JP2001165635A - Inspection device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To precisely measure the edge position of the protruding part of a device having an irregular pattern formed on the surface with high reproducibility. SOLUTION: In a line width measuring device 1, the circuit line of one CMOS to be measured is imaged in the state where the circuit line and the pixel train of a CCD image sensor are inclined in the measurement of the line width of the circuit line. A plurality of optical profiles crossing in the lateral direction of the circuit line is generated from the taken image. A fitting curve is determined from the profiles to specify a plurality of edge positions. The edge positions are averaged to determine the final edge position.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an inspection apparatus for inspecting a device in which a concavo-convex pattern having a linear convex portion is formed on a surface. For example, the present invention relates to a device for inspecting a wafer pattern formed on a semiconductor wafer. The present invention relates to an inspection device applied as a line width measuring device for measuring a line width.

[0002]

2. Description of the Related Art In a semiconductor wafer manufacturing process,
For example, an inspection process of a line width of a wafer pattern such as a gate line width in CMOS is one of the important processes in manufacturing because it directly affects the yield of semiconductor wafers. At present, inspection of the line width of a wafer pattern is performed by a scanning electron microscope (SEM). SEM
Is a microscope using a method of irradiating an inspection target with an electron beam, and therefore, it is necessary to place a semiconductor wafer as a measurement sample in a vacuum. However, in recent years, the vacuum tank of the SEM has been increased in size with the increase in the size of the semiconductor wafer. Therefore, it is not possible to secure a sufficient number of devices at the time of constructing a semiconductor manufacturing process due to the cost and maintenance of the SEM, and the line width of the wafer pattern can be inspected for all the wafers due to the deterioration of the throughput due to this. Was gone.

The present applicant has proposed a microscope using an ultraviolet laser in Japanese Patent Application No. 11-164448. A microscope apparatus using an ultraviolet laser proposed in Japanese Patent Application No. 11-164448 is disclosed in Nd: YAG (1).
064 nm) and an ultraviolet light laser such as a fourth harmonic (266 nm) of a laser. If such an ultraviolet laser is used, for example, the circuit line width of the wafer pattern is 0.18 nm.
Even in a deep submicron class CMOS of 0.15 nm or less, the circuit line width of the wafer pattern can be sufficiently observed. for that reason,
The line width of the wafer pattern can be measured using a microscope apparatus using an ultraviolet laser instead of the SEM. A microscope device using such an ultraviolet laser cannot obtain absolute accuracy as high as an SEM because it is an optical type, but can be easily incorporated into a semiconductor manufacturing process because it can be measured in air. In addition, the line width of the wafer pattern can be measured at a lower cost and at a higher speed as compared with the SEM.

Hereinafter, a method of measuring the line width of a wafer pattern of a semiconductor wafer using a microscope apparatus using an ultraviolet laser will be described.

When an unevenness pattern having a linear projection having a line width substantially equal to the laser wavelength to be irradiated is picked up by an optical microscope, the linear projection of the observed image is cut in the width direction. As shown in FIG. 8, the obtained optical profile has an overshoot at its edge. This overshoot is caused by the influence of light interference, diffraction and the like.

The position where this overshoot appears is theoretically different from the edge portion of the linear projection.
However, since the peak portion of the overshoot has reproducibility with respect to the shape of the linear convex portion, the distance between the peak portions of the overshoot obtained from the profile is changed by the line of the linear convex portion. It can be defined relatively as width. Further, in order to determine the line width of the linear convex portion as an absolute distance, for example, by obtaining a correlation with a measurement result by a high-precision measuring device such as an SEM,
What is necessary is just to convert to a true value.

Specifically, the wavelength of an ultraviolet laser is 266n.
m, the gate line width was 15
When observing the CMOS of 0 nm, the observed image is as shown in FIG. 9, and the optical profile in the line width direction is as shown by the line segment X.

[0008] The profile thus obtained is CC
Since it is obtained from each pixel of a D (charg coupled device) image sensor, it is a sampling value at each pixel position. Therefore, a profile obtained as a sampling value is fitted by, for example, a quadratic function or a cubic function, and a fitting curve as shown in FIG. 10 is obtained. Subsequently, the minimum peak value of the fitting curve obtained by the fitting is obtained, and the edge position is specified. Then, the line width can be measured by obtaining the distance between the two edge positions.

As described above, by obtaining a fitting curve from a profile obtained as a sampling value of a pixel unit of a CCD, even in a microscope apparatus using an ultraviolet laser, the distance accuracy is equal to or less than the pixel interval of the CCD. , The distance between edges can be obtained with sub-pixel accuracy, and highly accurate line width measurement can be performed.

[0010]

As described above, in a microscope apparatus using an ultraviolet laser as described above, even if optical conditions and the like are made the same so that the same profile can be obtained, the original profile can be obtained. Contains higher-order components, so even if fitting with a profile curve represented by a function lower than that higher-order component,
A quantization error occurs. Therefore, when specifying the edge position by a microscope device using an ultraviolet laser, from the influence of the quantization error by sampling in pixel units, as long as fitting to specify the edge position with sub-pixel accuracy, Its reproducibility is limited.

For example, if the interval between pixels on a semiconductor wafer is 24 nm, the edge position fluctuates between about ± 2 to 3 nm due to the limit of reproducibility.

Specifically, the reproducibility when fitting a higher-order original function with a lower-order fitting curve will be described.

For example, as shown in FIG.
At the = (0,0) coordinate, fitting is performed by a cubic function on a function having a minimum peak value, and the position of the minimum peak value is obtained. For the function corresponding to the profile, a quartic function higher than the cubic function is used for convenience. Here, the sampling interval corresponds to the pixel interval of the CCD.

FIGS. 12A to 12F show fitting functions when fitting is performed by shifting the sampling points by 0.2. These are 0.2 samples
This corresponds to a lateral shift in the pixel (4.8 mm) object plane.

In this case, the minimum peak value (edge position) with respect to the pixel shift at the sampling point fluctuates within an interval of about ± 0.15 pixels with respect to one pixel interval as shown in FIG. For example, one of the CCD pixels
If the pixel is 24 nm, it will fluctuate between about 7 nm.

The present invention has been made in view of such circumstances, and is an inspection method capable of measuring the edge position of a convex portion of a device having a concavo-convex pattern formed on a surface with high accuracy and high reproducibility. It is intended to provide a device.

[0017]

SUMMARY OF THE INVENTION In order to solve the above-mentioned problems, an inspection apparatus according to the present invention is an inspection apparatus for inspecting a device having an uneven pattern having linear projections formed on a surface. Supporting means for supporting the device, imaging means for imaging the surface of the device by an imaging element having pixels arranged two-dimensionally, and pixels of the imaging element traversing the width direction of the linear projections Measuring means for generating a two-dimensional profile curve of the concavo-convex pattern in the width direction of the linear convex portion based on image data obtained from the column, and measuring an edge position of the linear convex portion; The imaging means shifts the relative position between the linear convex portion to be measured and the pixel of the image sensor at a pixel interval or less in the width direction of the linear convex portion, and displays the table of the device. The measuring means measures a plurality of edge positions in the width direction of the linear convex portion to be measured at each position shifted in the width direction by a pixel interval or less, and measures a plurality of measured edges. An edge position obtained by averaging the positions is obtained.

More specifically, the inspection apparatus is arranged such that one direction of the pixels arranged two-dimensionally in a matrix of the image pickup device is at a predetermined angle with respect to the width direction of the linear projection. The imaging means is disposed at an angle with respect to the device supported by the support means. Then, the inspection device generates a plurality of the profile curves based on a plurality of image data obtained from a plurality of pixel rows of the image sensor that traverses the width direction of the linear protrusion by the measuring unit, A plurality of edge positions of the linear protrusion determined from each profile curve are measured, and an edge position obtained by averaging the measured plurality of edge positions is determined.

Further, in the inspection apparatus, the supporting device may move the device at a pixel interval or less in the width direction of the linear projection, and the imaging device may move the device at a pixel interval or less. To image the surface of the device at each location. Then, the inspection apparatus generates a plurality of the profile curves based on the image data at each position captured by the imaging unit by the measurement unit, and generates a plurality of the linear protrusions obtained from each profile curve. Are measured, and an edge position obtained by averaging a plurality of measured edge positions is obtained.

[0020]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, an embodiment of the present invention will be described with reference to the accompanying drawings.

A line width measuring apparatus according to an embodiment of the present invention captures an image of a semiconductor wafer with a microscope apparatus using an ultraviolet laser having a wavelength of 266 nm, and measures a circuit line width of a wafer pattern of the semiconductor wafer from the captured image. Device.
This line width measuring apparatus is an apparatus for measuring a line width state of a resist pattern formed on a surface of a semiconductor wafer by a semiconductor exposure apparatus. The line width measuring apparatus does not directly inspect the unevenness of the device by etching the semiconductor wafer in this way, but inspects the resist pattern before etching.For example, a defect of the circuit line width of the wafer pattern occurs. This is because the semiconductor wafer can be used again by removing the etching film.

The semiconductor wafer to be inspected is, for example, a CMOS semiconductor wafer having a line width of 0.15 μm.

FIG. 1 shows a configuration diagram of a line width measuring apparatus according to an embodiment of the present invention.

A line width measuring apparatus 1 shown in FIG. 1 comprises a movable stage 2, an ultraviolet laser light source 3, a high-sensitivity low-noise camera 4,
, An optical fiber probe 5, a beam splitter 6,
It comprises an objective lens 7, an acousto-optic device 8, and a computer 9. Such a line width measuring apparatus 1 measures the line width of a semiconductor wafer 10 having a resist pattern formed on its surface.

The movable stage 2 is a stage for supporting a semiconductor wafer 10 to be inspected. The movable stage 2 has a function of supporting the semiconductor wafer 10 to be inspected and a function of moving the semiconductor wafer 10 to a predetermined inspection target position.

Specifically, the movable stage 2 includes an X stage, a Y stage, a θ stage, a Z stage, a suction plate, and the like.

The X stage and the Y stage move in the horizontal direction. The X stage and the Y stage move the semiconductor wafer 10 to be inspected in directions orthogonal to each other, and change the device pattern to be inspected. It leads to a predetermined inspection position. The θ stage is a so-called rotary stage for rotating a semiconductor wafer. At the time of inspection of the semiconductor wafer, the semiconductor wafer is rotated by, for example, the θ stage such that the device pattern on the semiconductor wafer is at a predetermined angle with respect to the high-sensitivity low-noise camera 4. The Z stage is a stage that moves in the vertical direction, and adjusts the height of the stage. The suction plate is for sucking and fixing the semiconductor wafer to be inspected.

The ultraviolet laser light source 3 is an ultraviolet laser light source having a wavelength of 266 nm. For example, an Nd: YAG fourth harmonic all-solid-state laser is used. As the ultraviolet laser light source, for example, a laser beam of 193 nm having a shorter wavelength or an ultraviolet lamp may be used as a light source instead.

The high-sensitivity low-noise camera 4 is a camera with high sensitivity to ultraviolet light.
A CD image sensor is provided, and the surface of the semiconductor wafer 10 is imaged by the CCD image sensor. This high-sensitivity low-noise camera 4 cools the main body,
It suppresses thermal noise, readout noise, circuit noise, etc. generated in image sensors and the like.

The optical fiber probe 5 is a waveguide for ultraviolet laser light, and guides the ultraviolet laser emitted from the ultraviolet laser light source 3 to the beam splitter 6.

The beam splitter 6 reflects the ultraviolet laser light from the ultraviolet laser light source 3 and irradiates the semiconductor wafer 10 on the movable stage 2 via the objective lens 6 and, at the same time, is reflected from the semiconductor wafer 10. The reflected light is transmitted to irradiate the high-sensitivity low-noise camera 3. That is, the beam splitter 6 is a laser beam separator for separating the optical path of the optical system of the emitted light of the ultraviolet laser light source 3 and the like from the optical path of the optical system of the reflected light to the high-sensitivity low-noise camera 4.

The objective lens 7 is an optical element for expanding and detecting the reflected light from the semiconductor wafer 10. The objective lens 7 has, for example, an NA of 0.9 and a wavelength of 266.
The aberration is corrected in nm. This objective lens 7
Is disposed between the beam splitter 6 and the movable stage 2.

The acousto-optic device 8 controls the amount of ultraviolet laser light emitted from the ultraviolet laser light source 3.

The computer 9 controls the lighting of the laser beam from the ultraviolet laser light source 3, controls the acousto-optic device 8, controls the moving position of the movable stage 2, and the like. Further, the computer 9 captures an image of the semiconductor wafer 10 captured by the CCD image sensor provided in the high-sensitivity low-noise camera 4, analyzes the image, and measures the line width.

In the line width measuring apparatus 1 having the above configuration,
The semiconductor wafer 10 is irradiated with an ultraviolet laser emitted from the ultraviolet laser light source 3 via the optical fiber probe 5, the beam splitter 6, and the objective lens 7. The ultraviolet laser light applied to the semiconductor wafer 10 is reflected on the surface of the semiconductor wafer 10. The reflected light enters the high-sensitivity low-noise camera 4 via the objective lens 7 and the beam splitter 6. And the high sensitivity low noise camera 4
The incident reflected light is imaged by a CCD image sensor, and surface image information obtained by the imaging is supplied to the computer 9.

Here, in the line width measuring apparatus 1, an image capturing optical system including the objective lens 7 and the like is used so that the size of one pixel of the CCD image sensor on the semiconductor wafer 10 is, for example, 24 nm. Has been adjusted. In other words, a region of 24 nm square is sampled as one pixel, and the surface of the semiconductor wafer 10 is imaged. Therefore, for example, if the number of pixels of the CCD image sensor is 1000 × 1000, the size of the viewable field of view is 24 μm × 24 μm.

Further, as shown in FIG.
A circuit line to be measured as 0 (a resist is formed to have a convex shape, and the shape as viewed from above the semiconductor wafer 10 is linear) is represented by a line segment X in the drawing. The direction of the semiconductor wafer 10 is adjusted by the movable stage 2 such that an inclination of a predetermined angle (θ) occurs between the pixel row of the CCD image sensor on the semiconductor wafer 10 and the pixel row of the CCD image sensor. That is, the angle between the line connecting the imaging points of each pixel of the CCD image sensor with a straight line in the direction crossing the circuit line and the width direction of the circuit line whose line width is to be measured is set to a constant angle. Note that since a CCD image sensor has a plurality of pixels formed in a two-dimensional lattice, two orthogonal rows can be represented in the case of a pixel row. It refers to a pixel row that crosses the route in the width direction.

Then, in the line width measuring apparatus 1, the computer 9 controls the movable stage 2 to
The positional relationship with the CD image sensor is adjusted to the tilted state as described above, and the circuit line width is measured based on the surface image of the semiconductor wafer 10 captured in the tilted state.

Next, a method for measuring the circuit line width from the imaged surface image of the semiconductor wafer 10 will be described.

First, FIG. 3 shows an example of a surface image of the semiconductor wafer 10 taken by the high-sensitivity low-noise camera 4.

First, a band-like circuit line is projected on the surface image, and the width direction of the circuit line (X direction in the figure) is in relation to the pixel row (Y direction in the figure) of the CCD image sensor. The image is tilted at a predetermined angle (θ).

Subsequently, with respect to this surface image, data of a pixel row in a direction crossing the circuit line (Y direction in the figure) is taken in, and an optical profile in this direction is obtained. Since this profile is data sampled in pixel units of the CCD image sensor, the pixel interval (2
It is a discrete value representing the optical intensity at 4 nm intervals).
Then, this optical profile is obtained for the n pixel rows. For example, pixel column number Y shown in FIG.
For the pixel columns of up to ₁ to Y _n, determine the optical profile.

FIG. 4 shows an example of the obtained optical profile.
Shown in As shown in FIG. 4, the optical profile has two overshoots corresponding to the edge positions of the convex portions of the circuit lines due to the influence of light interference. The position where the overshoot appears is theoretically different from the edge portion of the linear convex portion. However, since the peak portion of the overshoot has reproducibility with respect to the shape of the linear convex portion, the distance between the peak portions of the overshoot obtained from the profile is changed by the line of the linear convex portion. It can be defined relatively as width. Hereinafter, the peak portion of this overshoot is regarded as the edge portion of the circuit line. In order to determine the line width of a circuit line as an absolute distance, for example, a correlation with a measurement result by a high-precision measuring device such as an SEM may be obtained and converted into a true value.

Subsequently, the overshoot portion appearing in the optical profile is fitted one by one using, for example, a quadratic function or a cubic function to obtain a fitting curve. Then, a peak value of this curve is obtained from the obtained fitting curve, and the position of this peak value is specified. The position of this peak value is determined to a unit accuracy equal to or less than the pixel interval. The position of this peak value is the edge position of the circuit line obtained from this optical profile.

Subsequently, the process of generating the fitting curve of the overshoot portion and specifying the edge position is performed on all of the plurality of optical profiles previously obtained. Thus, a plurality of edge positions are specified for one circuit line.

Subsequently, a plurality of edge positions specified from a plurality of optical profiles are averaged. Then, the average edge position obtained by the averaging is specified as the final edge position of the circuit line to be measured.

Subsequently, these averaged edge positions are obtained for two overshoots appearing in the optical profile, and the interval between these two edge positions is obtained.
The value obtained by multiplying the interval between the edge positions by cos (θ) becomes the circuit line width of the wafer pattern to be measured.

By performing the above-described processing, the line width measuring apparatus 1 generates a plurality of optical profiles crossing in the width direction of one circuit line to be measured. Since the positional relationship between the pixel row of the CCD image sensor and the width direction of the circuit line to be measured is inclined at a predetermined angle (θ), each generated optical profile is shifted with respect to the edge position of the circuit line. The sampling positions slightly fluctuate. It should be noted that the amount of this variation must be smaller than the pixel interval. That is, in this example, since the pixel interval is 24 nm,
It is necessary to vary by less than this 24 nm.

The line width measuring apparatus 1 determines a fitting curve from a plurality of profiles whose sampling positions fluctuate below the pixel interval, and specifies a plurality of edge positions. Then, the plurality of edge positions are averaged to obtain a final edge position.

As a result, in the line width measuring apparatus 1, the edge position of the circuit line can be specified with a detailed accuracy smaller than the pixel interval, and the quantization error caused by the influence of the discrete image data can be suppressed. The edge position can be specified with high reproducibility.

The inclination angle (θ) between the circuit line and the pixel row needs to be set so that the sampling position fluctuates within a pixel interval between a plurality of pixel rows to be averaged. Therefore, this tilt angle (θ) is
It is defined as follows.

When the number of pixel rows to be averaged is n and the pixel size of the CCD image sensor is δ ₁ × δ ₂ , the inclination angle (θ) is tan ⁻¹ (δ ₁ × δ ₂ ) [rad]. Integer multiples.

Further, since the relative position between the CCD image sensor and the semiconductor wafer 10 may have the above-described relationship of the inclination angle (θ), the semiconductor wafer 10 is rotationally moved by the movable stage 2. The adjustment may be performed by rotating the high-sensitivity low-noise camera 4.

Next, the results of the reproducibility of the edge position with respect to the tilt angle (θ) as measured by the above method will be described.

First, FIG. 5 is a graph showing the reproducibility (pixel) of the edge position with respect to the inclination angle (θ) when the number (n) of pixel rows to be averaged is set to 80. As shown in FIG. 5, when averaging 80 columns, the reproducibility of the edge position improves as the angle increases as the inclination angle (θ) changes from 0 degree to 5 degrees.
At the same time, it can be seen that there is a change every fixed period. From the graph shown in FIG. 5, the pixel interval is 24n.
m, and the circuit line width to be measured is 0.15 nm,
It can be seen that a sufficient reproducibility can be obtained with a tilt angle (θ) of about 2 to 3 degrees.

FIG. 6 is a graph showing the reproducibility (pixel) of the edge position with respect to the tilt angle (θ) when the number (n) of pixel rows to be averaged is set to 10. As shown in FIG. 6, when 10 columns are averaged, the reproducibility of the edge position is determined by the inclination angle (θ) from 0 to 15 degrees.
It can be seen that when the angle is changed, it becomes better as the angle becomes larger, and at the same time, there is a fluctuation at a constant period. From the graph shown in FIG.
It can be seen that in the case where the circuit line width to be measured is 0.15 nm in nm, sufficient reproducibility can be obtained with an inclination angle (θ) of about 2 to 3 degrees. As described above, when the number of pixel rows to be averaged decreases, it is necessary to increase the tilt angle (θ).

FIG. 7 is a graph showing the reproducibility (pixel) of the edge position with respect to the tilt angle (θ) when the number (n) of pixel rows to be averaged is set to 10. The graph shown in FIG. 7 is a graph in a case where the inclination angle (θ) is set larger than the graph shown in FIG. As shown in FIG. 7, even when averaging 10 columns, if the inclination angle (θ) is too large, very large quantization noise is periodically generated. The cause is considered to be quantization noise in a direction orthogonal to the pixel row to be averaged. For this reason, even if the inclination angle (θ) is too large, the reproducibility may deteriorate. Therefore, it is preferable not to make the inclination angle (θ) too large.

As described above, in describing the embodiment of the present invention, as a method of generating a plurality of optical profiles crossing in the width direction of one circuit line to be measured with respect to a pixel row, Thus, an example in which a circuit line to be measured is inclined at a predetermined angle has been described. However, a method for generating a plurality of optical profiles that cross a circuit line to be measured in the width direction of the circuit line is not limited to such a method. That is,
It is sufficient that the semiconductor wafer can be imaged so that a plurality of optical profiles whose sampling positions with respect to the edge positions of the circuit lines fluctuate minutely (in subpixel units) can be detected. Other such techniques include, for example,
The movable stage 2 supporting the semiconductor wafer 10 is minutely (subpixel interval) in the longitudinal direction of the circuit line to be measured.
And a plurality of optical profiles may be generated by imaging the semiconductor wafer 10 at each moving position. Further, needless to say, the high-sensitivity low-noise camera 4 may be similarly moved.

In the above description of the embodiment of the present invention, a line width measuring device for measuring the line width of a resist pattern is taken as an example. It is not limited to the one that measures. For example,
The line width of the semiconductor wafer after the etching may be directly measured.
The circuit line width of the wafer pattern such as T may be measured.

[0060]

According to the inspection apparatus of the present invention, it is possible to measure the edge position of the convex portion of the device having the uneven pattern formed on the surface with high accuracy and high reproducibility.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of a line width measuring apparatus according to an embodiment of the present invention.

FIG. 2 is a diagram for explaining a positional relationship between a circuit line whose line width is to be measured and a pixel row of a CCD image sensor.

FIG. 3 is a diagram for explaining a positional relationship between a surface image of a semiconductor wafer and a pixel row.

FIG. 4 is a diagram for explaining an optical profile of a circuit line whose line width is to be measured.

FIG. 5 is a graph showing a reproducibility of an edge position with respect to a tilt angle (θ) when the number (n) of pixel rows to be averaged is set to 80;

FIG. 6 is a graph showing the reproducibility of an edge position with respect to a tilt angle (θ) when the number (n) of pixel rows to be averaged is set to 10;

FIG. 7 is a graph showing the reproducibility of the edge position when the inclination angle is larger than the inclination angle (θ).

FIG. 8 is a diagram illustrating an optical profile obtained by cutting a linear convex portion of an observed image in a width direction.

FIG. 9 is a diagram illustrating an observed image when a CMOS device having a gate line width of 150 nm is observed.

FIG. 10 is a diagram illustrating a fitting curve obtained by fitting an overshoot portion of an optical profile.

FIG. 11 is a diagram showing a quartic function having a minimum peak value at (x, y) = (0, 0) coordinates and a curve of a cubic function for fitting the quaternary function.

FIG. 12 is a diagram illustrating each cubic function obtained when a sampling position in fitting a quartic function is shifted in units of 0.2 pixels.

FIG. 13 is a diagram showing the reproducibility of the position of the minimum peak value when the quartic function is fitted with a cubic function.

[Explanation of symbols]

1 line width measuring device, 2 movable stage, 3 ultraviolet laser light source, 4 high sensitivity low noise camera, 5 optical fiber probe, 6 beam splitter, 7 objective lens, 8
Acousto-optic device, 9 Computer, 10 Semiconductor wafer

Continued on the front page F-term (reference) 2F065 AA12 AA22 AA51 BB02 BB05 BB18 CC19 EE00 EE08 FF42 GG04 GG22 JJ03 JJ26 LL02 LL04 LL46 LL57 NN01 NN02 NN20 PP05 PP11 PP12 PP13 QQ00 QQ03 BAQ01 Q05 AQ01 Q03 A DA08 EA30 EC02 EC03 4M106 AA01 AA02 AB09 BA07 CA39 DB04 DB08 DB18 DB19

Claims (5)

[Claims] 1. An inspection apparatus for inspecting a device having an uneven pattern having a linear convex portion formed on a surface thereof, comprising: a support means for supporting the device; and an image sensor having pixels arranged two-dimensionally. Imaging means for imaging the surface of the device, based on image data obtained from a pixel array of the image sensor traversing the width direction of the linear protrusion, based on image data obtained from the pixel row of the linear protrusion, Measuring means for generating a two-dimensional profile curve and measuring an edge position of the linear convex portion, wherein the image-capturing means sets a relative position between the linear convex portion to be measured and the pixel of the image sensor. The position is shifted in the width direction of the linear protrusion at a pixel interval or less, and an image of the surface of the device is taken.The measurement unit shifts the position in the width direction at a pixel interval or less. A plurality of edge position in the width direction of the linear convex portion to be measured at each position is measured, the inspection device characterized by obtaining the averaged edge positions a plurality of edge positions measured. 2. The imaging device according to claim 1, wherein the direction of one of the pixel rows two-dimensionally arranged in a matrix of the imaging element is at a predetermined angle with respect to a width direction of the linear protrusion. The measuring means is disposed at an angle with respect to the device supported by the means, and the measuring means is based on a plurality of image data obtained from a plurality of pixel rows of the image sensor that traverses the width direction of the linear convex portion. Generating a plurality of the profile curves, measuring a plurality of edge positions of the linear protrusion determined from each profile curve, and determining an edge position obtained by averaging the measured plurality of edge positions. Item 2. The inspection device according to Item 1. 3. The image sensor has a pixel size of δ ₁ × δ ₂ , and the measuring means generates n profile curves based on a plurality of image data obtained from n pixel rows. The imaging means may be configured such that an angle between a pixel row crossing the width direction of the linear protrusion of the image sensor and the width direction of the linear protrusion is tan ⁻¹ (δ ₁ / nδ ₂ ) or more. 3. The inspection apparatus according to claim 2, wherein the inspection apparatus is arranged to be inclined so that 4. The support means moves the device at a pixel interval or less in the width direction of the linear projection, and the imaging means moves the device at a pixel interval or less to move the surface of the device at each position. Imaging, the measurement means generates a plurality of the profile curves based on the image data at each position imaged by the imaging means, and calculates a plurality of edge positions of the linear convex portion obtained from each profile curve. 2. The inspection apparatus according to claim 1, wherein an edge position obtained by measuring and averaging a plurality of measured edge positions is obtained. 5. The inspection apparatus according to claim 1, wherein the measuring means measures the width of the linear projection based on edge positions at both ends in the width direction of the linear projection. .