control command is given to the stage drive circuit 12 to drive the stage. here,
If the speed of the stage 11 is high, the photomask 3 will vibrate and the position of the photomask 3 will not be aligned properly, resulting in false detection (
Therefore, in order to eliminate erroneous detection due to this vibration, the speed of the stage 11 was set to a low speed at the defect position, which is the result of the first inspection. Note that the CPU 13 may be programmed to stop the stage 11 without slowing down. Further, the detection level of the defect detection and removal circuit 15 is the same as the first time.

Under such setting conditions, the same inspection as the first inspection is performed by slowing down the speed of the stage 11 to remove the influence of vibration of the photomask 3 only for the defect positions detected in the first inspection. As a result of this second inspection result, many erroneous detections are removed, and as a result, the only defective position in the photomask 3 is, for example, the position indicated by the X mark in FIG. 5, and this defective position is updated and stored in the memory 16.

Next, in the third inspection, only the defect positions remaining in the second inspection are inspected. In this case, the CPUI
3 gives a control command to the stage drive circuit 12 to set the speed of the stage 11 at the maximum speed up to the defect position, and then stop or slow down (5 an/in) at the defect position, and also controls the speed of the stage 11 when the defect is detected again at the second time. Change the threshold voltage for the position from a high value (for example, 0.6V) to a low value (for example, 0.6V).
A control command is given to the defect detection and removal circuit 15 to continuously vary the voltage to 4V). As a result, vibration errors (false detection due to vibration) at the defect location can be almost completely eliminated.
In the defect detection and removal circuit 15, this detection level (0,6 to 0
．． 4V) are detected as defects, and acceptable minute defects occurring below the detection level are removed. Therefore, the defect detection and removal circuit 15 changes the threshold voltage from 0.6 to 0.4 as shown in FIG. 6 for the defect position of the second inspection result, and determines at which voltage defect detection starts. By doing so, the size of the defect can be determined, and if you want to know this, you can also store this in the memory 16. The size of the defect that starts to be detected when the threshold voltage is in the range of 0.6V to 0.4V shown in Figure 6 is around 2 μm, and if there is a defect within this range, it is determined to be a defect and 0. Micro-defects (size: 1.6 μm or less) that are detected at a voltage of .4 V or less are allowed and do not pose a problem. As a result of the third inspection, the distribution of defect positions in the photomask 3 became the location indicated by the X mark in the 7th M, and
Compared to the second test result (4th @), almost all false positives are removed.

After the third automatic inspection is completed, a human makes a final visual judgment as to whether the defect position, which is the result of the third inspection, is an erroneous detection or a defect, and if there is an erroneous detection, it is removed.

As a result, the defect distribution diagram in the photomask 3 becomes as shown in FIG.

The appearance inspection method has been briefly explained above, but until now it took two hours to inspect the appearance of a photomask of a certain size and capacity (corresponding to the first defect detection inspection), and as a result, the number of defects detected was Assuming that there were 1000 detections (including false positives), it would take 80 minutes for the final visual judgment by a human, and a total of 200 minutes.

In contrast, in the embodiment of the present invention, the first inspection takes the same amount of time as before, and if the number of detected defects is 1000, including false detections, the second and third inspections are unprecedented. The time required for automatic inspection is 30 minutes. However, as a result of the third inspection, most of the false positives have been eliminated, and if the number of remaining defects is 100, the final visual judgment time for a human being is only 8 minutes, and the total visual inspection work is reduced. The time will also be 158 minutes. Therefore, the time required for one photomask is reduced by 42 minutes compared to the conventional method. Therefore, the throughput can be significantly improved compared to the conventional method.

Furthermore, in the first inspection, the detection level (threshold voltage) of the defect detection and removal circuit 15 is lowered without worrying about erroneous detection, so even if the number of erroneous detections increases, it is possible to detect even minute defects. Improved detection sensitivity for defects 1-1 In addition, the second and third inspections allow the previously detected defect positions to be more strictly inspected and almost all false detections can be removed, so the results of the third inspection The defect detection accuracy can be significantly improved compared to the conventional one-time detection.

Therefore, the appearance quality of the chip can be improved.

〔effect〕

If the present invention is used, the following effects will be achieved.

(1) Even if it takes time for the second and subsequent automatic inspections, this time will be short, and the results of the second and subsequent inspections will eliminate many of the defect positions that were incorrectly detected in the first inspection. Since it is removed as a detection, the final human judgment time is extremely short compared to the conventional method, and the overall appearance inspection work time can be significantly shortened compared to the conventional method, improving throughput. I can do it.

(2) As the first inspection, it is possible to increase the defect detection sensitivity and detect even minute defects without worrying about erroneous detections, but this will include many erroneous detections. However, in the second and subsequent inspections, the defect position based on the previous inspection results is inspected with higher defect detection accuracy, so in the end most false detections are removed and the defect detection accuracy can be improved. can. Therefore, the appearance quality can be improved.

Although the invention made by the present inventor has been specifically explained above based on Examples, it goes without saying that the present invention is not limited to the above Examples and can be modified in various ways without departing from the gist thereof. Nor. For example, in the embodiment described above, the second inspection method and the third inspection method may be different from each other as necessary. That is, in the second inspection, the speed of the stage 11 may be slowed down or stopped at the defect position, and the threshold voltage of the defect detection and removal circuit 15 may be set higher than that in the first inspection; In some cases, the speed of the stage 11' may be set as in the first inspection, and the threshold voltage in the defect detection and removal circuit 15 may be set higher than in the first inspection (to the extent that the thickness of unacceptable defects can be detected). setting)
It may be 1-.

Furthermore, in the third test, the second test method described above may be repeated as necessary until the number of false detections is reduced, or various other detection methods may be applied. Furthermore, although the above embodiments have referred to photomasks (full size photomasks and reticle) in the semiconductor manufacturing field, the present invention is not limited thereto and can also be applied to wafer mask patterns. In this case, in detecting a defect in a mask pattern on a wafer, even if the wafer is opaque, the reflected light from the wafer 19 on which the mask pattern 18 is formed is reflected by the half mirror 20a, as shown in FIG.
What is necessary is just to make it enter into optical sensor 5a, 5b via 20b. Furthermore, in the above embodiment, two adjacent
By mutual comparison of two chips, if there is a defect in one of the corresponding positions of the two compared chips, the corresponding position of both chips is determined to be defective, and by simultaneously scanning adjacent chips one after another, each chip is Pattern defects (appearance defects)
Although this method is suitable for cases where three or more defects rarely occur in a row at the same corresponding position on each chip, the present invention Without being limited to this, it is possible to compare two chips in order to obtain acceptable detection accuracy, taking into account the extent to which defects occur consecutively at the same corresponding position on each chip (how many times will they occur, the probability of such occurrence, and various other conditions). Various test procedure programs are conceivable that employ this method.For example, the outputs of the optical sensors 5a and 5b are delayed, and two chips are compared according to a program that combines them appropriately.In other words, for example, for any given chip For example, we compare this chip with another chip next to it, and then with the chip C next to it, and if there is even one original defect in the corresponding position of one chip, we treat the corresponding position of both chips as defective. A program with high defect detection accuracy is conceivable, in which the same detection method is performed for each chip one after another.

[Application field]

The above explanation has mainly explained the background of the invention made by the present inventor and its application to photomasks and Uni-Nos in the field of semiconductor device manufacturing, which is a recent field of application, but the present invention is not limited thereto. . In short, the present invention can be applied to the case of inspecting the appearance, and is particularly suitable when the contrast of an appearance defect (for example, a portion different from the designed shape) is small and noise is easily detected erroneously.

[Brief explanation of drawings]

Fig. 1 is a simplified explanatory diagram of a Flying Bont 2-chip comparison type visual inspection device, Fig. 2 is an explanatory diagram of a 2-chip comparison type, and Fig. 3 is a simplified configuration showing an embodiment of the visual inspection device according to the present invention. Figures 4, 5, 7 and 8 are distribution diagrams of defects in the mask based on the first, second and third inspection results and final judgment results, respectively. 9 is a characteristic diagram showing the relationship between the threshold voltage (detection level) and the size of a detected defect in the visual inspection of a photomask. FIG. 9 is a simplified main part explanatory diagram showing another embodiment of the present invention. 1...CRT, 2...Light beam, 3...Photomask,
4a. 4 b...Lens, 5a, 5b-light sensor, 6 (6
a. 6b)...Mask pattern, 11...Stage, 1
2... Stage drive circuit, 13... CPU, 14...
...Comparison circuit, 15...Defect detection and removal circuit, 16...
-Memory, 18...Mask pattern, 19...Leafer, 20a. 20b...Half mirror-0 Fig. 3 Fig. 4 Fig. 5 Fig. 6 Misalignment /, i+-ru) τ Pressure C] - Fig. 7 Fig. 91 No? 26. z Go56L' 20 Ding No. 8 Figure 1 δb 4b

Claims (1)

[Claims] 1. Detecting shape defects in a plurality of patterns formed on a substrate by scanning the substrate using a flying spot generating cathode ray tube and comparing the patterns two by two. In the visual inspection method, the first inspection is performed by increasing the defect detection sensitivity and performing a full-scale inspection to detect defect positions including false detections, and the second and subsequent inspections are performed by increasing the defect detection accuracy. A visual inspection method characterized in that false detections are further removed from defect positions based on previous inspection results. 2. As the first inspection, the images of the flying spots irradiated on the corresponding positions of the two patterns are respectively detected by photodetectors, and the detection level for comparing with the difference signal of the detection outputs is set low, 2. The external appearance inspection method according to claim 1, wherein if the difference signal of the detection output is equal to or higher than the detection level, it is detected as a defect. 3. As the second inspection, the speed of the stage on which the substrate is placed is made slower or stopped at the defect position detected in the first inspection, and
The images of the flying spots irradiated on the corresponding positions of the two patterns are detected by photodetectors respectively, and the detection level is compared with the difference signal of the detection output.By setting the detection level low, it is detected in the first inspection. 3. The external appearance inspection method according to claim 1, wherein the defect position is determined to be a defect if the difference signal of the detection output is equal to or higher than the detection level, and other than that is determined to be a defect and removed as an erroneous detection. 4. As the third inspection, the stage is stopped at the defect position according to the second inspection result, and the detection level is continuously changed from a high value to a low value, and the detection occurs when the detection level is below a predetermined value. 4. The visual inspection method according to claim 3, wherein acceptable defects are removed from said defect positions. 5. A substrate on which multiple patterns are formed is placed on a stage, and defects in the pattern shape are detected by simultaneously scanning and comparing two patterns on the substrate using a flying spot generating cathode ray tube. The visual inspection apparatus includes a stage drive circuit that drives and controls the moving speed of the stage based on a control command, and a stage drive circuit that drives and controls the moving speed of the stage based on a control command, and images of flying spots irradiated on corresponding positions of two patterns on the substrate, respectively. There are two photodetectors to detect, a comparison circuit that compares the outputs of these photodetectors and sends out a deviation output, and a comparison circuit that compares the deviation output of this comparison circuit with a detection level that can be varied based on control commands to detect pattern defects. detect,
A visual inspection device comprising: a defect detection and removal circuit for removing minute defects among the defects; and a storage circuit for storing the defect position on the substrate based on a defect detection signal from the defect detection and removal circuit. . 6. The visual inspection apparatus according to claim 5, wherein the control command is issued based on an inspection procedure program input to the conbeater.