JPH0694632A - Automatic apparatus for inspecting specular surface - Google Patents

Abstract

PURPOSE:To inspect a flaw and a relative roughness with high precision even when the flow is slight, by obtaining two-dimensional image information with movement of a sample in one direction by using a CCD line sensor. CONSTITUTION:An automatic apparatus for inspecting a specular surface is equipped with a laser light source 1, an optical system 3 which transforms a laser light from this laser light source 1 into a line from a point, a CCD line sensor 2 which senses a reflected light from a sample, a means 7 for moving the sample perpendicularly to the linear laser light, a frame memory which stores an output of the CCD line sensor 2 as two-dimensional image information, and a means for measuring a relative roughness and a flaw by an image analysis of the two-dimensional image information stored in this frame memory.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an optical line displacement type sensor and a wide range mirror surface scratch and roughness inspection device to which image analysis is applied.

[0002]

2. Description of the Related Art Demands for mirror finishing of silicon wafers, stainless steel plates, marble, etc. are increasing due to ultra-precision processing in the field of semiconductors and upsizing of building interior materials. These, the degree of finish of the mirror surface, that is, "roughness"
As a device for inspecting, there are a contact needle type roughness meter, a laser displacement meter type roughness meter and the like.

[0003]

However, the contact needle type roughness meter has the following problems. It is a destructive inspection that scratches the mirror surface with a stylus. Since the roughness is measured with one point contact, the roughness on the line is inspected. Not suitable for a wide range of inspections. The laser displacement meter type roughness meter solves the problem of the contact needle type roughness meter by using a non-contact optical point displacement type sensor, but there is no change in the measurement of one point, so the problem is Remain. None of them are suitable for the roughness inspection of a wide range of surfaces.

On the other hand, for the inspection of scratches on the finished mirror surface, an apparatus applying image processing has been manufactured. This is an apparatus that attempts to detect a flaw by analyzing an image captured by a CCD area sensor because a flaw requires a wide area inspection over the entire inspection target. However, many scratches that are problematic on the mirror surface are shallow and small, and have not been sufficiently detected. The shallowest of these scratches are those that can only be detected with a roughness meter. Therefore, the problem to be solved by the present invention is to inspect scratches and roughness with high accuracy even if the scratches are shallow.

[0005]

In order to solve this problem, an automatic mirror surface inspection apparatus according to the present invention includes a laser light source, an optical system for converting laser light from the laser light source into a line from a point, and a sample light source. A CCD line sensor for receiving the reflected light, a means for moving the sample perpendicularly to the linear laser light, a frame memory for storing the output of the CCD line sensor as two-dimensional image information, and the same frame memory Means for measuring roughness and scratches by image analysis of the stored two-dimensional image information.

It can be said that the apparatus of the present invention is an improvement of the laser displacement type roughness meter. The following improvements were made to the laser displacement gauge type roughness meter. The incident light for displacement detection is made into a line from a point. A CCD line sensor is used as a displacement detection sensor. , Are called optical line displacement type sensors. A precision stage that moves vertically with respect to linear light is used to obtain displacement information as a two-dimensional image. Roughness and scratches are measured by image analysis.

However, the following problems occur in image analysis. 1) Concerning roughness inspection-In the point displacement detection element (PSD) that has been used conventionally, the unevenness of the brightness that has been structurally removed appears as a brightness unevenness on the two-dimensional image. -Non-planarity such as slight warp or distortion of the mirror surface is detected as displacement and appears as uneven brightness. Both are problems of uneven brightness in a two-dimensional image.

On the other hand, since the roughness of the mirror surface is a monotonically increasing function with respect to the standard deviation of luminance, the following measures were taken. The uneven brightness of the two-dimensional image is removed by regression calculation. -Roughness is calculated from the standard deviation of the image after removing the uneven brightness.

2) Regarding scratch inspection-Since there are almost no scratches on the mirror surface, the ratio of the area of the scratched portion to the total area is small, and it is difficult to set the judgment standard.

In response to this, the following measures were taken. Maximum-value projection of a two-dimensional image in one direction. The scratched portion is detected from the one-dimensional maximum brightness information obtained as a result of projection. In order to deal with scratches along the projection direction, projection in another vertical direction is also possible. The location of scratches can be detected from the maximum brightness information in two vertical directions.

[0011]

EXAMPLES Prior to the detailed description of the present invention, the measurement principle of the present invention will be described. In order to apply the CCD to the light spot shift method, it is necessary to convert the positional displacement of the sample into the output of the CCD, that is, the exposure amount (light intensity × time).

FIG. 5 shows the light distribution on the slit when the laser light is obliquely incident on the surface to be measured and a slit is formed on the optical axis of the reflected light. When O is the reference position of the surface to be measured, when O changes to O ′ and O ″, the light intensity of the light passing through the slit changes as shown by the shaded area in the figure.
If this change in the light intensity is detected by the CCD, an output corresponding to the change in the position of the surface to be measured can be obtained. In addition, the light spot displacement method generally receives a light spot (point spot) displacement corresponding to the position displacement of the surface to be measured by a position detection element (PSD).

However, in this method, since information can be captured only at points, the sample must be scanned in two dimensions in order to inspect the surface. On the other hand, when a one-dimensional CCD is used, information can be captured by a line, so that the sample can be scanned in one dimension.

Therefore, since a one-dimensional CCD is used as the light receiving element, the laser light is linearly spread by the rod lens. The linear light is obliquely incident on the surface to be measured, and the reflected light is received by the one-dimensional CCD. The state of the light beam at this time is shown in FIG.

Since the laser light is expanded in the X-axis direction by the rod lens arranged as shown in the figure, the pixel lines of the one-dimensional CCD are arranged in parallel with the X-axis. As can be seen from FIG. 2B, since the linear light emitted from the rod lens is not a parallel light beam, there is a possibility that only a part of the reflected light from the surface to be measured can be captured by the one-dimensional CCD. As shown in (c), a convex lens is provided in front of the one-dimensional CCD to collect light.

However, with the above system, it is not possible to obtain a sufficient light spot displacement for the measurement of a sample having minute irregularities. That is, in FIG. 5, even if the surface to be measured changes on the order of several microns to several tens of microns, the position of the reflected light does not change sufficiently on the slit, so that the CCD also changes the light intensity distribution passing through the slit. Cannot be changed so much as to detect significantly.

Therefore, as shown in FIG. 7, cylindrical (cylindrical) lenses L ₁ and L ₂ are installed. L ₁ is arranged so that the linear light is focused on the sample surface (it becomes the thinnest). The reason for arranging L ₁ in this way is to improve the resolution in the Y axis, that is, the scanning direction. The linear light emitted from L ₁ is reflected by the sample surface and enters L ₂ . The light emitted from L ₂ is emitted at an emission angle corresponding to the incident position and angle on L ₂ . Now, in the figure, L when the measured surface is displaced from O to O ′ by d
Letting θ be the emission angle from ₂ and D be the displacement of the linear light on the screen, the following equation holds. D≈R · θ (1) Here, R is the optical path length from L ₂ to the screen.
In this way, the displacement d of the surface to be measured is converted into the emission angle by L ₂ , and is enlarged by taking R longer.

FIG. 8 shows an optical system of the mirror surface inspection apparatus to which the above-mentioned light spot displacement method is applied. An embodiment of a device to which this is applied is shown in FIG.

In FIG. 1, 1 is a laser light source, 2 is CC
D camera (light receiving part), 3 is a lens, 4 is a cylindrical lens, 5
Is a power source for a laser light source, 6 is a sample stage, 7 is a scanning stage,
Reference numeral 8 is a control box, 9 is a vibration isolation table, and 10 is a clean bench.

FIG. 2 is a block diagram showing the configuration of the apparatus shown in FIG. In FIG. 2, 11 is a frame memory, 12 is a personal computer, 13 is a monitor display,
Reference numeral 14 is a video printer, 15 is a keyboard, and 16 is a mouse.

Experiments have shown that in the optical line displacement sensor, the mirror surface roughness has a monotonically increasing relationship with the standard deviation of luminance. By applying this fact, the unevenness of the luminance is removed by the regression calculation, and then the standard deviation of the luminance is obtained, whereby the mirror surface roughness in a wide area can be measured.

The procedure of a program example for realizing this is shown below. -Roughness measurement program (Fig. 3) Measurement image acquisition An optical line displacement sensor scans the measurement area to obtain a two-dimensional image in the frame memory 11. Acquisition of Horizontal Line Luminance Data The luminance data for one line is acquired in the horizontal direction from the frame memory 11. Regression calculation Regression calculation is performed based on the data for one line. Here, a linear regression equation was used simply. y = ax + b x: lateral position y: luminance A and b are obtained by regression calculation. Brightness unevenness correction Subtract the value of the regression equation from the data for one line. The standard deviation of luminance is obtained from the result of calculating the standard deviation of one line. Standard deviation total The standard deviation σ of each line is totaled. Repeat for all lines. Standard deviation average σ _{ave The} standard deviation collected in the calculation is divided by the total number of lines to obtain the average standard deviation σ _ave of the entire area. Substitution in the calibration curve S is obtained by substituting σ _ave in the calibration curve C () indicating the relationship between the average standard deviation σ _ave and the roughness S. Calibration curve C ()
Is experimentally determined.

Scratch inspection program If the scratch portion is directly extracted from the two-dimensional image obtained by the optical line displacement type sensor, the detection sensitivity of the scratch decreases,
It takes processing time. Therefore, the flaw is detected by the following procedure. This is called the maximum intensity projection method.

・ Example of flaw inspection program by maximum intensity projection method (FIG. 4) Measurement image acquisition The optical line displacement type sensor scans the measurement region to obtain a two-dimensional image in the frame memory. Initialization of maximum brightness storage area The value of the storage area max [X] holding the maximum value of all lines is set to 0. Y line luminance data acquisition The luminance data for one line in the X direction of the Yth line is acquired from the frame memory. X position luminance data comparison / update Obtained luminance data of the Y line X position and max [X]
, And if the value of max [X] is smaller,
Change to the brightness data value. Repeat processing for one line Repeat for one line. When the repetition is completed, the maximum brightness at each X position up to that point is stored in max [X]. Repeat for all lines Repeat for all Y lines. When the repetition is completed, the maximum brightness at each X position in all images is stored in max [X]. Comparison and Count of Depth of Defect of Scratch The maximum brightness value of max [X] is compared with a reference value of depth of scratch which is a reference for detection of a defect. Max exceeding the depth reference value
Count the number of [X] s. This is the total length L of the wound. Comparing / Judging with Acceptance Criteria The total length L of scratches is compared with the acceptance criterion L _S, and the total length L of scratches is L
If it is smaller than _S , it passes, and if it judges, it fails. The above program is a case where the maximum value is projected in the Y direction. X
If Y and Y are exchanged, the program projects the maximum value in the X direction.

[0025]

As described above, according to the present invention,
By using the CCD line sensor, two-dimensional image information can be obtained by moving the sample in one direction, and thus even if the scratch is shallow, the scratch and roughness can be inspected with high accuracy.

[Brief description of drawings]

FIG. 1 is an overall view of a mirror surface inspection device.

FIG. 2 is a block diagram of the entire apparatus.

FIG. 3 is a flow chart of a roughness measurement program.

FIG. 4 is a flowchart of a flaw inspection program.

FIG. 5 is an explanatory diagram of a light distribution on a slit.

FIG. 6 is an explanatory diagram showing how light spreads.

FIG. 7 is an explanatory diagram of a light spot displacement method.

FIG. 8 is an explanatory diagram of an optical system of the mirror surface inspection device.

[Explanation of symbols]

1 laser light source, 2 CCD camera (light receiving part), 3 lens, 4 cylindrical lens, 5 laser light source power source, 6 sample stage, 7 scanning stage, 8 control box,
9 anti-vibration table, 10 clean bench, 11 frame memory, 12 personal computer, 13 monitor display, 14 video printer, 15 keyboard,
16 mice

Claims (3)

[Claims] 1. A laser light source, an optical system for converting laser light from the laser light source into a line from a point, a CCD line sensor for receiving reflected light from a sample, and the sample for linear laser light. And a unit for vertically moving the CCD line sensor, a frame memory for storing the output of the CCD line sensor as two-dimensional image information, and a unit for measuring roughness and scratches by image analysis of the two-dimensional image information stored in the frame memory. An automatic mirror surface inspection device characterized by being equipped with. 2. The automatic mirror surface inspection apparatus according to claim 1, further comprising means for removing the brightness unevenness of the two-dimensional image information by regression calculation and calculating the roughness from the standard deviation of the image after the brightness unevenness is removed. 3. Two-dimensional image information is maximum-value projected in one direction, a flaw portion is detected from one-dimensional maximum luminance information obtained as a result of projection, and maximum-value projection is also performed in a direction orthogonal to the one direction. 3. The automatic mirror surface inspection device according to claim 1, further comprising means for identifying a flaw position from the maximum brightness information in these two directions.