JPH08271437A - Method and equipment for inspecting foreign matter - Google Patents

Abstract

PURPOSE: To detect a foreign matter with high sensitivity by subjecting the source light to shading correction so that the illumination intensity will be substantially constant in the accurate linear direction thereby preventing erroneous detection owing to the mismatch in the intensity of detection signal in the chip comparison process. CONSTITUTION: A shading correction optical system, i.e., an illumination optical system 31, employs a shading correction plate 231 which is inserted into the optical path in front or rear of a cylindrical lens 34 where the luminous flux is high. In the following chip comparison process, a chip pattern is irradiated with a light beam and a diffraction pattern is detected. Corresponding points of adjacent chips are determined and an image data detected by a linear image sensor 40 is stored in a memory 189. The corresponding points are compared sequentially by a comparison means 190 thus realizing comparison inspection. Although a memory 189 requires a capacity for temporarily storing the acquired image data, no mechanism section is required thus realizing a highly reliable system.

Description

Detailed Description of the Invention

[0001]

The present invention relates to a semiconductor wafer, a TFT
When mass-producing semiconductors, TFTs, etc. using various process processing devices such as a sputtering device and a CVD device for forming a film on a semiconductor substrate such as a substrate, a dry etching device for forming a pattern, a resist coating device, an exposure device, etc. In another process processing apparatus, it is possible to inspect a foreign matter adhered on a sample on which repetitive chips such as a semiconductor wafer and a TFT substrate are formed with a simple structure, and a foreign matter inspection method and apparatus therefor.

[0002]

2. Description of the Related Art Japanese Patent Laid-Open No. 5-218163 and Japanese Patent Laid-Open No. 3-44054 are known as conventional techniques relating to the above-mentioned foreign matter inspection method and apparatus.

[0003]

However, the structure of the inspection head provided with the illumination optical system and the detection optical system is simplified, and the chip comparison process does not affect the repeated chip pattern formed on the sample. The problem of detecting foreign matter adhering to the sample with high sensitivity has not been sufficiently considered.

An object of the present invention is to solve the above problems.
Higher sensitivity to foreign matter by chip comparison processing by uniformizing the illumination intensity of the illumination light that is linearly focused and irradiated on the sample in the same direction and making the intensity of the diffracted light detected from the adjacent chips the same. (EN) Provided are a foreign matter inspection method and an apparatus therefor which can be inspected.

[0005]

In order to achieve the above-mentioned object, the present invention provides a linear width by a shading correction plate having a plurality of curved transmission parts for the intensity distribution of light emitted from a light source. Direction is adjusted so that the illumination intensity is almost uniform in the linear longitudinal direction, and the light is condensed linearly from the oblique direction onto the sample on which the chips are repeatedly formed by the condensing optical system. Then, the scattered reflected light from the irradiated sample is collected by the detection optical system, received by the linear image sensor and converted into a signal, and these converted signals are compared between repeating chips. The foreign matter inspection method is characterized by inspecting as a foreign matter on a sample due to a mismatch. Further, according to the present invention, with respect to the intensity distribution of the light emitted from the light source, a shading correction plate formed with a plurality of curved transmission portions collects a Gaussian distribution in a linear width direction and illuminates a linear longitudinal direction. The intensity is corrected so that it becomes almost uniform, and the light is collected and irradiated linearly on the sample on which chips are repeatedly formed by the condensing optical system from an oblique direction, and the scattered reflection from the irradiated sample Foreign matter characterized by condensing light with a detection optical system, receiving it with a linear image sensor, converting it to a signal, comparing these converted signals between repeating chips, and inspecting as a foreign matter on the sample due to mismatch It is an inspection method. Further, according to the present invention, with respect to the intensity distribution of the light emitted from the light source, a shading correction plate having a plurality of curved transmission portions forms almost the same phase distribution in the linear width direction and the illumination intensity in the linear longitudinal direction. The light reflected and scattered from the irradiated sample is radiated by converging the sample linearly on the sample on which the chips are repeatedly formed by correcting the light so that it becomes almost uniform. The scattered light reflected by the linear image sensor is received by the linear image sensor after being collected by the detection optical system and passed through the spatial filter that shields at least the 0th-order diffracted light generated from the sample, and the converted signal is repeated. It is a foreign matter inspection method characterized by comparing chips and inspecting them as foreign matter on a sample if they do not match. The present invention also provides
A shading correction plate that forms a plurality of curved transmission parts for the intensity distribution of the light emitted from the light source collects a Gaussian distribution in the linear width direction and the illumination intensity is almost uniform in the linear longitudinal direction. The light is scattered and reflected from the irradiated sample by irradiating the sample on which the chips are repeatedly formed by the converging optical system after converging the light in a straight line. The linear image sensor receives scattered reflected light obtained through a spatial filter that collects at least the 0th-order diffracted light generated from at least the sample, converts it into a signal, and compares the converted signal between chips that repeat. Then, the foreign matter inspection method is characterized by inspecting as a foreign matter on the sample due to the mismatch.

Further, according to the present invention, a shading correction plate having a plurality of curved transmissive portions with respect to the intensity distribution of the light emitted from the light source makes the phase distribution substantially the same in the linear width direction and the linear longitudinal direction. An illumination optical system that illuminates and linearly collects and irradiates a sample on which chips are repeatedly formed by a condensing optical system with the illumination intensity corrected to be substantially uniform. Between the detection optical system that collects the scattered reflected light from the sample irradiated with the light and receives it with the linear image sensor and converts it into a signal, and the chip that repeats the signal converted from the linear image sensor of the detection optical system. A foreign matter inspection apparatus comprising inter-chip comparison means for comparing and inspecting as foreign matter on a sample if they do not match. Further, according to the present invention, with respect to the intensity distribution of the light emitted from the light source, a shading correction plate formed with a plurality of curved transmission portions collects a Gaussian distribution in a linear width direction and illuminates a linear longitudinal direction. An illumination optical system that illuminates the sample on which a chip is repeatedly formed by correcting the intensity so that the intensity is almost uniform and then collects and irradiates the sample linearly from the oblique direction. Comparison is made between a detection optical system that collects scattered reflected light from the sample and receives it with a linear image sensor and converts it into a signal, and a chip that repeats the signal converted from the linear image sensor of the detection optical system. And an inter-chip comparison unit that inspects as foreign matter on the sample due to mismatch. Further, according to the present invention, with respect to the intensity distribution of the light emitted from the light source, a shading correction plate having a plurality of curved transmission portions forms almost the same phase distribution in the linear width direction and the illumination intensity in the linear longitudinal direction. Is corrected so as to be substantially uniform, and is repeatedly condensed by the condensing optical system onto the sample on which the chips are formed. And a detection optical system for collecting scattered reflected light from the sample and receiving the scattered reflected light obtained by a linear image sensor through a spatial filter that shields at least the 0th-order diffracted light generated from at least the sample and a signal. And a chip comparison means for comparing the signals converted from the linear image sensor of the detection optical system between repeating chips and inspecting them as foreign matter on the sample due to mismatch. Is a foreign substance inspection apparatus according to. Further, according to the present invention, with respect to the intensity distribution of the light emitted from the light source, a shading correction plate formed with a plurality of curved transmission portions collects a Gaussian distribution in a linear width direction and illuminates a linear longitudinal direction. An illumination optical system that illuminates the sample on which a chip is repeatedly formed by correcting the intensity so that the intensity is almost uniform and then collects and irradiates the sample linearly from the oblique direction. Optical system for collecting scattered reflected light from the sample and passing through a spatial filter that shields at least 0th order diffracted light generated from at least the sample by a linear image sensor to convert it into a signal And a chip-to-chip comparison means for comparing the signals converted by the linear image sensor of the detection optical system between repeating chips and inspecting them as foreign matter on the sample due to a mismatch. A foreign matter inspection apparatus characterized.

Further, the present invention is characterized in that, in the foreign matter inspection device, each of the plurality of curved transmission parts in the shading correction plate of the illumination optical system is formed by a curved opening.

[0008]

With the above structure, the illumination intensity of the illumination light that is linearly condensed and irradiated onto the sample in an oblique direction is made uniform, and the intensity of the diffracted light detected from the adjacent chips is made the same. Foreign substances can be inspected with high sensitivity by the comparison process. Particularly, as the shading correction plate in the illumination optical system, by having a plurality of curved transmissive portions (openings) in the linear width direction in which the transmissive area of the central portion is reduced with respect to the peripheral portion in the at least linear direction, Light fluxes with each phase will pass through on average, and as a result, light fluxes with almost the same phase distribution will reach each point in the linear width direction (y-axis) and it will be possible to collect a Gaussian distribution. Becomes As a result, the heights at which light is most condensed (positions on the z axis) at the respective points in the linear width direction (y axis) are the same (the focus positions are the same), and the accurate linear longitudinal direction (x It is possible to perform shading correction with substantially uniform illumination intensity in the axial direction), prevent erroneous detection due to mismatch of detection signal intensities in the chip comparison process, and inspect foreign matter with high sensitivity.

Further, the shading correction plate can be manufactured at low cost, and as a result, the cost of the entire foreign matter inspection head can be significantly reduced.

[0010]

Embodiments of the present invention will be specifically described with reference to the drawings.

First, a first embodiment will be described with reference to FIGS. As a typical process processing apparatus, a CVD apparatus for depositing a thin film such as an insulating film on the work 4, a sputtering apparatus for depositing a metal thin film on the work 4, a deposited metal thin film There is an etching processing apparatus that forms a circuit pattern by performing etching processing on the above. FIG. 1 is a schematic view showing a foreign matter inspection device mounted on the process processing apparatus. This apparatus includes a process processing chamber 7, a work supply station (supply station (loader) 8a, recovery station (unloader)) 8 such as a loader or unloader.
There is one or a plurality of a and 8b, and the detection head 1a
And the scanning stage 1b and the work 4 are mounted, and the θ stage 3 performs rotational positioning (rotational correction) in at least the rotational θ direction (the θ stage 3 electrically rotates the image signal detected by the linear image sensor). Since it can be corrected, it is not always necessary.), And an inspection control device 11 for driving and controlling the scanning stage 1b and the Θ stage 3 and a foreign matter inspection device 1 provided with a data processing unit (CPU) 9 are installed. It has been configured. Before the work (wafer) 4 is transferred from the work supply stations 8a and 8b to the process processing chamber 7 through the buffer chamber 6 by the robot mechanism 5, the foreign matter inspection device 1 attaches foreign matter to the work 4. Is measured and the work 4 is transferred to the process processing chamber 7 by the handling robot mechanism 5 through the buffer chamber 6 to the process processing chamber 7. At this time, the foreign matter inspection device 1
The data processing unit (CPU) 9 of calculates the measurement result of foreign matter and the like, and when the value exceeds the management standard value Mp,
A method of controlling the robot mechanism 5 for transmitting and handling an alarm signal to the control device 10 for controlling the process processing device so as not to process the process in the process processing chamber 7 and returning it to the work supply stations 8a, 8b. It is also possible to take By doing so, it is possible to reduce the production of defects in the work in the process processing chamber 7, and it is also possible to improve the operating rate of the process processing apparatus. On the other hand, the work 4 transported to the process processing chamber 7 is subjected to film forming processing or etching processing in the process processing chamber 7. Next processed workpiece 4
When the robot mechanism 5 for handling returns the work to the work supply stations 8a and 8b, the foreign matter inspecting apparatus 1 measures the adhered state of foreign matter and the like to the work 4 after the process treatment, and then the work 4 is removed. The robot mechanism 5 for handling stores the work in the work supply stations 8a and 8b. At this time, the measured result of the adhered state of foreign matter or the like to the work 4 is processed by the CPU 9a of the data processing unit 9 shown in FIG. 2 and stored as a database in the memory 9b or a hard disk (not shown). The CPU 9a of the data processing unit 9 uses a memory 9b or a hard disk (not shown) as a database.
Before being supplied to the process processing chamber 7 stored in the process processing chamber 7 (before process processing), a map of foreign particles before processing showing a state of adhesion of foreign particles and the like on the work and when the processing processing chamber 7 is processed and discharged (process processing). The number of foreign matters or the foreign matter detection position is compared with a post-process foreign matter map indicating the state of adhesion of foreign matter or the like to the work piece), and the foreign matter or the like on the work piece in the process processing in the process processing chamber 7 is performed. The increased foreign matter map (increased foreign matter number and its position) by the processing indicating the adhered state is displayed on the monitor 9e or the printer 9f. (For the difference between the foreign matter map after the processing and the foreign matter map before the processing as described above, see Japanese Patent Laid-Open No. Hei 2
-170279 publication. In this way, the CPU 9a of the data processing unit 9 obtains a distribution (map) of foreign matter including the number and size (for example, three stages of large, medium, and small) of the foreign matter adhering to the work in the process processing apparatus.
Can be calculated and stored in the memory 9b or a hard disk (not shown).

As an example of the detection head 1a,
As described in JP-A-5-218163, there is a configuration shown in FIGS. 3 (a) and 3 (b). That is, FIG.
As shown in an enlarged view in (b), the illumination optical system 31 includes a semiconductor laser 32 that outputs laser light with high brightness (high intensity).
And a beam diameter expanding optical system 33 for expanding the beam diameter of the laser beam output from the semiconductor laser 32, and a laser beam expanded by the beam diameter expanding optical system 33 is linearly (slit-shaped) condensed. A uniaxial direction condensing lens (cylindrical lens) 34 and a mirror 35 for reflecting the laser beam condensed by the uniaxial direction condensing lens 34 and irradiating the work 4 linearly (slit shape) are provided. The laser beam is shaped to irradiate the surface of the work 4 at a shallow angle. Although the light is linearly condensed (slit-shaped) by the uniaxial condenser lens 34, it is possible to irradiate the laser light linearly (slit-shaped) with a scanning optical system such as a galvanometer mirror. Since it is necessary to perform high-speed scanning by the scanning optical system, the optical system becomes complicated, but the beam diameter expanding optical system becomes unnecessary, and a semiconductor laser that outputs laser light with high brightness (high intensity) can be used. Further, the detection optical system 36 has a wide field of view (a high NA (Numerical NA such as 0.4 to 0.6) that collects scattered reflected light generated from the surface of the work 4 by the irradiated linear laser light.
Aperture: numerical aperture) telecentric optics 37, 38 and the telecentric optics 37, 3
8 is provided with a variable spatial filter 39 arranged almost on the Fourier transform plane and a linear image sensor 40, and the scattered reflected light or the spatial frequency of the repetitive circuit pattern is extracted from the edge of the circuit pattern existing on the surface of the work 4. The variable spatial filter 39 shields the light, and the linear image sensor 40 receives the scattered and reflected light from the foreign matter. Then, if the work 4 or the detection head 1a is scanned by the scanning stage 1b in the uniaxial direction (x-axis direction) on the work 4 under the control of the control device 11, it is possible to inspect the adhered state of foreign matter or the like on almost the entire surface of the work 4. like,
As shown in FIG. 3A, the detection head 1a is configured by arranging six sets of the illumination optical system 31 and the detection optical system 36 in a zigzag pattern. Therefore, when the workpiece 4 or the detection head 1a is relatively scanned once in the uniaxial direction (x-axis direction), the adhered state of foreign matters or the like should be inspected over the entire surface of the workpiece 4 as shown in FIG. 4A. You can

On the other hand, as the detection head 1a, three sets of the illumination optical system 31 and the detection optical system 36 are arranged, and the work 4 or the detection head 1a is relatively uniaxially arranged. When scanning once in the (x-axis direction), FIG.
Although there is a region that is not inspected as shown in (b), it is possible to inspect the adhered state of foreign matter or the like in almost the entire region of the work 4. According to this, the detection head 1a can be simplified and the detection head can be scanned at high speed. Further, as the detection head 1a, two sets of the illumination optical system 31 and the detection optical system 36 are arranged, and the work 4 or the detection head 1a is controlled by the control device 11. When the scanning is performed three times in the uniaxial direction (x-axis direction) while relatively shifting in the y-axis direction, it is possible to inspect the adhered state of foreign matter or the like over the entire surface of the work 4 as shown in FIG. 4C. . However, in this case, it is necessary to scan the workpiece 4 or the detection head 1a three times in the x-axis direction while relatively shifting in the y-axis direction, and the number of sets of the detection heads can be significantly reduced, while the scanning is performed. The mechanism becomes complicated and the inspection time is long.

In FIG. 2, information of a large number of foreign matter inspection devices 1 and control devices of process processing devices (process gas flow rate, process gas pressure, work (wafer) temperature, voltage applied to the work, etc., is also input. .) 10 and the inspection device 22 including the foreign substance inspection device 22a and the pattern inspection device 22b are connected to the scanning electron microscope (SEM) 21.
a, secondary ion mass spectrometer (SIMS) 21b, scanning tunneling microscope (STM) 21c, spectroscope (ST
S) shows an entire system configuration including a foreign matter data analysis computer 20 and the like by the analysis device 21 configured by 21d and the like. The foreign matter data analysis computer 20 includes a CPU 20a having a memory 20b, a keyboard 20c and a mouse 20d for inputting data, a foreign matter analysis result, and an abnormal process device name for which an alarm should be displayed and an abnormal starting lot or wafer. A display device 20e capable of displaying, a foreign matter analysis result, an abnormal process device name for which an alarm must be displayed, and an output device 20f such as a printer for outputting an abnormal start lot or a start wafer, and the start of each process processing device. The external storage device (hard disk) 20g stores the correlation between the generation state of foreign matter and the estimated or confirmed cause of foreign matter generation in lot units or in wafer units. Then, the foreign matter data analysis computer 20 is input with data from a large number of foreign matter inspecting apparatuses 1 and the control unit 10 of the process processing apparatus in which the inspecting apparatuses are installed, and the starting lot exceeding the foreign matter management values Mp and Mq is started. Inspection wafer 22 and analysis device 2
The result of detailed analysis in 1 and its presumed cause of failure (input by the input means 20c) are input, and the state of occurrence of foreign matter is estimated in each processing lot unit or starting wafer unit in each process processing apparatus or An external storage device (hard disk) 20g stores the correlation with the confirmed cause of the foreign matter.

Next, the second embodiment will be described with reference to FIGS.
This will be described with reference to FIG. That is, FIG. 5A shows an embodiment in which the on-machine foreign matter inspection device (on-machine foreign matter monitor) 1 is mounted on the process film forming apparatus. This process film forming apparatus includes a loader (L) unit 102, an unloader (U / L) unit 103, a transfer chamber 104, and a reaction chamber 101. The transfer arm mechanism 5 is provided inside the transfer chamber 104 and above the transfer chamber 104. It has a detection head 1a 'for on-machine foreign matter monitoring. The wafer 4 is received from the loader unit 102 and is transferred to the reaction chamber 101 by the transfer arm mechanism 5. The wafer 4 processed in the reaction chamber 101 is returned to the unloader chamber 103 by the transfer arm mechanism 5. At this time, the surface of the wafer 4 which is being moved by the transfer arm mechanism 5 is monitored (inspected) by the on-machine foreign matter monitor detection head 1a ′ mounted on the upper side of the transfer chamber 104. In this embodiment, the surface of the wafer 4 after processing is monitored, but it is also possible to monitor the surface of the wafer 4 before or after processing.

FIG. 5B shows a cross section of the detection head unit 1a 'for monitoring the on-machine foreign matter. The inside of the transfer chamber 104 is in a vacuum atmosphere, and the wafer 4 which is being moved by the transfer arm 5 a of the transfer arm mechanism 5 is monitored (inspected) by the detection head 1 a ′ mounted on the upper side of the transfer chamber 104. The detection head 1a ′ is mounted on a vacuum compatible standard flange 106 mounted on the upper lid 105 of the transfer chamber 104. The detection head 1a ′ is composed of an illumination optical system 31, a detection optical system 36, and a wafer rotation detection system (details will be described later) 110. The wafer rotation detection system 110 detects the rotation of the wafer 4 so as to perform software. The surface of the wafer 4 which has been corrected (by image processing) and illuminated by the illumination optical system 31 is monitored (inspected) by the detection optical system 36. In addition,
Detection head 1a 'in the vacuum compatible standard flange 106
Illumination unit 106a, detection unit 106b, and detection unit 106c
Is composed of transparent parts. The mirror 35 may be mounted on the vacuum compatible standard flange 106. In the above embodiment, the monitoring (inspection) is performed via the vacuum compatible standard flange 106, but it is also possible to install the detection head main body 1a 'in the transfer chamber 104, that is, in the vacuum atmosphere by downsizing. is there.

Next, an embodiment for performing the chip comparison process will be described with reference to FIGS. Wafer rotation correction will be described. That is, the wafer rotation direction detector 110 (shown in FIG. 5) is provided immediately before the foreign matter detection optical system (detection head) 1a, 1a 'shown in FIGS. The wafer rotation amount obtained by the wafer rotation direction detector 110 is obtained when the wafer rotation stage (Θ stage) 3 is included, or when the hand of the robot mechanism 5 that handles the wafer has a rotation mechanism, or the foreign matter detection optical system. When the (detection head) 1a rotates, it is mechanically corrected accordingly. If there is no rotation correction mechanism, rotation correction is performed by an electric circuit or software processing. In the case of rotation correction by an electric circuit or software processing, since the rotation correction mechanism by the Θ stage or the hand of the robot mechanism or the foreign matter detection optical system is not required, the mechanism can be simplified and downsized. At the same time, it is possible to easily cope with changes in the dimensions of the wafer. First, in order to detect the rotation direction of the wafer, (1) the orientation flat direction of the wafer is detected. (2) The circuit pattern direction of the wafer is detected. Diffracted light detection Circuit pattern image detection (feature extraction) Strictly speaking, since the circuit pattern direction and the orientation flat direction are different, if the circuit pattern direction needs to be detected accurately, the circuit pattern direction should be detected. There must be.

The wafer 4 has a wafer rotation direction detector 110.
The direction of rotation of the wafer is detected when it passes underneath or is temporarily stopped. (1) A method for detecting the orientation flat direction of the wafer will be specifically described. A detection method of the first wafer rotation direction detector 110a will be described with reference to FIG. That is, the wafer 4 passes under the illumination system having several light emitting points 152 along the wafer moving direction V and moves from the position 153 to the position 154. The locus 15 on the wafer 4 of the illumination light emitted from the light emitting point 152 of the illumination system of the wafer rotation direction detector 151 is shown in the figure.
5 is shown. In the case of the light emitting point A, the time As when the illumination light starts to hit the wafer 4 and the time Ae when the wafer 4 deviates are measured, and this is performed for the other light emitting points B to G. Based on the above data and the movement time of the wafer 4, the orientation flat 1 of the wafer 4
The direction of 56 is obtained, and the rotation deviation amount θ1 of the wafer 4 is calculated. Further, as a method of detecting the rotation direction of the wafer 4, there are scribe area detection, chip detection, and special mark detection such as alignment mark.

From FIG. 7, the second wafer rotation direction detector 11 is shown.
0b and the detection method of the third wafer rotation direction detector 110c will be described. That is, as shown in (b), the second wafer rotation direction detector 110b emits the reflected light reflected from the outline including the illumination light source 162 that linearly illuminates and the orientation flat edge of the wafer 4 that is linearly illuminated. And a sensor 163 arranged linearly for detection, and based on the reflected image signal detected by the sensor 163, to obtain the direction of the orientation flat edge 156 (rotational deviation amount θ2 of the wafer 4) as in FIG. You can Here, in the case of such reflected light detection, in addition to detecting the orientation flat edge, the characteristics of the image of the circuit pattern (including scribe area detection, chip detection, alignment marks, etc.) on the surface of the wafer 4 are extracted. By doing so, the amount of rotation deviation of the wafer can be calculated and detected. Further, the second wafer rotation direction detector 110c, as shown in (c), has a linear sensor 167 for detecting the light source 166 for linearly illuminating and the light shielded by the contour including the orientation flat edge of the wafer 4. The direction of the orientation flat edge 156 can be obtained based on the shielded light image signal detected by the sensor 167 as in the case of FIG.

(2) A method of detecting the circuit pattern direction of the wafer will be specifically described.

First, a method of detecting the wafer rotation deviation amount by detecting the diffracted light from the circuit pattern on the wafer will be described. In FIG. 8, in order to detect the wafer rotation direction,
A method of detecting diffracted light from the circuit pattern on the wafer 4 by the wafer rotation direction detector 110d will be described. Light emitted from an illumination light source 171 composed of a laser light source or the like is condensed by a condenser lens 172 and reflected by a mirror 173 to illuminate a circuit pattern on the wafer 4 by oblique illumination.
The objective lens 174 provided with the diffracted light above and the detector 1
And 75. Here, the detector 175 is installed at the Fourier transform plane position of the objective lens 174, and the wafer 4 may be moving uniaxially or may be temporarily stopped. Detection images on the Fourier transform plane are shown in FIGS. FIG. 6B shows the case where the wafer is not rotated (wafer reference position), and the diffracted light (0th order light) 177 from the main circuit pattern or the repeating circuit pattern on the wafer is at the center of the detector 175 (Y-axis direction). Image in the center). FIG. 3C shows the case where the wafer is rotated, and the diffracted light (0th order light) 1 from the main circuit pattern or the repeated circuit pattern on the wafer is shown.
The image of 78 is deviated from the center of the detector 175 by Δd to form an image. The deviation amount Δd has a correlation with the wafer rotation deviation amount θ3, and the wafer rotation deviation amount θ3 can be obtained from the deviation amount Δd. That is, the CPU 176 connected to the detector 175
Can calculate the deviation amount Δd from the image signal based on the diffracted light detected by the detector 175, and can calculate the wafer rotation deviation amount θ3 from the deviation amount Δd. Here, since the light emitted from the illumination light source 171 is condensed and projected (projected) onto the surface of the wafer 4 from an oblique direction, the point light source as the illumination light source 171 has sharper diffracted light. Since the image is formed, the accuracy of detecting the rotation deviation amount of the wafer is increased. A TV camera, a one-dimensional linear sensor, a position sensor, or the like can be used as the detector.

Next, FIG. 9 shows a case in which vertical illumination is used in place of the oblique illumination in FIG. 8 as a method for detecting the diffracted light from the circuit pattern on the wafer in order to detect the wafer rotation deviation amount. (Wafer rotation direction detector 110e) is shown. That is, an illumination light source 181 and a half mirror 182 are used instead of the illumination light source 171, the condenser lens 172 and the mirror 173 in FIG. In this case, the unit main body can be downsized as compared with FIG. 8, and all the rotation deviation angles of the wafer (360 degrees) can be detected. That is, since the circuit pattern formed on the wafer 4 is mainly composed of an X-direction component and a Y-direction component, the wafer does not rotate (wafer reference position).
In the case of, the cross-shaped diffracted light (0th order light) 183 is imaged in the X and Y axis directions of the detector 175 as shown in (b). Diffracted light 18
4 is imaged with an angle θ4 offset from the X and Y axis directions of the detector 175. The CPU 176 connected to the detector 175 is
The angle deviation amount θ4 is calculated from the image signal based on the cross-shaped diffracted light detected by the detector 175, and the angle deviation amount θ is calculated.
4, the wafer rotation deviation amount θ4 can be calculated.

Next, in FIG. 10, as a method of detecting the diffracted light from the circuit pattern on the wafer in order to detect the amount of wafer rotation deviation, the detector and lens installed above in FIG. 8 are installed in an oblique direction. The case (wafer rotation direction detector 110f) is shown. That is, the detector 17 in FIG.
Instead of 5 and the lens 174, the detector 192 and the lens 193 were installed obliquely. The illumination optical system 194 including the illumination light source 195 and the condenser lens 196 is also arranged so that the detection optical system 191 including the detector 192 and the lens 193 faces each other.
In this case, when the wafer is not rotated (wafer reference position), the diffracted light (0th order light) 197 is focused on the center of the detector 193 (center in the Y-axis direction) as shown in FIG. In the case of rotation, the diffracted light 198 is imaged with the deviation Δd ′ and the angle Δθ5 deviated from the center (center in the Y-axis direction) of the detector 192 as shown in (c). The combined deviation amount Δd ′ + Δθ5 has a correlation with the wafer rotation deviation amount θ5, and the wafer rotation deviation amount θ5 is obtained from the combined deviation amount Δd ′ + Δθ5. That is, the CPU 176 connected to the detector 193 causes the shift amount Δd ′ synthesized from the image signal based on the diffracted light detected by the detector 192.
It is possible to calculate + Δθ5 and calculate the wafer rotation deviation amount θ5 from the deviation amount Δd ′ + Δθ5.

Next, FIG. 11 shows a method of using the circuit pattern image on the wafer by the wafer rotation direction detector 110g to detect the wafer rotation deviation amount. That is, the upper illumination optical system 201 configured by the illumination light source 202, the condenser lens 203, the mirror 204, and the half mirror 205.
Illuminates a repeated circuit pattern on the wafer 4, and an optical image of the illuminated circuit pattern on the wafer 4 is obtained by the objective lens 207, the imaging lens 208, and the detector 20.
It is detected by a detection optical system 206 composed of 9 and 9. When the detector 209 is a TV camera, a temporary still image of the wafer is obtained. When the detector 209 is a linear sensor, the wafer is uniaxially scanned, and an optical image of the circuit pattern formed on the wafer 4 is formed. It can be picked up and detected. The detected image is shown in FIG. The CPU 176 connected to the detector 209 can calculate the rotation deviation amount of the wafer 4 by detecting the inclination angle of the chip 210 or the direction of the scribe area 211 in the image shown in FIG.

Next, a technique relating to rotation alignment will be described. The subject of the present application is basically a semiconductor, but may be a liquid crystal display element and a substrate in the process of forming the fine pattern repeatedly formed, or a substrate on which other patterns are formed, for example. In the foreign matter inspection apparatus of the present application, comparison of adjacent chips is an important technology. This adjacent chip comparison technology knows the corresponding points of the adjacent chips, compares the detection output signal levels of the corresponding points, and if the signal levels are the same, it is determined that there is no defect or foreign substance in any chip, and the signal level is If the difference is large, it is determined that there is a defect or a foreign substance on the chip with the larger signal level. This technique requires finding corresponding points between adjacent chips as described above. Of course, it does not matter how to find this corresponding point. As an example, after acquiring all the data, the correlation function of the data may be obtained for the directions x and y which are considered to be the array direction of the chips, and the chip size p and the array direction θ may be calculated from the cycle of the correlation function. Since this method requires a large capacity in the storage device for storing data, the following method can be considered.
That is, the chip pitch p is taken in from the design data, and the direction of the chip arrangement is detected for each wafer, whereby the corresponding points between adjacent chips can be known. (FIG. 13) In this method, it is necessary to know the direction of the chip arrangement, that is, the rotation direction of the wafer. As one method, the orientation of the orientation flat (orientation flat) 156 formed on the wafer 4 may be measured. This is premised on the basis of the orientation of the orientation flat when forming the pattern. Further, as another method, the direction θ of the chip formed on the wafer may be directly measured. However, the pattern formed on the wafer is often complicated, and it is difficult to measure the pattern forming direction from the shape of the pattern. Here, the wafer rotation direction detector 11 is shown in FIG.
As indicated by 0e, when a light beam is irradiated onto the formed pattern, diffraction occurs due to the shape of the pattern, and the diffraction pattern can be detected. Since patterns on many wafers are often formed by repeating basic patterns having x and y directions, this diffraction pattern is mainly emitted in the x and y directions, in other words, shown in FIG. Like 0
The secondary diffracted light 184 is diffracted in the x and y directions. Therefore, by measuring the direction of the 0th-order diffracted light 184 from the detected image (diffraction image) 185, the direction θ of the pattern on the wafer 4 can be accurately known.

The measured chip arrangement direction θ is the corresponding point between the adjacent chips on the acquired data (a, shown in FIG. 14).
a'coordinate relationship 187) (to correct the chip arrangement direction θ rotation with respect to the chip pitch p by the rotation correction operator in the wafer rotation correction 187 by the electrical processing shown in FIG. 12). The obtained image data (the image shown in the memory range in FIG. 15) is sequentially stored in the memory 189, and the comparison means 190 is used.
Are these corresponding points (a and a ′, b and b ′ shown in FIG. 15).
The above comparative inspection (defect inspection of foreign matter etc.) is realized by sequentially comparing the above. This method requires a storage capacity for temporarily storing the captured image data in the memory 189, but can realize a highly reliable system in which the mechanical section is unnecessary. As another method, as shown in FIG.
By rotating the entire detection system 36 so that the longitudinal direction of the linear image image sensor (one-dimensional detector) 40 is parallel to the chip arrangement direction θ, the corresponding point on the adjacent chip is always on the detector 40. Can come to
With this method, the capacity for temporarily storing the image data in the memory 189 to know the corresponding points can be minimized. vice versa,
In this method, the entire detector 36 needs to be rotated, so a rotation mechanism section is required.

The above method has an effect that the method of arranging the chips of the work (wafer) in which the orientation flat or notch or the like is not formed on the work (wafer) can be detected. In addition, since this method directly measures the pattern on the work (wafer), it is possible to obtain a higher-precision measurement and a higher-precision corresponding point than the method using an orientation flat or the like. Hereinafter, a specific configuration of the rotation detection unit 110 will be described based on FIG. 12 (a similar configuration is shown in FIG. 9). The rotation detector 110e uses the light source 18 that is as close to the point light source as possible.
1, a half mirror 182, an imaging optical system 174, and a two-dimensional detector 175. The point light source 181 passes through the half mirror 182, the imaging optical system 174, and the wafer 4,
Image on detector 175. Where work (wafer)
4 can be basically considered as a mirror.

An infinity system (telecentric optical system) is provided between the imaging optical system 174 and the work (wafer) 4.
It is good to be configured. When this portion is configured as an infinity system, there is an effect that even if the position of the work (wafer) 4 in the optical axis direction slightly changes, the detection result is not significantly affected. An image (diffraction image) 185 detected by the detector 175 is shown in FIG. In many semiconductor patterns and liquid crystal display patterns, the 0th-order diffracted light 184 in the x and y directions as shown in the figure is detected as a cross-shaped straight line. Here, the direction of the diffracted light 184 corresponds to the direction of the chip arrangement θ on the wafer 4. From the image 185 detected here, the well-known Hough transform 186
Can measure the direction of the straight line that forms the cross shape. What is characteristic here is that when the work (wafer) 4 is placed perpendicularly or almost perpendicularly to the optical axis, the Hough transform, which is usually two-dimensional, can be compressed into one-dimensional. Next, the formulas (2) of two straight lines are shown.

Xsin θ ₁ −ycos θ ₁ = 0 (Equation 1) xcos θ ₂ + ysin θ ₂ = 0 (Equation 2) where θ ₂ = (π / 2) + θ ₁ The straight lines θ ₁ and θ ₂ are orthogonal to each other. Since we know, there is effectively one variable. Here, with the center of the optical axis as the origin of the image, x in the equations (1) and (2)
By substituting the coordinates on the straight line for y and y, θ can be calculated. FIG. 18A shows a histogram in which this θ is calculated for all points on the image and weighted by the detection output on the image at that time.
Create as shown in. The peaks of this histogram are θ ₁ and θ ₂ of the above straight lines. This is because a large signal is detected in pixels on a straight line.

Here, in reality, in order to improve the accuracy of θ, it is necessary to make the steps of the histogram have high resolution. However, if the resolution of the histogram steps is increased, the value on the vertical axis becomes smaller and the curve becomes less smooth,
The accuracy when calculating the peak will be reduced. Therefore, as a method of solving these trade-offs, the centroid calculating method shown in the following (Formula 3) is preferable. The formula (3) is a formula for calculating θ ₁ , but θ ₂ can be calculated similarly.

[0031]

(Equation 3)

Note that I (x, y) represents the intensity of the 0th-order diffracted light 184 of the detection image 185 shown in FIG. 12, for example. The method ranges rotational position theta ₁ or the theta ₂ is present (θ ₁ -δ ₀ <
It is necessary to know θ <θ ₁ + δ ₀ or θ ₂ −δ ₀ <θ <θ ₂ + δ ₀ ) in advance, and the detection accuracy improves when the existence range can be predicted more accurately. Therefore, it is necessary to align the orientation of each work (wafer), that is, the chip arrangement direction θ within a certain range in advance by aligning the orientation flat. Also,
If there is no diffracted light of the first or higher order, a perfect value can theoretically be calculated even if the addition range is 0 to 90 degrees. Therefore, it is desirable that the diffracted light intensity of the first or higher order is small. Needless to say, two-dimensional Hough transform may be used if the calculation time does not matter.

In the above equation, θ was calculated to obtain the histogram, but from the viewpoint of shortening the calculation time, t
The histogram may be obtained by calculating anθ, sinθ, or cosθ. If there is a slight misalignment of the optical axis,
θ may be calculated with an error. In the case of the method using this diffracted light, the fine shift of the optical axis is specifically a peak crest phenomenon of the histogram (see FIG. 18).
It is shown in (c). ). In the case of the diffraction pattern 184, as shown in FIG.
The angle with the error calculated at the time of misalignment is
It occurs because it exists on the + side and the-side. Therefore, as described above, when calculating the center of gravity, the deviation of the optical axis is canceled. In this sense as well, the method of obtaining the center of gravity based on the equation (3) can be expected to have high precision calculation.

The above method using the one-dimensional Hough transform can be applied even when the optical axis center is not on the detected image. Specifically, the center of the optical axis existing outside the image may be represented by the same coordinate system as the image. In this method using diffracted light, the detection signal output at the center of the optical axis is often extremely large, and the output of the detector is often saturated over a wide range around the center of the optical axis on the image. Therefore, it is often the case that the detection accuracy is improved by masking an appropriate range of the center of the optical axis and not using it for calculating θ. This mask is
It may be a light-shielding plate installed on the optical system, or a software s mask that does not use the data of the central portion of the detected signal. Actually, as shown in the detection image 185 of FIG. 12, the ± 1st-order or ± 2nd-order diffraction patterns 19 are formed.
2 is formed and detected. However, since these patterns 192 are usually weaker than the 0th-order diffracted light 184, they can be ignored by the method described below. The first is a method of binarizing using a threshold value, and the second is a method of using multivalued data. Although the case where the detector is two-dimensional has been described, the rotation detection system can be realized as follows even if the one-dimensional detector is used.

That is, the wafer rotation deviation can be corrected by the processing apparatus shown in FIG. This processing device (correction device) includes a rotation detection optical system 110e (detector 175), a rotation detection calculation system 186, an adjacent chip corresponding point vector calculation system (coordinate relationship between a and a ') 187, and a memory for storing image data. 189 and comparison means 190. The rotation detection calculation system 186 calculates the direction θ of the chip arrangement by the method described above. An adjacent chip correspondence vector is calculated from this θ and the data of the chip pitch p input from the design data. In the chip comparison means 190, corresponding points (a and a ′, b and b ′ shown in FIG. 15) are extracted from the memory information stored in the memory 189 by the adjacent chip correspondence vector in the longitudinal direction of the linear image sensor 40, and the signal is output. The outputs are compared, and the foreign object signal 188 due to the mismatch is detected. Here, the field of view must be larger than twice the chip pitch. Also, work (wafer)
Let θ be the allowable inclination (the range of the inclination of the work that can be allowed in performing the foreign matter inspection), the range 1 of the memory stored in the memory 189 is the length obtained by multiplying the visual field size of the linear image sensor 40 by sin θ. You need the capacity to record the range.

Next, the rotation correction by the work scanning direction comparison method will be described with reference to FIG. That is, although the adjacent chip to be compared has been described in the case of the longitudinal direction of the linear image sensor 40 as shown in FIG. 15A, the comparison target chip is the linear image sensor 40.
Need not be in the longitudinal direction, as described below,
It may be a direction perpendicular to the longitudinal direction of the linear image sensor 40, that is, the scanning direction of the work. Similar to the comparison of the linear image sensor 40 in the longitudinal direction, the corresponding points of the adjacent chips can be obtained by calculating the adjacent chip corresponding vector in the direction perpendicular to the longitudinal direction of the linear image sensor 40. The memory size (memory range 1) can be calculated by the same calculation.
In this case, chip comparison can be realized even when the detection field of view does not have a sufficiently large value with respect to the chip size, but on the other hand, the feeding accuracy of the work by the handling mechanism or the like is increased so that the adjacencies existing in the scanning direction are increased. It is necessary to know exactly the corresponding point of the chip. Also, in this scanning direction comparison,
When θ is close to 0 degree, there is an effect that influences such as illumination unevenness and detection lens unevenness are eliminated.

Next, the rotation correction by the rotation adjusting mechanism will be described with reference to FIGS. That is, with respect to the rotation correction by the processor using the memory, the direction θ of the chip arrangement calculated by the rotation detection calculation system 186 (see FIG. 17).
Shown in The rotation of the entire optical system 36 (the illumination system 31 and the rotation detection optical system 110 may be combined) or the work (wafer) based on (4) can simplify the calculation of the corresponding points of the adjacent chips. it can. Then, in the chip comparison means 191, the corresponding points are extracted from the memory information accumulated in the memory 189, the signal outputs are compared, and the foreign matter signal 188 due to the mismatch is detected. Further, as an effect of this method, even if the precision of the mechanism for scanning the work (wafer) (uniformity of the feed rate, vibration in the optical axis direction, vibration in the scanning direction detector direction, etc.) is poor, it can be dealt with. Further, by using the rotation adjusting mechanism as described above, it is not necessary to make the scanning direction and the chip arrangement direction vertical. Further, when the spatial filter 39 is used, the angle of the spatial filter 39 exactly matches the chip arrangement, so that the effect of the spatial filter 39 can be maximized.

Next, the 0th-order cut spatial filter system + chip comparison will be described. That is, when the chip comparison method is used, it is not always necessary to use the spatial filter 39 that blocks the n-th order diffracted light. In particular, the wafer 4
When the minimum cell size of the chip pattern formed above becomes small and the 1st or higher order diffracted light does not enter the detection lens 37, only the 0th order light may be blocked. In this case, there is an advantage that it is unnecessary to deal with the warp of the wafer 4 and the tilt of the wafer 4 which are necessary when the spatial filter 39 is used. FIG. 19 and FIG. 20 show the states of the diffracted light and the aperture of the detection lens when the chip arrangement direction on the wafer is as shown in FIG. FIG. 19 shows an incident angle α of illumination and an angle of view γ of the detection lens 37, and a certain spherical surface 282 is assumed.
83, a line of intersection 284 with the view angle of the detection lens 37 is shown in FIG.
0 is shown. 283 is the incident point on the spherical surface 282,
Reference numeral 5 is an emission point on the spherical surface 282. Here, by rotating the wafer 4 in the θ direction so that the 0th-order diffracted light does not enter the detection lens 37, the information of the main pattern on the wafer 4 can be erased. As a result, it is possible to emphasize and detect the foreign matter or defect attached to the pattern in the wafer 4. If the above chip comparison is carried out after the detection in this way, the above-mentioned foreign matters and defects can be detected. The condition for the detection optical system 38 to escape the pattern is the wafer 4 relative to x0 obtained by the following formula (4).
To rotate. That is, the wafer 4 is relatively rotated by x0−θ.

Sin α · sin (x0)> sin γ (Equation 4) At this time, the maximum value x0 (max) of the rotation angle x0 is obtained from the above Equation (4) by the following equation (5). . sin (x0 (max)) = sinγ / sinα (Equation 5) That is, the relative rotation stroke between the wafer 4 and the detection optical system 36 including the detector 40 is x0 (ma) above.
It suffices if it is chosen so that x) can be realized. Specifically, the numerical aperture (N of the objective lens 37 of the detection optical system 36 (N
A = sinγ is about 0.1 (focal length f = about 70 mm
(This is because a focal length f of about 50 mm or more is required to detect foreign matter on the work (wafer) 4 from the outside of the transfer chamber 104 through the transparent window 106 as shown in FIG. 5B. ), The depth of focus is approximately ± 60 μm (a depth of focus of approximately ± 40 μm to ± 30 μm is required to detect foreign matter on the work (wafer) 4 without automatic focusing), and α is 60 degrees. If x0 (ma
x) is about 6.6 degrees, and the numerical aperture (NA = sinγ) of the objective lens 37 of the detection optical system 36 is about 0.08 (focal length f = about 90 mm, depth of focus is about ± 60 μm), and α
Is 85 degrees, x0 (max) is about 4.6 degrees,
x0 (max), that is, 5 as a relative rotation stroke
It is sufficient if a rotation allowance range of about 10 degrees to 10 degrees can be realized. When the workpiece (wafer) 4 is θ, θ> x0 (ma
It is clear that it is not necessary to rotate the work (wafer) 4 or the detection optical system 36 if the wafer is conveyed under the condition of x).

The above method eliminates the need to install the 0th-order cut filter 39 in the detection optical system 36.
There is an effect that the imaging performance of the detection optical system 36 can be maintained. Here, it suffices that the 0th-order diffracted light can be shielded, so that the rotation angle θ of the wafer 4 may be set to 0 degree. In this case, the 0th-order cut filter 39 shown in FIGS. 3 and 22 is required. In this case, the 0th-order cut filter 39 is experimentally required to have a wider width, specifically, to sufficiently shield the 0th-order diffracted light. In this case, as described above, the method of detecting the rotation of the wafer 4 and performing the rotation correction to calculate the chip array vector to realize the chip comparison, and the method of not detecting the rotation of the wafer and not performing the chip comparison. Can be considered. However, the method of performing the chip comparison has lower detection sensitivity than the method of not performing the chip comparison. Particularly, in the pattern formed on the wafer 4, the 0th-order cut filter 3 has various directional components such as the rounded corners.
You can't remove everything at 9. Therefore, by comparing the chips, the ones that cannot be erased by the 0th-order cut filter 39 can be removed.

Next, the shading correction will be described. In the present invention, as described above, since the signal outputs are compared between the adjacent chips, the illumination needs to be irradiated with a more uniform intensity. That is, when there is no foreign substance between the adjacent chips to be compared, it is necessary that the linear image sensor 40 detect signals at substantially the same level. For that purpose, the linear light flux 41 which is condensed and irradiated linearly from the oblique direction needs a uniform (uniform) illumination intensity in the linear longitudinal direction (y-axis direction). However, the beam emitted from the light source 31 such as a semiconductor laser light source is
Generally, the center is strong and the periphery is weak. As a result, FIG.
As shown in (a), the illumination intensity is strong in the vicinity of the center of the linear visual field and weak in the surroundings. Therefore, a correction plate 231 that reversely corrects this intensity distribution is inserted near the exit position of the illumination optical system 31 (in front of the cylindrical lens 34) shown in FIG. The correction plate 231 may be a correction plate 231a having a curved slit 241 shown in FIG. 24, and has no wavelength characteristic and has a transmittance near the center (shown in FIG. 25B) as shown in FIG. .) Dropped N
It may be a D (Neutral Density) filter 231b. However, such an ND (Neutral Density) filter 231b is much more expensive than the correction plate 231a having the curved (hand-shaped) slits (openings) 241. Therefore, as the shading correction plate 231, a correction plate 231a having an inexpensive curved (drum-shaped) slit (opening) 241 is used.

FIG. 23 shows an illumination optical system 31 using a shading correction plate 231a as a shading correction optical system.
Indicates. The shading correction plate 231a may be located anywhere in the optical path, but it is easy to adjust the shading correction plate 231a.
Considering the manufacturing precision of a, etc., it is preferable to insert it before and after the cylindrical lens 34, which is a position where the luminous flux is large. Reference numeral 232 is a beam diameter expanding optical system (beam expander) that expands the luminous flux of the laser light emitted from the semiconductor laser light source 32.

FIG. 26 shows the illuminance distribution of illumination.
By inserting the shading correction plate 231a, the illumination having the distribution shown in FIG. 9A becomes a flat distribution (uniform distribution) as shown in FIG.

However, when the shading correction plate 231a shown in FIG. 24 is used, the cross-section in the width direction (x-axis direction) of the linearly focused light beam is weak near the center as shown in FIG. It has a light intensity distribution. Therefore, the illumination intensity fluctuates extremely due to a slight positional deviation of the illumination and a displacement (focus position variation) of the substrate (sample) 4 in the z direction (height direction). This is due to the phase distribution of the illumination light. Since this optical system uses the cylindrical lens 34, an equiphase surface as shown in FIG. 29 is formed. In FIG. 29, this equiphase surface is corrected by a correction plate 231.
Shown above a. The light passing through the correction plate 231a is linearly condensed on the y axis by the cylindrical lens 34. At this time, a part of the equiphase surface m of the illumination light flux is corrected by the correction plate 23.
Since the light is shielded by 1a, at each point of the y-axis in the light collecting surface,
Light fluxes having different phase distributions will pass through. As a result, the position on the z-axis where light is most condensed at each point on the y-axis is different (the focal position is different), and at a certain position on the y-axis where light is not condensed, the light near the center as shown in FIG. 27 is weak. It becomes a waveform of intensity distribution. Therefore, the illumination intensity fluctuates extremely due to a slight positional deviation of the illumination and a displacement (focus position variation) of the substrate (sample) 4 in the z direction (height direction).

Therefore, the present invention uses a shading correction plate 231c having a plurality of curved (hand-shaped) slits (openings) 241 'as shown in FIG. This allows the light beams having the same phase distribution to reach each point on the y-axis as a result.

Therefore, if the shading correction plate 231c having the shape shown in FIG. 30 is used, the phase of the light beam passing through the correction plate 231c becomes substantially the same at each point on the y axis, as shown in FIG. A Gaussian distribution can be condensed, and it is possible to prevent the illumination intensity from fluctuating due to a slight displacement of the illumination or a displacement (focus position variation) of the substrate (sample) 4 in the z direction (height direction).

Here, a shading correction plate having two curved (hourglass) slits (openings) 241 'as shown in FIG. 30 is shown, but in the present invention, as described above, the equal phase is obtained. It is needless to say that the same effect can be obtained by using two or more slits (openings) having a curved shape (a drum shape) because the purpose is to pass a light flux having a surface.

Further, in the above-described embodiment, the shading correction plate 231c is provided with a plurality of linear (hand-shaped) light-shielding plates.
Although the slit (opening) 241 'is formed, it is inexpensive even if a light-shielding film for shielding other than a plurality of curved (hand-shaped) slits (openings) is attached to a transparent glass plate. Obviously, it can be manufactured.

[0049]

According to the present invention, a semiconductor wafer, a TFT
Semiconductor devices such as semiconductor wafers and TFT substrates in various process processing devices such as a sputtering device and a CVD device for forming a film on a semiconductor substrate such as a substrate, an etching device for forming a pattern, a resist coating device, an exposure device, and a cleaning device. The foreign matter attached to the substrate can be inspected with high sensitivity by the simplified configuration.

Further, according to the present invention, the illumination intensity of the illumination light that is linearly condensed and irradiated on the sample in an oblique direction is made uniform, and the intensity of the diffracted light detected from the adjacent chips is the same. Thus, the foreign matter can be inspected with high sensitivity by the chip comparison process. Particularly, as the shading correction plate in the illumination optical system, by having a plurality of curved transmissive portions (openings) in the linear width direction in which the transmissive area of the central portion is reduced with respect to the peripheral portion in the at least linear direction, Light fluxes with each phase will pass through on average, and as a result, light fluxes with almost the same phase distribution will reach each point in the linear width direction (y-axis) and it will be possible to collect a Gaussian distribution. Next to
As a result, the heights at which light is most condensed (positions on the z axis) at the respective points in the linear width direction (y axis) are the same (the focus positions are the same), and the accurate linear longitudinal direction (x It is possible to perform shading correction with substantially uniform illumination intensity in the (axial direction), prevent erroneous detection due to inconsistency of detection signal intensities in the chip comparison processing, and inspect foreign substances with high sensitivity. Further, according to the present invention, the shading correction plate can be manufactured at low cost, and as a result, the cost of the entire foreign matter inspection head can be significantly reduced.

[Brief description of drawings]

FIG. 1 is a diagram showing an example in which a foreign matter inspection apparatus is installed in a process processing apparatus according to the present invention.

FIG. 2 is a diagram showing an overall system configuration according to the present invention.

FIG. 3 is a diagram showing a specific configuration of an inspection unit (detection head) according to the present invention.

FIG. 4 is a diagram showing the relationship between the number of channels and the scanning method in the inspection unit (detection head) according to the present invention.

FIG. 5 is a diagram showing an embodiment different from FIG. 1 when a foreign matter inspection device is installed in the process processing device according to the present invention.

FIG. 6 is a diagram for explaining an embodiment of detecting the orientation flat of the wafer and detecting the wafer rotation direction according to the present invention.

FIG. 7 is a diagram for explaining an embodiment different from FIG. 6 in which the orientation flat of the wafer according to the present invention is detected to detect the wafer rotation direction.

FIG. 8 is a diagram for explaining an embodiment in which a chip lattice formed on a wafer according to the present invention is detected to detect a wafer rotation direction.

FIG. 9 is a view for explaining an embodiment different from FIG. 8 in which the chip lattice formed on the wafer according to the present invention is detected to detect the wafer rotation direction.

FIG. 10 is a view showing a rotation direction of a wafer by detecting a chip grid formed on the wafer according to the present invention.
FIG. 6 is a diagram for explaining an example different from that of FIG.

FIG. 11 is a view showing a rotation direction of a wafer by detecting a chip grid formed on the wafer according to the present invention.
FIG. 11 is a diagram for explaining an example different from FIG. 10.

FIG. 12 is a diagram for explaining a method of detecting a rotation angle of a wafer from a detection image according to the present invention.

FIG. 13 is an explanatory diagram for applying a rotation correction operator in chip comparison according to the present invention.

FIG. 14 is a diagram showing a chip comparison system executed by software processing according to the present invention.

FIG. 15 is an explanatory diagram of FIG. 14.

FIG. 16 is a view showing a chip comparison system using a rotation mechanism according to the present invention.

FIG. 17 is an explanatory diagram of FIG. 16.

FIG. 18 is an explanatory diagram for detecting a rotation angle of a wafer from a detection image according to the present invention.

FIG. 19 is a front view showing the relationship between the illumination light, the diffracted light from the wafer, and the numerical aperture of the objective lens.

FIG. 20 is a plan view showing the relationship between illumination light, diffracted light from a wafer, and the numerical aperture of an objective lens.

FIG. 21 is a view showing a chip arrangement direction of a wafer.

FIG. 22 is an explanatory diagram in which a spatial filter blocks 0th-order diffracted light.

FIG. 23 is a diagram showing a case where a shading correction plate is inserted in the illumination optical system in the inspection unit according to the present invention.

FIG. 24 is a diagram showing a shading correction plate in which one curved (drum-shaped) opening is formed.

FIG. 25: Expensive ND as a shading correction plate
It is a figure for demonstrating a (Neutral Density) filter, (a) is a circular shape and (b) is a figure which shows the transmittance | permeability.

FIG. 26 is a diagram showing the intensity distribution of illumination light in the longitudinal direction (y-axis direction) of a linear light beam, where (a) shows the case without a shading correction plate and (b) shows the case without a shading correction plate. FIG.

FIG. 27 is a diagram showing an intensity distribution of illumination light in a width direction (x-axis direction) of a linear light beam when a shading correction plate having one curved (hand-shaped) opening is used. is there.

FIG. 28 is a Gaussian distribution intensity of illumination light in the width direction (x-axis direction) of a linear light beam when a shading correction plate having a plurality of curved (hand-shaped) openings according to the present invention is used. It is a figure which shows distribution.

FIG. 29 is a diagram showing an equiphase surface on a shading correction plate.

FIG. 30 is a diagram showing an example of a shading correction plate having a plurality of curved (hand-shaped) openings according to the present invention.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Foreign matter inspection apparatus, 1a, 1a '... Inspection unit (detection head) 1b ... Scanning stage, 3 ... Θ stage, 4 ... Work (wafer) 5 ... Handling mechanism (transfer robot), 5a ... Transfer arm 6 ... Buffer chamber , 7 ... Process processing chamber 8a, 8b ... Work supply station (loader, unloader) 9 ... Data processing unit (CPU), 10 ... Control device, 11 ...
Inspection control device 20 ... Foreign material data analysis computer, 20 g ... Database (hard disk) 21 ... Analysis device, 22 ... Inspection device, 31 ... Illumination optical system, 35 ... Mirror 36 ... Detection optical system, 39 ... Spatial filter 37 ... Telecentric optical system (Objective lens), 40 ...
Linear image sensor 41 ... Illuminated linear light flux, 101 ... Reaction chamber, 10
2 ... Loader section 104 ... Transport chamber, 105 ... Top lid, 106 ... Vacuum compatible standard flange 186 ... Rotation detection processing, 187 ... Rotation correction operator,
189 ... Memory 190 ... Comparison processing means 231, 231a, 231b, 231c ... Shading correction plate 241, 241 '... Opening

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Keiju Sakai, Hitachi Electronics Engineering Co., Ltd. 3-16-3 Higashi, Shibuya-ku, Tokyo (72) Inventor Hiroshi Morioka 292 Yoshida-cho, Totsuka-ku, Yokohama-shi, Kanagawa Bunch Co., Ltd., Hitachi, Ltd.

Claims (9)

[Claims] 1. A shading correction plate having a plurality of curved transmissive portions with respect to an intensity distribution of light emitted from a light source provides a substantially same phase distribution in a linear width direction and an illumination intensity in a linear longitudinal direction. The light is scattered and reflected from the irradiated sample by correcting it so that it becomes almost uniform and repeatedly collecting and irradiating the sample on which chips are repeatedly formed by the converging optical system in an oblique direction. Is detected by a linear image sensor and is converted into a signal, and the converted signals are compared between repeated chips and inspected as a foreign matter on the sample due to a mismatch, which is a foreign matter inspection. Method. 2. A shading correction plate having a plurality of curved transmissive portions with respect to an intensity distribution of light emitted from a light source collects a Gaussian distribution in a linear width direction and illuminates a linear longitudinal direction. The intensity is corrected so that it becomes almost uniform, and the light is collected and irradiated linearly on the sample on which chips are repeatedly formed by the condensing optical system from an oblique direction, and the scattered reflection from the irradiated sample A foreign substance characterized by condensing light with a detection optical system, receiving it with a linear image sensor, converting it to a signal, comparing these converted signals between repeating chips, and inspecting as a foreign substance on the sample due to mismatch Inspection methods. 3. A shading correction plate having a plurality of curved transmissive portions with respect to an intensity distribution of light emitted from a light source so that the phase distribution is substantially the same in the linear width direction and the illumination intensity is in the linear longitudinal direction. The light is scattered and reflected from the irradiated sample by correcting it so that it becomes almost uniform and repeatedly collecting and irradiating the sample on which chips are repeatedly formed by the converging optical system in an oblique direction. Is scattered by the detection optical system, and the scattered reflected light obtained through the spatial filter that blocks at least the 0th-order diffracted light generated from at least the sample is received by the linear image sensor and converted into a signal, and the converted signal is repeated. A method for inspecting foreign matter, characterized by comparing chips and inspecting them as foreign matter on a sample due to inconsistency. 4. A shading correction plate having a plurality of curved transmissive portions with respect to an intensity distribution of light emitted from a light source collects a Gaussian distribution in a linear width direction and illuminates a linear longitudinal direction. The intensity is corrected so that it becomes almost uniform, and the light is collected and irradiated linearly on the sample on which chips are repeatedly formed by the condensing optical system from an oblique direction, and the scattered reflection from the irradiated sample The linear image sensor receives scattered reflected light obtained through a spatial filter that condenses light with a detection optical system and shields at least 0th-order diffracted light generated from at least a sample, and converts the converted signal into a signal. A foreign matter inspection method characterized by inspecting as a foreign matter on a sample due to a mismatch between repeated chips. 5. A shading correction plate having a plurality of curved transmissive portions with respect to an intensity distribution of light emitted from a light source provides a substantially same phase distribution in a linear width direction and an illumination intensity in a linear longitudinal direction. And the illumination optical system that collects and linearly collects and irradiates the sample on which the chips are repeatedly formed by the condensing optical system from the oblique direction after correcting the A detection optical system that collects scattered reflected light from the sample and receives it with a linear image sensor and converts it into a signal is compared with a chip that repeats the signal converted from the linear image sensor of the detection optical system. A foreign matter inspection apparatus, comprising: inter-chip comparison means for inspecting foreign matter on a sample due to a mismatch. 6. A shading correction plate having a plurality of curved transmissive portions with respect to the intensity distribution of light emitted from a light source collects a Gaussian distribution in a linear width direction and illuminates a linear longitudinal direction. An illumination optical system that corrects the intensity so that the intensity is almost uniform and repeatedly collects and irradiates the sample on which the chips are repeatedly formed by the condensing optical system from an oblique direction, and the irradiation optical system. Comparison is made between a detection optical system that collects scattered reflected light from the sample and receives it with a linear image sensor and converts it into a signal, and a chip that repeats the signal converted from the linear image sensor of the detection optical system. And an inter-chip comparison unit that inspects as a foreign substance on the sample due to a mismatch. 7. A shading correction plate having a plurality of curved transmissive portions with respect to an intensity distribution of light emitted from a light source so that the phase distribution is substantially the same in the linear width direction and the illumination intensity is in the linear longitudinal direction. And the illumination optical system that collects and linearly collects and irradiates the sample on which the chips are repeatedly formed by the condensing optical system from the oblique direction after correcting the And a detection optical system for collecting scattered reflected light from the sample and receiving the scattered reflected light obtained by a linear image sensor through a spatial filter that shields at least the 0th-order diffracted light generated from at least the sample and a signal. A chip-to-chip comparison means for comparing the signals converted from the linear image sensor of the detection optical system between repeating chips and inspecting them as foreign matter on the sample due to mismatch. Foreign matter inspection apparatus for the. 8. A shading correction plate having a plurality of curved transmissive portions with respect to an intensity distribution of light emitted from a light source collects a Gaussian distribution in a linear width direction and illuminates in a linear longitudinal direction. An illumination optical system that corrects the intensity so that the intensity is almost uniform and repeatedly collects and irradiates the sample on which the chips are repeatedly formed by the condensing optical system from an oblique direction, and the irradiation optical system. Optical system for collecting scattered reflected light from the sample and passing through a spatial filter that shields at least 0th order diffracted light generated from at least the sample by a linear image sensor to convert it into a signal And a chip-to-chip comparing means for comparing the signals converted by the linear image sensor of the detection optical system between repeating chips and inspecting them as foreign matter on the sample due to a mismatch. Particle inspection apparatus according to claim. 9. A shading correction plate of the illumination optical system, wherein each of the plurality of curved transmission parts is formed by a curved opening.
Alternatively, the foreign matter inspection device according to item 8.