JP3275425B2 - Defect detection apparatus and method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a manufacturing process for forming an object by forming a pattern on a substrate, such as a semiconductor manufacturing process, a liquid crystal display device manufacturing process, and a printed circuit board manufacturing process.
The present invention relates to a defect detection apparatus and method for detecting, analyzing, and taking measures for detecting a defect such as a generated foreign substance in a manufacturing process for analyzing the state of occurrence of the defect such as a foreign substance.

[0002]

2. Description of the Related Art In a conventional semiconductor manufacturing process, the presence of foreign matter on a semiconductor substrate (wafer) causes defects such as wiring insulation failure and short-circuiting. When present, the foreign matter causes destruction of an insulating film and a gate oxide film of the capacitor. These foreign substances are generated from various parts such as those generated from the operating part of the transport device, those generated from the human body, those generated by reaction in the processing device by process gas, and those mixed in chemicals and materials. Mixed in various states.

[0003] Even in the same liquid crystal display element manufacturing process, if foreign matter is mixed into the pattern or any defect occurs, the pattern cannot be used as a display element. The situation is the same in the manufacturing process of the printed circuit board, and the intrusion of foreign matter causes a short circuit of the pattern and a defective connection.

As one of the conventional techniques for detecting foreign matter on a semiconductor substrate of this type, as described in Japanese Patent Application Laid-Open No. Sho 62-89336, a semiconductor substrate is irradiated with a laser to illuminate the semiconductor substrate. Detects scattered light from foreign matter generated when foreign matter is attached to the substrate, and compares it with the inspection result of the same type of semiconductor substrate inspected immediately before, eliminating false alarms due to patterns, providing high sensitivity and high reliability. Japanese Unexamined Patent Publication (Kokai) No. 63-1358 allows inspection of foreign matter and defects.
As disclosed in Japanese Patent Application Publication No. 48-48, a semiconductor substrate is irradiated with a laser to detect scattered light from the foreign substance generated when the foreign substance adheres to the semiconductor substrate. Some are analyzed by an analysis technique such as luminescence or secondary X-ray analysis (XMR).

Further, as a technique for inspecting foreign matter, a wafer is irradiated with coherent light to remove light emitted from a repetitive pattern on the wafer by a spatial filter, and foreign matter and defects having no repeatability are emphasized and detected. A method is disclosed.

[0006]

In the above prior art, the mass production line in the semiconductor manufacturing process is not distinguished from the mass production line, and the inspection apparatus used in the mass production start-up work is directly applied to the mass production line. In a mass production line, it is necessary to quickly detect the occurrence of a defect such as a foreign substance and take a countermeasure. However, since the conventional inspection device was large in scale and had to be installed independently, the semiconductor substrate, liquid crystal display element substrate, and printed circuit board processed on the production line were brought to the inspection device to remove foreign substances and the like. The defect was inspected. Therefore, there is a problem that it takes time to transport these substrates and inspect for defects such as foreign substances, and it is difficult to inspect all of the substrates, and it is difficult to obtain a sufficient inspection frequency even with a sampling inspection. In addition, such a configuration has a problem that human labor is required.

SUMMARY OF THE INVENTION It is an object of the present invention to solve the above-mentioned problems of the prior art by enabling high-speed, high-precision, full-automatic detection of defects such as foreign substances, so that a total number of defect inspections and a sufficient defect inspection frequency can be achieved. It is an object of the present invention to provide a defect detection apparatus and method for realizing a sampling inspection and obtaining a highly efficient substrate manufacturing line.

[0008]

In order to achieve the above object, the present invention provides an illumination system for linearly irradiating plane wave light to a substrate having a repetitive pattern having a different pitch, and an illumination system comprising: An imaging optical system that forms an image of the reflected light from the irradiated substrate, and a spatial filter that is installed in the middle of the imaging optical system so as to shield the diffracted light from the small-pitch repetitive pattern on the substrate. A detector that detects the light image obtained through the spatial filter and formed by the imaging optical system, and among the signals detected by the detector,
Erasing means for comparing and erasing signals generated based on a repetitive pattern having a large pitch on the substrate obtained through the spatial filter; and defect detection for detecting a defect on the substrate based on a signal obtained from the erasing means. Means and a method for detecting a defect.

Further, the present invention provides a transporting means for transporting a substrate having a repetitive pattern, an illumination system for irradiating the substrate transported by the transporting means with a plane wave of light in a straight line, and an illumination system. An imaging optical system that forms an image of the reflected light from the illuminated substrate; and A spatial filter composed of a plurality of linear light-shielding objects having variable intervals, a detector that detects the light image obtained through the spatial filter and formed by the imaging optical system, and a detector that detects the light image. And a method for detecting a defect on the substrate based on the detected signal.

That is, in order to achieve the above object, the present invention has a configuration in which a foreign substance and defect inspection apparatus is arranged on a mass production line, and foreign substance and defect inspection is performed on all or almost all the substrates. As a specific configuration for realizing this, a small foreign substance monitoring device is arranged on a line for manufacturing a substrate such as a semiconductor, and is mounted in an input / output port of a processing device or a transport system between the processing devices.

Further, the present invention provides an illumination system, an imaging optical system, and a spatial filter disposed on a Fourier transform surface of the imaging optical system in a substrate production line for a mass-produced semiconductor or the like having a plurality of processing apparatuses. A foreign substance monitoring device that includes a detector disposed at an image forming position of the image forming optical system and detects a state of generation of a foreign substance on a semiconductor substrate; A semiconductor manufacturing method, comprising: a foreign substance occurrence state analyzing apparatus in a semiconductor manufacturing process, wherein the foreign substance occurrence state analyzing apparatus is provided in a transfer system between the apparatuses and detects a state of generation of foreign substances on a semiconductor substrate by the processing apparatus. That system.

More specifically, the present invention relates to a manufacturing system for a substrate such as a semiconductor provided with a plurality of processing apparatuses, a lighting system coherent in at least one axial direction, and means for detecting light from the illuminated substrate. Means for acquiring information of a repetitive pattern on the substrate from the detected light; an imaging optical system; a spatial filter disposed on a Fourier transform surface of the imaging optical system; A foreign matter monitoring device for detecting a state of occurrence of foreign matter on the semiconductor substrate, the foreign matter monitoring device including a detector arranged at an imaging position of the system.

Further, according to the present invention, in an apparatus for inspecting foreign substances on a semiconductor substrate, an illumination system for illuminating the semiconductor substrate with linear wave light having substantially a single wavelength and a plane wave, and illumination by the illumination system. An imaging optical system that forms an image of light reflected from the semiconductor substrate, and a spatial filter that is provided in the middle of the imaging optical system so as to block diffracted light from a repetitive pattern on the semiconductor substrate. Detector for detecting the light image obtained by the detection, erasing means for erasing a signal repeatedly generated on the semiconductor substrate among the signals detected by the detector, and a semiconductor substrate based on the signal not erased by the erasing means. A defect detecting device for foreign matter or the like, comprising: foreign matter detecting means for detecting the above foreign matter.

Further, the present invention is characterized in that, in the foreign matter and defect inspection apparatus, the image forming optical system is constituted by a refractive index variable lens array.

The present invention is also directed to an apparatus installed on a transport system for transporting a sample, comprising: an illumination optical system for illuminating the sample transported on the transport system; Optical system that forms an image at the detected position on a detector and detects the same, and a plurality of linear optical systems that are installed at the image forming positions of the illumination optical system and the light source of the illumination by the detection optical system and that have a variable distance from each other. And a processing circuit for processing a signal detected by the detector.

Further, the present invention is characterized in that the detection optical system is composed of two Fourier transform lens groups, and when an image of an object is formed, the image side is configured to be telecentric. It is a detection device.

Further, in the present invention, the two Fourier transform lens groups are constituted by lens groups having different numerical apertures, and have a structure in which one of the Fourier transform lens groups can be replaced with a lens group having a different numerical aperture. This is a feature of detecting defects such as foreign matters.

The present invention also provides a mechanism in which the direction of the linear structure of the spatial filter is configured to be parallel to the repetition direction of the pattern formed on the sample, and the sample is set substantially parallel immediately before the sample is inspected. And a mechanism for measuring the tilt of the sample after being installed almost in parallel, and a pattern formed on the sample by tilting the illumination optical system and the detection optical system according to the measured tilt so that the direction of the linear structure of the spatial filter is formed on the sample. A defect detecting device for foreign matter or the like, characterized by having a mechanism for making it parallel to the repetition direction.

Further, according to the present invention, the detection optical system has a linear detector, and the detection optical system has at least two or more repetitive patterns such as a chip pattern transferred onto a sample in its visual field. This field of view has the size
A defect detecting device for foreign matter or the like, characterized by having means for comparing and erasing pattern information other than a foreign matter or a defect by comparing two or more repeated patterns.

Further, according to the present invention, the comparing and erasing means determines whether the signal of one pixel on the detector is a foreign substance or a defect, and determines the signal level of one pixel on the detector. And comparing a signal level of a pixel at a corresponding position of an adjacent repetition pattern captured by the same detector as the detector with a signal level of a plurality of pixels adjacent to the corresponding position, and When a pixel having a value equal to the signal level of the one pixel is present in the signal level of a location or an adjacent location, the signal detected by one pixel on the detector is a signal from a repetitive pattern. A defect detecting device for foreign matter or the like, characterized by having a processing means for judging the defect.

Further, according to the present invention, the linear filter of the spatial filter may be so arranged that the repetition pitch of the repetitive pattern shaded by the spatial filter is larger than the width of the corresponding pixel or the area of the area where adjacent pixels are combined. A defect detection device for foreign matter or the like, characterized in that the pitch of the pattern is set.

The present invention further comprises means for performing a Fourier transform on a signal detected by the detector during the transport,
A means for calculating a pitch of a Fourier transform image by a repetitive pattern formed on the sample from a calculation result by the Fourier transform means; and a means for changing a pitch of the spatial filter based on the calculation result. It is a defect detection device.

The present invention also provides means for continuously changing the pitch of the spatial filter, and calculating the pitch at which the signal obtained from the detector takes a minimum while continuously changing the pitch. An apparatus for detecting a defect such as a foreign matter, comprising: means for changing a pitch of a spatial filter so as to have a pitch.

[0024]

In order to achieve the above object, a foreign substance and defect inspection apparatus is arranged on a mass production line to realize real-time sampling. It is configured so that it can be placed in the input / output port or in the transport system between the processing devices. That is, the present invention relates to an oblique illumination system including an illumination array, an imaging optical system including a lens array or a microlens group, and an imaging optical system including a plurality of processing apparatuses. A foreign matter monitoring device for detecting the occurrence of foreign matter on a semiconductor substrate, comprising a spatial filter arranged on a Fourier transform surface of the above and a detector arranged at the image forming position of the image forming optical system, By installing the processing apparatus at the entrance or the exit thereof, or in a transport system between a plurality of processing apparatuses, it is possible to detect a state of occurrence of a foreign matter defect on a substrate by the processing apparatus in a plate manufacturing process. By processing the foreign matter defect occurrence status and classifying the occurrence mode, and analyzing the component of the foreign matter defect generated, the cause of the foreign matter defect occurrence can be determined.

This is because the conventional apparatus has a large scale and requires a long inspection time. To realize a real-time monitor using these conventional apparatuses, it is necessary to arrange a large number of large-scale apparatuses. Was difficult. Realistically,
Inspection of one lot, one lot, or one semiconductor substrate every day has been the limit. In the inspection of the foreign matter and the defect at such a frequency, it cannot be said that the occurrence of the foreign matter was detected sufficiently early. In other words, it was far from ideal real-time sampling for mass production lines. Therefore, the foreign matter defect inspection apparatus having the configuration of the present invention can reduce the number of processes and the number of facilities in a mass production line by installing necessary and sufficient monitors at necessary and sufficient locations.

Further, by realizing a foreign substance inspection apparatus that can be placed on a fully automatic transport system, an inspection line without human intervention can be configured.

One of the main tasks for mass production start-up of LSI
Finally, there is an operation of investigating the cause of the generation of these foreign substances and taking countermeasures. In that work, detecting the generated foreign substance and analyzing elemental species and the like is a great clue for searching for the cause of the generation. On the other hand, in a mass production line, it is necessary to quickly detect the occurrence of these foreign substances and take measures. If the time elapses from the generation of foreign matter to the detection of the generation of foreign matter, the number of defective occurrences increases and the yield decreases. Therefore, in order to maintain a high yield, it is indispensable to reduce the elapsed time from the generation of a foreign substance to its detection. In other words, a foreign substance defect inspection apparatus capable of shortening the monitor sampling time enables real-time sampling in a mass production line, and can maximize the effect of foreign substance and defect inspection.

According to the present invention, a foreign substance on a semiconductor substrate can be detected in real time by installing the processing apparatus at the entrance or exit of the processing apparatus or at the transport system between a plurality of processing apparatuses.

Further, the present invention uses a foreign substance detection and analysis system in which a sampling semiconductor substrate is devised so that the evaluation at the time of mass production start-up proceeds smoothly and quickly. And take measures to prevent dust from the equipment.The results are fed back to each material, process, equipment, etc. ,
A foreign substance monitor on a semiconductor substrate is installed as necessary at a place where foreign substances are likely to be generated and used for mass production line inspection and evaluation specifications, or a specification that monitors only the increase or decrease of a specific foreign substance at a specific location Things. As a result, at the time of mass production start-up of semiconductor manufacturing processes, each process, equipment, etc. are evaluated by expensive and high-performance evaluation equipment in order to evaluate and improve (debug) materials, processes, equipment, designs, etc. It is possible to reduce the number of processes and equipment of the line as much as possible, and in particular, reduce the number of inspection and evaluation items, thereby shortening the cost of equipment and the time required for inspection and evaluation.

By separating the mass production line from the mass production line as described above, foreign matter detection, analysis,
The evaluation device can be operated efficiently and mass production can be started up quickly.In addition, foreign matter inspection and evaluation equipment used in the mass production line can be reduced to the necessary and simple monitoring device to reduce the weight of the mass production line. .

In the above-described mass production line monitoring apparatus of the present invention, the following method was focused on in order to solve a high-speed and small-sized inspection apparatus having a function equivalent to that of a conventional large-sized apparatus with the current technology. First, we focused on the repeatability of the memory. Conventionally, a method for detecting a defect by repeatedly removing a pattern is known. This method can ensure the detection performance. However, it is not mentioned that this method is advantageous for realizing the above-described monitoring device. Further, the monitor in this case does not need to monitor all points on the semiconductor substrate, but only needs to monitor the semiconductor substrate at a specific ratio. We focused on the fact that monitoring alone was very effective.

In the repetitive pattern, when the coherent light is irradiated, light is emitted only in a specific direction. That is, in the case of a memory, light emitted from a repetitive portion in a specific direction can be blocked by a spatial filter, and foreign matter that does not repeatedly occur can be detected with high sensitivity. At this time, if liquid crystal is used as the spatial filter, the shape of the spatial filter can be arbitrarily changed by turning on and off the liquid crystal, so that an inspection of an arbitrary repetitive pattern can be automatically performed.

The above-mentioned means improves the yield in semiconductor manufacturing for the following reasons. Through a strict detection experiment of the number of foreign substances on the semiconductor substrate, it was newly found that the number of foreign substances did not gradually increase or decrease but suddenly increased or decreased. Conventionally, the number of foreign substances was considered to gradually increase or decrease.
Defects such as foreign matter were inspected as frequently as once a day. However, at this inspection frequency, a sudden increase in foreign matter may be overlooked or may be detected after a while while increasing, resulting in a considerable number of defects. That is, in a mass production line, it is necessary to detect the occurrence of foreign matter as soon as possible and take a countermeasure. If time elapses from the occurrence of foreign matter to the detection of the occurrence of foreign matter, the number of defective occurrences increases and the yield decreases. Therefore, a high yield can be maintained by shortening the elapsed time from the generation of a foreign substance to its detection. In other words, by shortening the sampling time of the monitor, ideally, real-time sampling can maximize the effect of inspection for defects such as foreign matter.

Further, in the conventional apparatus, the semiconductor substrate is extracted and inspected. At this time, a new foreign matter adheres to the semiconductor substrate, which also lowers the yield.
With the foreign matter and defect inspection apparatus according to the present invention, the semiconductor substrate can be inspected without being extracted, so that the yield can be prevented from lowering due to foreign matter adhering to the semiconductor substrate.

In realizing a high-speed and small foreign matter inspection device,
The method using this spatial filter is based on the prior art (
The reason why the method is more suitable than the polarization detection method described in Japanese Patent Application Laid-Open No. 2-89336 will be described with reference to FIGS.

In the method of illuminating a sample with light and detecting scattered light from a foreign substance, scattered light from a pattern formed on the sample surface becomes noise. This noise increases as the size of the pixel (the minimum unit detected as one signal) of the detector 2006 increases, as shown in FIG. Since the pattern serving as a noise source is formed on almost the entire surface of the sample, the noise increases in proportion to the pixel size.

On the other hand, since the inspection time increases as the number of pixels increases, it is necessary to increase the pixel size in order to realize a high-speed inspection. Therefore, it is necessary to increase the pixel size and reduce the noise level. As a method of reducing this noise level, Koizumi et al., "Analysis of Light Reflected from LSI Wafer Pattern", Transactions of the Society of Instrument and Control Engineers, 17-2, 77/82 (1981), analyze the method using polarization. Have been. According to this, scattered light (noise) from a pattern is obtained by using polarized light.
Can be attenuated. However, the attenuation rate of the scattered light by this method depends on the direction of the detector as analyzed in the above-mentioned paper. For this reason, when condensing light emitted in various directions as in the case of using an imaging optical system, the respective attenuation rates are integrated to obtain an attenuation rate of 0.1% to 0.01%.
About.

On the other hand, in the method using the spatial filter of the present invention, the attenuation rate is reduced from 0.001% to 0.0001.
%. The reason will be described with reference to FIGS. 65 and 66. The wafer 2001 on which the repetitive pattern is formed is illuminated with the illumination light 2002, and the illuminated area is set in the lens system 200.
3 and 2005, an image is formed on the detector 2006. Here, FIG. 66 shows the intensity distribution of light emitted from the pattern on the Fourier transform surface on which the spatial filter 2004 is mounted.
Light emitted from the repetitive pattern is concentrated at a position corresponding to the pitch of the pattern. As an example of calculating this concentration ratio, the diffracted light intensity distribution in the case of a double slit is written by Hiroshi Kubota,
This is described in "Applied Optics" (Iwanami). According to this, when the number of slits (in the present application, the number of repetitive patterns that are simultaneously illuminated) increases, the concentration ratio increases. This ratio can also be calculated using the Fourier transform F []. Assuming that the shape of the illuminated pattern is a (x, y), the light intensity distribution at the position of the spatial filter is F [a (x, y).
y)]. Assuming that the shape of the spatial filter is p (u, v), p (u, v) * F [a (x, y)] is light passing through the spatial filter. If the shape of the figure complementary to the spatial filter is ￣p (u, v), then ￣p
(U, v) * F [a (x, y)] is a light component that is shielded by the spatial filter. The ratio of these two components becomes the previous attenuation rate. When this attenuation rate is calculated when the number of pattern repetitions is 3, it is about 0.001%. When the number of repetitions is 5, it becomes about 0.0001%, and when the number of repetitions is further increased, the attenuation rate decreases. Therefore, the attenuation rate can be reduced as compared with the case of using polarized light.

The above calculations are for the case where the pattern shape and other conditions are ideal, and may not always agree with actual experimental results. However, experimental results have shown that the attenuation rate is reduced by one to three orders of magnitude compared to the polarization method, and pattern noise can be reduced.

[0040]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The construction of a specific embodiment of an online monitor according to the present invention will be described below with reference to FIGS.

In this embodiment, as shown in FIG. 1, the illumination means 102, the detection optical system 103, the rotation adjusting mechanism 105, the spatial filter unit 106, the detector 107, the rotation detection means 108, the operational amplifier 201, the A / D converter Head 101, pitch detecting means 21 composed of detector 202
2. Operator processing system 203, foreign matter data memory 20
6, a pattern memory 208, a software processing system 210, a parameter transmission unit 209, a foreign substance memory 211, a coordinate data creation unit 232, a microcomputer 229, and a display unit 230.

Further, as shown in FIG.
Is a beam expander including a semiconductor laser 112, a collimator lens 113, a concave lens 114, and a receiver lens 115, a cylindrical lens 116, a mirror 11
8, the detection optical system is a Fourier transform lens 1
08, spatial filter unit 106, rotation detecting means 1
08 and a Fourier transform lens 111.

As shown in FIG. 3, the spatial filter unit 106 includes coil springs 121 and 122, a plurality of linear spatial filters 141, a coil spring support 119,
120, guide 125, right-hand thread 127, left-hand thread 12
8, worm gears 129, 130,
It is composed of a motor 140. The spatial filter unit 106 includes detectors 123 and 124 for detecting rotation.
Is installed.

As shown in FIG. 4, the operator processing system 203 includes a four-pixel adding unit 214 and an octalizing unit 21.
5. Cutting means 2 including a plurality of line memories 216
04, buffer memory 217, determination pixel extracting means 2
18, a comparison circuit group including a plurality of foreign matter comparison circuits 220, a threshold setting circuit 221, an OR circuit 224, and an AND circuit 226.

As shown in FIG. 5, the pitch detecting means 212 includes an FFT circuit 242, an operator pitch calculating means 241, a filter pitch calculating means 244, and a spatial filter control system 243.

As shown in FIG. 6, the rotation adjusting mechanism 105 includes a rotation guide 151, a rotation bar 152, and a spring 1.
53, piezo element 154, piezo element controller 1
55 and a frame 156.

(Relation) The substrate 1 is illuminated by the illuminating means 102, and scattered light or diffracted light from a foreign substance, defect or pattern on the surface is taken in.
At 06, an optical filtering process is performed, and the light is detected by the detector 107 in the detection optical system 103. The detected signal is amplified by an operational amplifier 201 in the detection head 101 to a signal having a large impedance and a low noise.
The signal is converted into a digital signal by the A / D converter 202 and transmitted to the operator processing system 203. The rotation direction of the substrate 1 is measured by the rotation detection unit 108, and the rotation direction of the substrate 1 is adjusted in advance by the rotation adjustment mechanism 105 controlled by the rotation control unit 213. The detection optical system 103 is
Since the substrate 1 has a sufficiently large depth of focus, the automatic focusing is basically unnecessary if the substrate 1 is transported by a transport system (not shown) with mechanical precision. Specifically, about 800 nm
And a numerical aperture of 0.08, the depth of focus is about ± 1.
00 microns. Of course, there is no problem even if an automatic focusing mechanism is provided.

The pitch detecting means 212 measures the repetition pitch of the pattern on the substrate 1 and the pitch of the chip from the detection signal. On the basis of the information such as the chip repetition pitch transmitted by the parameter transmission means 209 in the operator processing system 203, the pattern information is removed using the repetition of the chip pitch. The result is stored in the foreign matter data memory 206 and the pattern memory 208, and further, has no repeatability between chips based on information such as the repeat pitch of the position coordinate chips of the test element group transmitted by the parameter transmitting means 209. The pattern information such as the test element group is removed by the software processing system 210 and stored in the foreign substance memory 211. Here, coordinate data is created by the coordinate data creating means 232, and is stored together with the foreign substance information as necessary. The above processing is managed by the microcomputer 229 and displayed by the display unit 230.

Also, as shown in FIG.
Then, the light from the semiconductor laser 112 is collimated as a plane wave by the collimator lens 113, the concave lens 114, and the receiver lens 115, and illuminates the substrate 1 through the cylindrical lens 116 and the mirror 118. Here, the illumination is collimated by the cylindrical lens 116 only in the x direction, and is focused on the substrate in the y direction, as shown in the figure. In the detection optical system 103, the light beam Fourier-transformed by the Fourier transform lens 110 is subjected to an optical filtering process by the spatial filter unit 106, and an image on the substrate is formed on the detector 107 by the Fourier transform lens 111. You.

As shown in FIG. 3, in the spatial filter unit 106, while being guided by the guide 125, the screw 12 having the right-hand thread 127 and the left-hand thread 128
Coil spring supports 119 and 120 which are moved by the rotation of No. 6
Thereby, the pitch between the black linear spatial filters 141 passed between the coils of the coil springs 121 and 122 is changed. Power is supplied from a motor 140 via worm gears 129 and 130.

The tilt of the substrate 1 in the direction of rotation of the substrate 1 with respect to the detection head 101 of the substrate is measured by the rotation detecting detectors 123 and 124 installed on the spatial filter unit 106. FIG. 3 is described so that the detection light is incident from the upper two directions for the sake of easy viewing.

The operator processing system 203 shown in FIG.
Then, the pixels 2 × 2 around the detection signal are added to the four-pixel adding means 2.
14 and are averaged. This processing is intended for stable detection by averaging, but the detection performance (detection sensitivity)
Since the device itself falls slightly, bypass means is provided so that the device can be bypassed if necessary. The added signal is 8
The data is octalized by the value conversion means 215 and stored in the buffer memory 217 as two-dimensional image data through the cutout means 204 including a plurality of line memories 216. After being stored, necessary data is cut out from the two-dimensional image data by the judgment pixel cutout means 218 and the operator cutout means 219 and 231, and the comparison circuit 21
9 Here, the detector 107 uses a one-dimensional linear sensor so that high-speed detection by high-speed stage scanning is possible. The line memory 216 and the buffer memory 217 convert the data from the detector 107 into a two-dimensional image. Each time a signal from the detector 107 is sent one pixel at a time, the entire image moves one pixel in the x direction. I do. This is so-called pipeline processing. A comparison circuit group including a plurality of foreign matter comparison circuits 220, a threshold setting circuit 221,
The foreign matter signal is extracted by the OR circuit 224 and the AND circuit 226 according to the logic described later.

In the pitch detecting means 212 shown in FIG. 5, the FFT circuit 242 performs a Fourier transform process on the detected image, and from this result, the operator pitch calculating means 24
1, the spatial pitch is calculated by the filter pitch calculating means 244, and the spatial filter control system 243 and the operator cutout means 2 are calculated.
19, 231.

In the rotation adjusting mechanism 105 shown in FIG. 6, the rotation guide 151 is used as a guide to rotate the rotation bar 152.
The detection optical system head 101 provided with the
08 based on the information from the piezo element controller 15
The rotation is controlled by the expansion and contraction of the piezo element 154 controlled by 5. The spring 153 is provided so that the piezo element 154 and the rotary bar 152 are in contact with each other.
Even if the piezo element 154 and the rotary bar 152 are directly fixed without the 53, there is no problem. With this configuration, the gantry 1
The rotation of the detection optical system head 101 is controlled with respect to the rotation guide 151 installed on 56. Here, a drive system using a piezo element has been shown, but the drive system is not necessarily a piezo element. Alternatively, any other rotary or linear drive mechanism using ultrasonic waves may be used. Here, the piezo element is used because the piezo element 154 is small and has high-precision driving performance.

(Principle) The principle of a parameter compression type spatial filter (PRES (Parameter Reduction Spatial) filter) of the present invention will be described.

Conventionally, there has been disclosed a technique for detecting non-repeatable foreign substances or defects by using the repeatability of the pattern on the wafer surface. However, the shape of the diffraction pattern differs depending on the repetition period and the shape of the basic pattern even though the pattern has repetition. Therefore, it is necessary to change the shape of the spatial filter, which is the light shielding plate, according to the shape of the target repeated pattern. As a method of changing the spatial filter, a method using a photographic plate is disclosed. With these methods, it took time to create a spatial filter tailored to the target, or a large-scale device was required.

Specifically, when illuminated with coherent light, that is, a plane wave, from an oblique direction as shown in FIG. 7A, for example, a diffraction pattern as shown in FIGS. Assume that it was observed. In this case, when the pitch of the pattern on the substrate changes, not only the pitches px and py of the diffraction pattern but also the entire phase φ changes. Further, when the basic shape of the pattern on the substrate changes, the arrangement of the point patterns forming the diffraction pattern changes. That is, there are many parameters describing the Fourier transform plane diffraction pattern, and it is difficult to correspond to the pattern shape.

Here, instead of the plane waves as shown in FIGS. 8A and 8B, x waves as shown in FIGS. 8D and 8C are used.
It is assumed that the aperture is narrowed down on the sample 1 in the direction and the coherent or plane wave is illuminated in the y direction. In this case, on the Fourier transform plane, an image is not formed in the u-axis direction but becomes a diffraction pattern having a shape compressed in the u-axis direction. As a result, the spatial filter 106 is compressed to a one-dimensional parameter only in the v-axis direction. Here, the pitch p in the v-axis direction of the compressed diffraction pattern is a pitch corresponding to the pitch in the y-axis direction of the area illuminated on the substrate surface. In addition, one one
The thickness w of the diffraction pattern on the book line is determined by the numerical aperture sinβ of the front Fourier transform lens 110 to the Fourier transform plane. Specifically, it is determined by the exit-side numerical aperture of the illumination system 102 and the numerical aperture of the front-side Fourier transform lens 110. Therefore, the illumination system 102 and the Fourier transform lens 11
This is determined when 0 is determined, and is not affected by the pattern on the substrate 1 to be inspected. However,
In some cases, the numerical aperture of the illumination is changed, and in some cases, the width of the linear spatial filter 106 is preferably variable.

In practice, a one-dimensional image sensor capable of continuously scanning a stage (not shown) is suitable as the detector 107 in order to realize a high-speed inspection. When this one-dimensional image sensor is used, the shape of the one-dimensional sensor, that is, linear illumination on the surface of the sample 1 is suitable for improving the efficiency of illumination. In order to realize such illumination, it is necessary to narrow down at least one direction. That is, one-way coherent illumination has a great effect also for improving the efficiency of illumination intensity.

As described above, conventionally, the shapes of patterns on a substrate vary widely, and a spatial filter corresponding to each pattern is required in order to correspond to the various patterns. . However, according to the present invention,
From a different point of view, these various spatial filters can be considered as functions of only the pitch p, which means that the spatial filter having multidimensional parameters is compressed to one dimension. As described above, by compressing the dimensions of the parameters of the spatial filter, it is possible to simplify the spatial filter which has been difficult to cope with a shape change due to a complicated shape, and to be able to cope with all repetitive patterns.

The above configuration can be applied not only to detection of foreign matter or defects on a wafer or a liquid crystal display element, but also to any inspection object in which a non-repeatable portion is to be detected from a repeatable pattern. is there. Specifically, the present invention can be applied to a semiconductor mask, a reticle, a micromachining component using a semiconductor process, other micromachining components, a printed circuit board, and the like. The present invention realizes high-speed inspection by implementing high-illumination illumination while applying spatial filtering technology without replacing a spatial filter for each object when inspecting these objects. .

(Spatial Filter Control and Operator Pitch Control) Based on FIGS. 9A and 9B, (A) a pattern erasing method by the spatial filter 106 and (B) a pattern erasing by the shot comparison operator in the operator pitch processing system 203 Method and software processing system 210 (206-
The TEG pattern erasing method in (C) software in (211) will be described. In the present invention, the repeatability of cells having a pitch of several hundreds of microns or less is erased using the spatial filter 106, and the operator processing system 203 (21
7B) The pattern comparison is repeated by using the shot comparison operator at a pitch of several hundred microns or more by using the repetition between adjacent chips (in some cases, between shots meaning one exposure), and the repetition is further performed. Chip (TEG pattern) without
In (206 to 211) and the like, the data is erased so as not to be inspected using the coordinate / matrix data. Here, there are parameters required for each erasure. A spatial filter pitch is required for erasing by the spatial filter 106, an inter-chip pitch is required for erasing by inter-chip repetition, and chip position information is required for erasing a chip having no repetition (TEG pattern). . Therefore, the detection head 10 of the present invention
It is desirable that 1 can simultaneously detect at least two chips.
That is, the visual field size of the detection optical system 103 of the detection head 101 needs to be at least two chips or more. Of course, it is only desirable to have this visual field size. The positional relationship between the plurality of installed detection heads 101 is accurately known, and this positional relationship is stored in the parameter transmission unit 209, and the operator processing system When the comparison process is performed between the plurality of detection heads 101 at 203 or the like, the visual field size does not need to be two chips or more. However, considering the required accuracy of the optical system (detection head) 101 and the complexity of the circuit system for data processing in the operator processing system 203 and the software processing system 210, the field of view has a size of two or more chips. It is desirable. Also, here, the description has been made on the assumption that two or more chips are used. However, when two or more chips are written on a reticle used as a mask when a pattern is transferred onto the wafer 1 by a stepper, a gap between these chips is provided. In many cases, a pattern called a test element group (TEG) is written. To erase these patterns, instead of using the pitch between chips, a shot ( The pitch between the pattern printed in one exposure and the pattern on the reticle (stored in parameter transmission means 209).
Must be used. Of course, this method is not always necessary, and the TE formed in one shot is not necessary.
There is no problem if the G pattern is erased in a later process.

These pieces of information are measured in advance in association with the substrate 1 and stored in the parameter transmitting means 209.
From the stored information, a parameter corresponding to the substrate is selected and fed back to the foreign matter defect inspection apparatus (the operator processing system 203 via the software processing system 210 and the parameter transmission unit 209) of the present invention. Therefore, it is necessary to identify the substrate when using this method. For the purpose of this identification, a number or symbol corresponding to the substrate is described on the substrate. Read this symbol before the inspection, and print the product number, lot number,
Knowing the type, knowing the process from the data at the place where the foreign substance inspection device of the present invention is installed,
, The pitch of the spatial filter 106 and a threshold value may be set by setting the pitch detection means 212 via the.

In implementing the foreign matter defect method of the present invention, it is not always necessary to acquire parameters as described above and send them to the apparatus of the present invention as described above. Rather, as described below, it may be more desirable to obtain the data independently by the apparatus of the present invention. In the above method, it is necessary to know the value of the parameter to be input in advance, but if it is obtained independently, such trouble is not required. Of course, it is not necessary to read the number written on the substrate.

In the present invention, as described above, a three-stage pattern removal is performed to extract and detect a foreign substance or a defect by distinguishing the foreign substance or the defect attached to the substrate having a complicated background pattern from the background pattern. Has a function. In the pattern removing function, a portion that is actually determined to be a pattern is discarded without being an inspection target.
More specifically, the repetition with a pitch of several hundred microns or less is eliminated by the spatial filter 106 and the parameter transmission
Are repeated using the repeatability between chips in the operator processing system 203 based on the parameters given through The data is erased based on the matrix data so as not to be inspected by the software processing system 210 or the like.

The reason why the area where the pattern is formed is excluded from the inspection target is as follows.
Even if a pattern is formed, adjacent chips have the same shape and are formed with patterns that emit the same amount of light in the same emission direction. Therefore, if the detected light intensities of the light from the two patterns are compared, the spatial filter 106
Inspection of foreign matter or defects should be possible even in a region where a pattern of a shape that cannot be erased by the above is formed. However, the intensity of the detection light tends to be unstable in these patterns particularly when scattered light is detected. If the above-described pattern removal by comparison is performed, a false report (pattern information that is not a foreign substance is detected as a foreign substance). )
But more. Therefore, it may be more effective to exclude the area where the pattern is formed from the inspection target. In other words, considering stability, especially when detecting scattered light, a region where a complicated pattern is formed is excluded from the inspection target, or the detected light intensity of light from an adjacent chip pattern is compared. It should be decided whether to inspect for foreign substances.

(Parameter Acquisition Method) Hereinafter, a specific parameter acquisition method will be described with reference to FIG. The spatial filter control system 243 changes the pitch of the spatial filter 106 from the maximum position to the minimum position at the next point where the wafer 1 is transported to a position where the detection optical system 101 can capture the repetitive pattern of the wafer 1. At this time, the all-pixel adding circuit 245 converts the signal captured by the one-dimensional detector 107 into a signal.
All the values of each pixel are added, and the pitch at the position where the added value is the minimum with respect to the change in pitch is determined by the pitch calculation circuit 24.
6 is selected. This value is sent to the spatial filter driving mechanism 106 ', and the spatial filter 106 is set to a predetermined pitch.

In selecting the pitch of the spatial filter 106, the spatial filter 106
Can be calculated even if the frequency analysis shown in FIG. The signal detected by the detector 107 at the next point where the detection optical system 101 can take in the repetitive pattern of the wafer 1 at the next point where the wafer 1 has been conveyed is subjected to frequency analysis by the FFT circuit 242. The calculating means 244 selects the pitch of the spatial filter so that the spatial frequency becomes a peak in the frequency domain. This value is the spatial filter control system 243
The spatial filter 106 is sent to the spatial filter driving mechanism 106 ′ via the. And the spatial filter 106 is set at a predetermined pitch. In this frequency analysis, fast Fourier transform is most desirable in terms of processing speed and the like, but it is not necessarily fast Fourier transform, and other methods such as Hadamard transform, frequency analysis by integration, method by autocorrelation function operation, and the like are problematic. Absent. In the method based on the frequency analysis, not only the pitch of the spatial filter but also the chip pitch (operator pitch) for the method for removing components that cannot be removed by the spatial filter is calculated by the operator pitch calculator 2.
41, the arithmetic processing is performed simultaneously. It is desirable to use the largest pitch between chips calculated by frequency analysis and smaller than 1/2 of the field of view of the detection optical system. This is because the largest one often corresponds to the pitch between shots described below.

After the parameter values for the inspection are set as described above, the inspection is performed. The above-described method has a problem that the first portion of the wafer 1 being transferred cannot be inspected for foreign substances. On the other hand, there is an effect that the inspection device can be made independent of other signal transmission systems.

(Telecentric Optical System) In the present invention,
As described above, the repetition of several hundred micron pitch or less is erased by the spatial filter 106, the repetition of several hundred micron pitch or more is erased by using the repetition between chips, and the chip having no repetition is not inspected. The data is erased as follows. Here, in order to erase the detection signal of the pattern of the chip by using the repeatability between the chips, the detection signal between the chips is compared, and when the difference is larger than a certain value, the detection signal is detected as a foreign substance. . That is, it is assumed that the scattered light or the diffracted light from the patterns of adjacent chips does not have the same intensity. Therefore, it is necessary to stably detect light from a corresponding position of an adjacent chip. However, since the diffracted light from the pattern has directivity, if the field of view is wide and the angle of view of the lens from each position within the field of view is greatly different, the light intensity differs depending on the position within the field of view due to this directivity.

Here, the telecentric optical system 103 '
Is a technique developed by making the principal rays from each point on the object 1 parallel to each other so that the magnification of imaging does not change even when the focal position is shifted. By using the telecentric optical system 103 'in the present invention, the intensity of the detection light from each point of the object is stabilized by taking measures against the change in the intensity of the detection light due to the directivity of the scattered light or the diffracted light from the foreign matter. And keep it constant. According to the present invention, all points of the object 1 can be illuminated from the same direction and all points can be detected from the same direction. Therefore, even if the diffracted light or scattered light from the pattern has directivity, the pattern shape is the same. This is because the intensity of the detection light becomes the same.

The reason why the magnification is changed is to change the pixel size. If you increase the pixel size,
The inspection speed can be increased as a result of an increase in the area to be detected as one signal, but the resolution of the detection system is reduced, making it difficult to detect small foreign matter or a defect.
Conversely, reducing the pixel size will increase the resolution and allow you to inspect smaller defects or foreign matter,
The inspection time is long. Of course, in this case, it is necessary to increase the resolution of the optical system.

A lens exchange mechanism will be described with reference to FIG. In the present embodiment, as described above, the one-to-one
The telecentric optical system 103 'having an imaging magnification of? It is important to be telecentric in order to sufficiently obtain the effects of the present invention, and it is not necessary to have a 1: 1 magnification as shown in FIG. Therefore, an optical system of another magnification can be used. In realizing an optical system of another magnification, as shown in FIG.
On the other hand, the magnification can be changed by exchanging the Fourier transform lens 161 (specifically, the lens on the object side is optimal). With such a configuration, the image-side lens 1
11 and the detector 107 need not be replaced,
As a result, optical systems having different magnifications can be supplied at low cost.

As described above, the telecentric optical system 1
03 ′, or an optical system in which a principal ray emitted from each point on the substrate 1 passes through the center of the pupil of the detection optical system 103 (the surface on which the spatial filter 106 is disposed).
A large effect can be expected when used in a foreign matter defect inspection apparatus using the method described above. This produces the effect of being detectable. This is particularly effective when an optical system having a large field of view is used.

(Basic Concept of PRES Filter) As described above, the PRES filter according to the present invention can exert the maximum effect when used together with a telecentric detection lens and a coherent illumination system with only one axis. When realizing a defect inspection apparatus for foreign matter or the like using the spatial filter 106, which is the original purpose, it is not always necessary to use them together. The illumination is made coherent on one side only because coherence is required when using a spatial filter and one side is sufficient. Furthermore, since one side is not coherent, the illumination light beam can be narrowed down on the object, and the illumination intensity can be increased. Conversely, if sufficient illumination light intensity can be obtained, the illumination may be coherent on both sides in the x and y directions. In other words, the essence of the present invention is that spatial filtering can be performed by adjusting the rotation direction of the substrate (wafer) 1 even if the spatial filter 106 is one-dimensionally compressed.

What is more important here is the illumination means 102.
In the case of illuminating obliquely, in order to make the parameter of the spatial filter one, it is necessary to make the linear spatial filter 106 parallel to the incident surface of the illumination, and it is not to use illumination that makes only one side coherent. That is, it is essential that the incident surface of the illumination, the longitudinal direction of the linear spatial filter, and the repetition direction of the pattern on the substrate (wafer) are matched. When it is not necessary to use one parameter, there is no need to make the linear spatial filter parallel to the plane of incidence of the illumination. By using the linear spatial filter, the pitch and the phase or the pitch and the rotation direction can be matched. A spatial filter that can correspond to all patterns can be configured. Furthermore, with respect to illumination from above, there is also an effect that only the repetition direction of the spatial filter and the pattern on the substrate (wafer) needs to be matched because the direction of incidence of the illumination always coincides with the direction of the linear spatial filter.

However, in any of the above cases, a linear beam is used for illumination,
If a dimensional sensor is used, it is necessary to adjust this direction. However, the alignment in this case is to obtain uniformity of illumination, or to eliminate repetition of a large cycle using inter-chip repetition,
This is not necessary for erasing pattern information having a smaller period in the spatial filter.

Optical Telecentric Detection Optical System 10
3 'also shows both telecentric optics here,
It is not always necessary to be both telecentric, and it is sufficient if at least the object side is telecentric. In addition, even if it is not telecentric, the illumination system 1 at each point on the substrate
It is sufficient that the 02 principal ray, that is, the 0th-order diffracted light from each point on the substrate passes through the center of the pupil plane (the plane on which the spatial filter is installed) of the detection optical system. Even with such a configuration, it is possible to avoid fluctuations in the pattern output due to the distribution of the zero-order diffracted light from the pattern at each point. However, the performance of this method is somewhat lower than that of the telecentric optics described above, since the direction of incidence of the illumination on the pattern will be different. However, for some subjects, this method may be sufficient.

Further, the principal ray of the illumination system at each point on the substrate,
That is, the present invention can be applied to the case where the zero-order diffracted light from each point on the object does not pass through the center of the pupil (the surface on which the spatial filter is installed) of the detection optical system, that is, even if a normal high-field lens is used. It is possible to realize a defect inspection apparatus such as a foreign substance using a spatial filter, which is the original object of the present invention.

(Method of Preliminary Measurement of θ) In the above inspection apparatus, it is necessary to adjust the inspection apparatus to the angle of the wafer or the substrate. Specifically, it is necessary to set the optical axes of the detector and the illumination perpendicular or parallel to the repetition direction of the pattern formed on the substrate. To achieve this, the angle at which the substrate is transported is accurately detected by an angle detection mechanism, and based on the result, the entire detection optical system is rotated around the normal to the surface of the substrate as an axis, and the direction of the pattern is detected. Match the direction of the vessel.

FIG. 12 shows a specific configuration. FIG.
Shows the arrangement of the spatial filter unit 106 and the rotation direction detectors 123 and 124 formed on the Fourier transform plane, and is a view of the detection optical system 103 as viewed from the substrate side. An aperture 142 for limiting is also shown. The detection optical system 103 has a slightly larger numerical aperture, and the diffraction pattern from the substrate 1 whose parameters have been compressed protrudes outside the stop 142 and is detected by the detectors 123 and 124. Therefore, if the detectors 123 and 124 detect the zero-order diffracted light in the diffraction pattern and detect a change in the peak position, the substrate 1
, The rotation direction of the detection optical system head 101 is measured. Specifically, the distance between the two detectors 123 and 124 is Lp, and the distance from the center of the detectors 123 and 124 is detected.
The distances between the peak positions of the diffracted light are hp1 and hp2. Since the diffracted light when the rotational position is shifted has a shape as shown in FIG. 12B on the Fourier transform plane, the rotational angle θp is substantially expressed by the following equation (1). FIG. 12 (b) shows a spherical surface including the Fourier transform surface when viewed from the direction of the Fourier transform surface.
Indicates a ray on the spherical surface and the pupil surface 142, and point 5 indicates an intersection of the zero-order diffracted light of the illumination light, that is, the reflected light and the spherical surface.

[0082]

Sin θp = hp1 / Lp (Equation 1) Strictly speaking, since the pitch Ldp of the diffracted light is an unknown number, the position between the peak positions of the diffracted light at the position rotated by a known minute rotation angle θk is strictly speaking. Re-measure the distance, hp11,
If simultaneous equations are established as hp21, Lpd and
θp can be calculated. Also, as another method, hp1, h
There is also a method of adjusting while rotating θ so that both p2 become 0. Here, when detecting the direction of the detection optical system 103 and rotating the detection optical system 103 in a cheap direction, the purpose is to erase the pattern formed on the substrate 1 by a spatial filter, so that the entire optical system 101 is necessarily rotated. It is not necessary to rotate the spatial filter 106. Also,
Regarding the rotation of the optical system 101, there is no problem if several units are simultaneously rotated or each unit is rotated.

What is important in this configuration is that the detection optical system 101 is movable to adjust the rotation with the substrate without rotating the stage supporting the wafer or the substrate 1.
Here, the foreign matter detection device according to the present invention can be realized because the foreign matter detection device has a configuration capable of detecting the foreign matter even if it is not completely perpendicular to the flow direction of the substrate (wafer) 1. In addition, the rotation of the optical system 101 corresponds to the rotation of the substrate 1, and the scanning of the optical system 101 with respect to the substrate 1 is performed by using a transport system (not shown) of the substrate 1 to reduce the two degrees of freedom. What is important is that each mechanism is simplified by having two mechanisms independently.

The angle detecting mechanism 108 may calculate the direction of the pattern formed on the substrate 1 from the image captured by the detector 107 without using the method as shown in FIG. In this case, it is difficult to measure in real time, but there is an effect that mechanisms such as the detectors 123 and 124 become unnecessary. Further, as shown in FIG.
01, instead of being mounted inside
A configuration in which the measurement is cleared in advance may be used. Also,
The detection result is transmitted to the detection head 101 by the rotation mechanism 181.
Alternatively, the detection head array 180 is rotated to adjust the rotational deviation. In such a configuration, since measurement has been completed in advance, there is an effect that inspection can be performed on the entire substrate. In this case, the rotation detecting head 162 detects the diffracted light, as described above, forms an image of the substrate 1 and processes the image, or uses another sensor such as a wafer orientation flat. The edge of the substrate 1 may be detected.

As described above, in the embodiment of the present invention, the angle detection and the substrate transfer are described as cases where the transfer is performed in a straight line. However, the present invention is not necessarily limited to this.
It is also applicable to a rotating arm or the like. However, in the case of such an arm transfer arm, it is desirable that the length direction of the detection area of the detection optical system 101 is set so as to pass through the center axis of the rotating arm. With such a configuration, the present invention can be applied to a transport mechanism of a rotary system without any consciousness.

(Rotation Control of Optical System) In the present invention, since the spatial filter 106 is used, the substrate 1 and the optical system 10
One rotation direction alignment is required. This rotation adjustment is desirably performed in advance for setting of other parameters. According to the present invention, as shown in FIG. 3, FIG. 12 or FIG.
4 or 162. These optical systems 123 and 12
Reference numeral 4 or 162 detects the rotational deviation of the wafer at the first portion of the substrate 1 being transferred, sends the detection result to the optical system rotating mechanism 181, and based on this information, the optical system 101 or 18
Rotate 0. FIG. 6 shows the rotation adjusting mechanism 105.

Further, instead of rotating the optical system 101 or 180 based on the detection result of the rotational deviation, the rotational deviation can be corrected by electric processing. As shown in FIG. 4, the rectangular operator cut out in the buffer memory 217 is shifted in the θ direction according to the detected rotational deviation of the wafer, thereby mechanically correcting the rotational deviation of the substrate (wafer) 1. The effect is as follows. This method has an effect that the time required for correction can be reduced because the optical system 101 or 180 does not need to be moved.
Also, by using this circuit, the operators 219 and 231 are always moved in the θ direction without measuring the rotational deviation θ, and under the condition that the detected foreign matter is minimized (this condition is the substrate 1 and the detection head 101). (In which there is no rotational displacement). Needless to say, this method has a high-speed signal processing system. Alternatively, even if a high-speed signal processing system is not used, the above-described conditions are set in advance by the above-described method, and then the inspection is performed, so that the inspection is performed as if the rotational displacement of the substrate (wafer) 1 was mechanically corrected. Can be implemented.

Of course, the method disclosed here is not always necessary, and for example, a certain tolerance for the rotational displacement of the substrate (wafer) 1 in the transport direction so that the substrate (wafer) 1 is carried into the reduction projection exposure apparatus. When transported after mechanical adjustment within the range, the detection control system is not necessary as described above.

(Log Scale Threshold) FIG. 14 shows a comparison test using an optical processing method such as a spatial filter as preprocessing, and a comparison test using only electric signals without using such processing. The appearance of the detection signal at this time is schematically shown in FIG. The method using the spatial filter 106 can remove only the information on the pattern without losing the information on the defect in the pattern portion.
Since the signal is an electric signal, a foreign object defect signal can be detected only within the dynamic range at the time of photoelectric conversion. That is, when the pattern signal is extremely large and the foreign matter defect signal is extremely small, the foreign matter defect signal is buried in the pattern signal, and it is difficult to detect the foreign matter defect signal separately from the pattern signal.

FIG. 14 shows the detection position on the horizontal axis and the detected signal intensity on the vertical axis. On the left side, a signal 18 containing foreign matter or foreign matter defect information 4 is shown, and on the right side, a signal 19 not containing foreign matter defect information to be compared. Here, when the pixel size detected as one signal is detected as 13,
Detection light corresponding to the hatched areas 16 and 17 is detected. In this case, since the foreign matter defect information 4 is smaller than the total area, the comparison between the two detection signals 16 and 17 cannot be performed stably. Specifically, it is buried in noise. In this case, even if the light intensity of the illumination is increased, a detector having a large dynamic range is required to be able to detect the foreign matter defect information 4. Here, the pixel size is changed from 13 to 14.
Then, the detection signals become 7 and 8, and the foreign matter defect information 4 can be detected by comparison. Reducing the pixel size produces such an effect. Conversely, if the detection signals 18 and 19 can be stably eliminated (by essentially comparing them without changing them into electrical signals, etc.), the detection signals can be detected at, for example, 10 or more positions. If possible, the detection signals will be 5 and 6, which are levels that can be compared and inspected. In this case, since the pixel size remains at 13 as described above, high-speed detection with large pixels becomes possible. If the light intensity of the illumination is increased, the defect information 4 such as a foreign substance can be detected even with a detector having a small dynamic range. The method using the spatial filter 106 of the present invention is to detect minute foreign matter defect information 4 while using the pixels 13 described above.

As described above, even if the pattern signal in the portion between adjacent chips to be compared can be extremely stabilized by devising the optical system or the like, the light intensity can be reduced to 1: 100 or 1: 1.
The limit is to detect with a dynamic range of about 000. Therefore, when the foreign object defect signal is small or the pattern signal is large so as to require a further dynamic range, the signal intensity between the adjacent chips is compared, and the foreign object defect signal is included in one of the signals. It cannot be determined whether it is included or not. Only when the ratio of the foreign matter defect signal to the pattern signal is sufficiently large, the presence or absence of foreign matter can be inspected by comparison. When this ratio is small, a foreign object is missed, or a false threshold increases when the threshold value is reduced to inspect the foreign object.

Therefore, it is difficult to inspect foreign substances existing on these patterns without false reports. There is no choice but to eliminate false reports or to reduce foreign substance detection sensitivity so that only large foreign substances can be detected. The purpose of using the embodiment in which the area where the pattern is formed is excluded from the inspection target in the present invention is to eliminate the false information described above.

Further, in order to reduce the foreign substance detection sensitivity so that only large foreign substances can be detected, a log scale comparison inspection as described below is required. Certainly, even if adjacent chips are formed with patterns having the same shape and emitting the same amount of light in the same emission direction, the detected light from these two patterns is not completely the same. Therefore,
The detected light intensities of the light from these two patterns are highly likely to vary and are difficult to compare. Therefore, at the time of comparison, a cut out by the operators 219 and 231 and p cut out by the operator 218 are compared with a plurality of foreign matter comparing circuits 220.
And a plurality of noise comparison circuits 22,
When the expression (2) is satisfied, the two signals are different, and it can be determined that a foreign substance exists.

[0094]

(Ap)> δ (equation 2) However, in this method, when the absolute level of the signal is large, if there is a variation that fluctuates at a ratio to the absolute amount, there is no foreign matter but there is a foreign matter The possibility of so-called false alarms to be judged increases. Therefore, when the ratio of the two signals satisfies the expression (3), it is determined that the signal is a foreign substance.

[0095]

(A / p)> δ (Equation 3) However, in practice, the division of two signals increases the scale of the arithmetic circuit.
When the expression (4) is satisfied, it is determined that a foreign substance exists.

[0096]

Log (a / p) = loga-logp> δ (Equation 4) As described above, by using the expression (4), if the quantization threshold is set using the logarithmic axis, The operation that should originally be divided can be processed in the comparison circuit 220 by subtraction. The circuit configuration shown in FIG. 4 implements the logic described above. For the logarithmic processing as described above, the threshold value of the octalization processing system 215 shown in FIG.

In FIG. 4, a comparison circuit 220 using logarithmic octalization processing in the octalization processing system 215 described above.
Although the subtraction processing is shown in the above, this method is not necessarily used, and even if the method of the above division processing is used,
It is possible to use multi-value conversion other than octal conversion. In this case, if ternarization is used, foreign substances on all patterns are discarded without being detected, and if further multi-level conversion is used, if the optical system is stable, detection of smaller foreign substances on the pattern is performed. Becomes possible.

Here, the above-mentioned δ is set in the threshold value setting circuit 221, and the comparison based on the above (Equation 2), (Equation 3) or (Equation 4) is made in the plural foreign matter comparison circuits 220, and the operator ( (Pixel) 218 and operator (pixel)
The OR circuit 224 performs an OR operation on the difference between 219 and 231 or more than the threshold value δ, and outputs “0” from all the foreign matter comparison circuits 220 to erase the repeated pattern between chips (shots). Foreign matter comparison circuit 22
Anything from 0 to the threshold δ is output as a foreign substance as “1”,
It is stored in the foreign object data memory 206. In the plurality of foreign matter comparison circuits 220, the above (Equation 2) or (Equation 3)
Alternatively, a comparison based on equation (4) is performed, and the operator (pixel) 218 and the operators (pixel) 219 and 23 are compared.
When all the differences from 1 are equal to or smaller than the threshold value δ, “0” is output from all the foreign matter comparison circuits 220, AND is obtained by the AND circuit 226, detected as a repetitive pattern between chips (shots), and stored in the pattern memory 208. It is memorized.

FIG. 30 shows the state of quantization using the logarithmic threshold described above. The horizontal axis indicates the detection position, and the vertical axis indicates the detection signal. Logarithmic threshold values 50, 51, 52, 53, and 54 are set, and signals located at portions separated by a pitch p are compared. Here, by multi-valued, for example, by betting one allowable range that is determined to be the same at the time of comparison, not only the pattern signals 55 and 58 quantized to the same value but also to one different value. Pattern signals 56, 59, and 57, 60 are determined to be patterns and are not false alarms.
In other words, by increasing the ratio of the logarithmic threshold value during quantization, the allowable range is increased and false alarms can be reduced, but only larger foreign substances can be detected on the pattern. Further, both the foreign matter signals 61 and 62 can be detected. Further, by spreading the operator 231 in the plane direction, a quantization error in the plane direction can be allowed.

As described above, the pattern erasing by the spatial filter 106 and the pattern erasing by the chip comparison include:
There are essential differences. In other words, the method using the spatial filter can detect defects in the pattern part by emphasizing them.However, the method using the chip comparison has a large dynamic range because the foreign matter defect information in the pattern part is photoelectrically converted and detected as it is and then compared. It is a necessary point. Just, the method using the spatial filter has a form in which only defects are emphasized by cell comparison using interference.

(Quantization Error in Planar Direction and Quantization Error in Depth Direction) Here, the method using the repetition between the chips used is basically a comparative inspection. The following configuration is used in order to stably realize such a comparative inspection by scattered light detection using a laser light source.
An operator who realizes a pattern removal method using repetition between chips is formed of a plurality of pixels in both the x and y directions in the plane direction. The comparison of whether the signals should be determined to be at the same level or whether the signals should be determined to have different signal levels due to the presence of a defect or foreign matter in (Equation 2), (Equation 3), Equation (4) is used. By expanding the sampling of the comparison numerical values in the plane direction and the light intensity direction in these comparisons, it is possible to stably distinguish the foreign matter from the pattern.

(Filter Size) The design concept of the spatial filter 106 will be described below. In the foreign matter and defect inspection apparatus according to the present invention, pattern information of a pattern having a large pitch is erased using inter-chip repetition by an operator. Therefore, the "spatial frequency erasable by the spatial filter" needs to be larger than the "spatial frequency erasable by the operator". Hereinafter, the reason will be described.

Assume that patterns of pattern pitches L1 and L2 (L1 <L2) are formed on substrate 1 as shown in FIG. With this pattern, the illumination light becomes θ
1. The light is diffracted in the direction of θ2 to form a diffraction pattern on the spatial filter 106 with pitches pf1, pf2 (pf1> pf2). Therefore, the following equation is established.

[0104]

Df / (2 · N.A.) = Pf1 / sinθ1 (Equation 5)

[0105]

Where Df is the diameter of the optical systems 110 and 111 on the pupil plane, and N.f. A. (NumericalAparture) is the numerical aperture of the optical system 110.

Among the diffraction patterns, the diffraction pattern having the pitch pf1 is shielded from light by the spatial filter 106, but the diffraction pattern having the pitch pf2 is too small in pitch.
It is assumed that the light cannot be shielded by the spatial filter 106. That is,
The pattern on the substrate having the pattern pitch L1 is erased by the spatial filter and is not imaged on the detector 107, but the pattern with the pattern pitch L2 is imaged on the detector 107 without being shaded by the spatial filter and is not erased. (Hereinafter, “the pattern is erased” means that the light emitted from the pattern, that is, the light having the pattern information is shielded by the spatial filter 106 so that the detector 1
07 to prevent light from reaching, and not to form an image of the pattern on the detector. " )
From the equations (5) and (6), the following equation (7) holds.

[0107]

L1 = (Df.lambda.) / (2.pf1.NA) (Formula 7) Here, the spatial filter 106 has a certain size due to limitations in the numerical aperture of the illumination 102 and the like as manufacturing accuracy. You need more. Therefore, from equation (7), if pf1 is kept large, the pitch of the pattern that can be eliminated by the spatial filter 106 will be small.

Here, as shown in FIG. 16, a plurality of chips are formed on the substrate 1 at a pitch Lt. The information of the pattern of the pattern pitch L2 is erased by using the repetition between the chips. Specifically, a signal repeatedly detected at the chip pitch Lt is a pattern pitch L2
And the detection signal itself is deleted. (In this case, “erasing the pattern information” literally means “deleting or discarding the signal.”) Here, in order to avoid a quantization error in the plane direction described later, the explanation will be given first. Operators 219, 231
Has the size of nx pixels in the x direction and ny pixels in the y direction,
If a pattern exists in this operator, the determination pixel is determined to be a pattern. Therefore, the ratio α of the detection area according to the size of the operators 219 and 231 is represented by the following equation (Equation 8).

[0109]

Α = 1− (w · nx / L2) (Equation 8) On the other hand, from Equation (8),

[0110]

L 2 = 1− (w · nx / α) (Equation 9) Therefore, if the detection area ratio α is kept large, the pitch of the erasable pattern becomes large.

Here, in order to erase all the pitch patterns on the substrate, as shown in FIG. 18, the pattern pitch L1 which can be erased by the above spatial filter should be larger than the pattern pitch L2 which can be erased by the operator. . That is, as shown in FIG. 18, it is necessary that the erasure by the spatial filter and the erasure by the operator overlap. As a result, even if a repeated pattern having a certain pitch appears on the substrate, the pattern can be erased by the spatial filter 106 or the operator, and as a result, only the minute foreign matter can be detected.

Here, as shown in FIG. 17, the variable pitch PRES filter 141 has a minimum pitch pf1 = 1 m.
The pitch is continuously changed from m to a maximum pitch of 2 · pf1 = 2 mm. In order to prevent or suppress the interference phenomenon on the detector 107 due to the spatial filter 106 (141), the number of filters,
It is necessary to limit the pitch and filter width. That is,
The minimum spatial filter pitch described above is not only limited in terms of manufacturing, but also in order to suppress interference phenomena. At this time, the relational expression among the number of filters, the field size, and the operator size becomes important.

Whether the pattern can be erased by the spatial filter 106 (141) depends not on the pitch of the pattern to be erased but on the number of linear spatial filters 141, as described below. The size of the spatial filter surface is D, the minimum spatial filter pitch is pfs,
The number of linear spatial filters is nf, and the maximum pattern pitch that can be shielded by a spatial filter of pfs is Lm. Let X be the field of view of the detector, w be the pixel size, and N be the number of pixels. The operator size when comparing two chips by the operator is no.

[0114]

Pfs / D = λ / (N.A. · Lm) (Formula 10) In order to make the detection area ratio α sufficiently large,

[0115]

(11) k · no · w = Lm (k >> 1) (11)

[0116]

D / pfs = nf (Equation 12) is required.

If the pixel size is set near the resolution of the optical system, it is necessary and sufficient from the viewpoint of inspection time and detection performance.

[0118]

[Mathematical formula-see original document] w = k0. [Lambda] /N.A.

[0119]

(14) pfs = D / (k · k0 · no) (14)

[0120]

Α = 1− (1 / k) = 1 − ((no · k0) / nf) (Equation 15), and the detection area ratio α is determined only by the operator size and the number of linear filters. Assuming that k0 is about 1 and the operator size no is 3, in order to set the detection area ratio to, for example, 80% or more, it suffices to have 15 or more linear spatial filters. The above examination results show that regardless of the pitch of the pattern to be detected,
If the pixel size is made equal to the detector size and approximately 15 or more linear spatial filters are used, it means that patterns of all pitches can be handled.

(Production method) Hereinafter, the spatial filter 1
The production method of No. 06 will be described. Spatial filter 1 of the present invention
06 may have a shape in which a plurality of linear filters are provided on a spring-like support, and the method does not necessarily depend on the method described here. It can be manufactured in

FIG. 19 shows a manufacturing method. In the present invention, a filter sheet 165 in which linear filters are formed at equal intervals by etching is placed on the coil springs 121, 121, and each linear filter 141 is soldered or installed with an adhesive or the like. Here, the coil springs 121 and 122 and the linear filter 141 are desirably made of stainless steel from the viewpoint of strength. However, in the case of stainless steel, there is a drawback that solder does not easily adhere. Therefore, it is preferable that the surface of stainless steel is plated with chromium and gold is further plated to prevent oxidation of chromium. However, these platings are not indispensable and are merely intended to make solder easier to apply. Therefore, a flux that acts directly on the surface of the stainless steel to facilitate soldering can be used. In such cases, chromium and gold plating are not required.

The linear filter 141 in the filter sheet 165 has a size at the Fourier transform position of the image of the light source 112 irradiated onto the substrate 1 in the optical system 103 described above, or a margin of about 10% to 100%. It is desirable to set the size to have Here, it is desirable that the position where the filter sheet 165 comes into contact with the coil springs 121 and 122 is as thick as the coil springs. Therefore, it is necessary to change the thickness of the filter portion and the contact portion of the linear filter 141. In the portion where the thickness changes, it is necessary to smoothly change the thickness in order to avoid stress concentration.

The filter sheet 165 is mounted on the coil spring 1 previously installed at an equal interval with the filter sheet 165.
21 and 122, and an appropriate amount of solder paste 166 mixed with solder and flux is applied to the contacts. Here, the solder paste may be put on the filter sheet 165 in advance.

Here, in order to keep the position of the central linear filter unchanged even when the interval between the linear filters 141 is changed, that is, to make the configuration that always blocks the zero-order diffracted light, the filter is used. When the sheet 165 is placed, the center of the filter sheet 165 needs to come to the center of the coil spring movable mechanism 131. Therefore, the movable mechanism 131 is operated in advance, and the contact portions of the coil springs 121 and 122 with the filter sheet come to the central part of the movable mechanism 131 (as shown by a dashed line perpendicular to 24 and 25 in FIG. 19). If not, it is necessary to rotate the coil spring around the center axis 23 of the coil spring itself so that the center 25 of the filter sheet 165 comes to the center 24 of the coil spring moving mechanism 131 as shown in FIG.

The filter sheet 165 on which the solder paste 166 is placed is placed on the above-described coil spring movable mechanism 131, and the whole is heated to melt the solder paste 166, and is then naturally cooled to complete. The reason why the entire coil spring moving mechanism is heated is to prevent residual stress due to heating.

Since the spatial filter mechanism 106 of the present invention uses coil springs 121 and 122 having a small elastic coefficient, the strain increases with respect to the residual stress. Filtering performance is inferior. In order to remove the strain due to such residual stress, it is preferable to heat the entire movable mechanism. Also, this spatial filter
Since the light is used to shield the diffracted light, it is desirable that the light be processed to be black in order to eliminate irregular reflection and the like when the light is shielded. This blackening treatment may be one in which a paint is applied, or may involve a heat treatment called a blackening treatment. Further, this black processing may be performed on the filter sheet 165 only for the portion other than the joint portion on the filter sheet 165 and the light shielding portion, even after the soldering.

The method using the solder paste 166 has been described above. However, it is not always necessary to use a solder paste, and this is a method of soldering the linear filters in the filter sheet one by one. Needless to say, it may be.

(Image restoration by convolution) FIG.
As shown in FIG. 1, in the apparatus according to the present invention, the light beams are diffracted by the spatial filters 141 arranged at equal pitches to form a diffraction pattern on the image plane. Specifically, a diffraction image of a point image is formed on the imaging surface. In this image, ± 1st-order diffracted light appears around the 0th-order diffracted light. When such diffracted light appears, for example, a signal 26 having a ± 1st order peak around the original peak is formed, and not only foreign substances are detected by increasing to three, but also in the case of a pattern. In this case, the location where erasing or detection sensitivity is reduced as a pattern is increased. In order to avoid this, the width of the linear spatial filter may be reduced. Specifically, when the width of the linear spatial filter is に 対 し て of the pitch of the spatial filter 141, the first-order diffracted light is 倍 times the 0th-order diffracted light, whereas the linear diffracted light is linear with respect to the pitch of the spatial filter. The width of the shape space filter is 1 /
In the case of 8, the first-order diffracted light is reduced to 1/30 times the 0th-order diffracted light. These results are calculated by Fourier transforming the shape L (u, v) of the spatial filter 106. Therefore, it is necessary to select the width of the linear spatial filter 141 with respect to the pitch of the spatial filter as needed. Also, particularly when it is necessary to know the ratio small, the method of condensing the light source irradiated on the substrate 1 by the illumination system 102 at the position of the Fourier transform so that the diffraction pattern is sufficiently shielded by the spatial filter. Also need to be changed. Specifically, this is achieved by increasing the numerical aperture of the y-direction component of the light incident on the substrate 1 for illumination. At this time, the size of the image of the illumination on the Fourier transform plane is calculated by the image formation theory of the coherent light.

The above-described effect of diffraction can also be avoided by an image processing method known as a Wiener filter. Specifically, as shown in FIG. 21, assuming that the shape of the spatial filter is L (u, v), the value of 1 / L (u, v) is obtained in advance and its Fourier transform is calculated. May be convolved with the image detected by the convolution unit 251. By convolving the image detected by the convolution means 251 in this way, the diffracted light by the spatial filter 141 can be removed. Here, since the value of 1 / L (u, v) diverges to infinity, it is necessary to approximate this value to a sufficiently large value. The value of the complex number obtained as a result of the Fourier transform is approximated so that the value whose phase inverts substantially is positive or negative and the magnitude is the absolute value of the complex number. In addition, it is necessary to set the size of the image to be convoluted so as to have a minimum size that is sufficiently effective when cutting out the image.

In the above-described pattern removal using the spatial filter and the repetitive chip, the smaller the pitch of the pattern removed by the spatial filter is, the more the pattern is removed by the method using the repetitive chip. The larger the pitch of the pattern, the better. In particular, in order to reduce the area that is removed as a pattern, it is desirable that the pitch of the pattern removed by a method using a repetitive chip is large. Therefore, it is preferable to determine the rotation direction of the substrate during the transfer of the substrate according to the shape of the pattern formed on the substrate so as to reduce the area that is removed as a pattern in the inspection. Specifically, as shown in FIG. 22, when there are pattern pitches L11 and L12 that cannot be eliminated by the spatial filter, it is better to scan in the direction 28 so that the direction of the larger pattern pitch L11 is the direction of the sensor. Good. The inspection is performed by rotating the substrate 1 by 90 degrees in advance so as to be in such a direction.

(Illumination from Directly Above) Here, for example, in order to realize miniaturization and securing low resistance, the aspect ratio of a current or future semiconductor device is increasing.
For this reason, a foreign substance or defect hidden behind the step cannot be detected. The present invention can inspect such an object because the depth of focus is large. Further, by using an illumination optical system that illuminates from above using a mirror 168 and a half mirror 169 as shown in FIG.
Foreign matter or defects hidden behind shadows can also be illuminated and detected.

(Application to TFT) A substrate such as a TFT substrate having a large substrate and a small thickness has a large distortion when supported.
In such a case, as shown in FIG. 24, the substrate 30 is supported at both ends by the supporting portions 171 and 172, and the longitudinal direction 31 of the illumination spot and the longitudinal direction of the detector extend in the direction 32 connecting the supporting ends. The substrate is set to be vertical and the substrate is scanned in parallel to the direction 32. With such a configuration, even when the distortion can be increased only by supporting the peripheral portion, since the direction of the distortion can be made perpendicular to the direction of the illumination and the detector, no distortion occurs in the direction of the illumination and the detector. In addition, the inspection object can be put within the depth of focus within the detection field of view. Further, even when the distortion of the entire substrate is large, there is an effect that the shape of the distortion of the substrate can be easily calculated. For this reason, based on parameters such as the thickness, length, width, longitudinal elastic modulus, and lateral elastic modulus of the substrate, the strain of the free-end support at both ends is calculated, and the focus is adjusted according to this shape. Inspection can be performed while driving the stage in the focal direction. Such a configuration has an advantage that an automatic focus detection mechanism is not required. With the above configuration, it is also possible to inspect a large substrate 30 such as a TFT for a defect such as a foreign substance.

When a plurality of rows of detection head arrays are used as indicated by reference numerals 33 and 34, the distortion differs at each position. Need to be adjusted. Further, it is needless to say that such a configuration is not always necessary when the board is fixed on a flat support and the board is inspected. Further, by supporting the four corners of the substrate 30, distortion can be reduced.

(High-Accuracy Foreign Inspection Apparatus) As described above, the inspection apparatus using the spatial filter is not intended to realize only a high-speed and compact size. With the same configuration as above, if the resolution is improved by replacing the Fourier transform lens on the object side as shown in FIG.
It can inspect any foreign matter or defect at a high speed from 0.3 microns. Further, if the resolution is set to about 3 microns, any foreign matter or defect having a minimum of 0.3 to 0.8 microns can be inspected at high speed. In such a configuration, a signal of a pattern which becomes a noise can be skillfully erased by utilizing the repeatability of the pattern, so that a larger pixel size is used as compared with a pattern inspection apparatus for design data comparison, cell comparison, or chip comparison. However, since a small foreign substance or a defect can be inspected, a high-speed inspection can be realized as a result.

Even in such an inspection, since it is basically an inspection of only a repeated pattern, it has no repeatability.
The pattern part is out of the inspection target. It is necessary to devise such a part that is not to be inspected, for example, to inspect it with another inspection device or to perform a visual inspection.

(Utilization of Detecting Head) As described above, the foreign matter defect detecting device is most effective to inspect the substrate 1 while the substrate 1 is being transported in one direction. In order to perform the inspection, it is desirable to arrange a plurality of inspection devices (detection heads) in parallel as shown in FIG. However, as shown in FIG. 26, even if only a part 35 of the substrate is inspected using one unit of the detection head 101, the effect is often sufficiently exhibited. This is because some foreign matter defect inspections can detect a situation rather than the occurrence of foreign matter defects. Of course, this need not be one unit, and multiple units can be arranged as needed. A configuration in which one unit or more detection heads are used and the entire substrate is inspected by xy scanning of the stage as shown in FIGS. 27 (a) and 27 (b) may be used.

(Dynamic Inspection According to Pattern Shape) FIG. 28 shows an inspection apparatus in the case where the pitch of the repetitive pattern changes in the substrate. In this embodiment, the repetitive pattern pitch detection unit 174 and the detection head 101
Consists of

When the substrate is conveyed, first, the pitch of the repetitive pattern is calculated by the parameter calculating means 212 using the above-described frequency analysis method based on the signal detected by the repetitive pattern pitch detecting section 174. . The calculated pitch is sent to the detection head 101,
The pitch of the spatial filter and the pitch of the operator are changed. In the detection head 101, the inspection is performed when the portion in which the pitch detection unit 174 calculates the pitch is conveyed to the inspection position of the detection head 101. The inspection proceeds while the pitch is constantly calculated. Here, when the pattern pitch does not change discontinuously, inspection can be performed by this method.
When the pitch changes discontinuously, there is a possibility that the setting of the pitch may not be in time. In such a case, it is necessary to stop the feeding of the transport system until the pitch setting is completed.

(Diffraction Interferometry Using Sine-Curve Diffraction Grating) A pattern inspection method combining diffracted light and interference light will be described below with reference to FIG. This apparatus is basically the same as the inspection apparatus using the spatial filter described above, and includes an illumination system 102, a detection optical system 103, a spatial filter unit 106, and a detection mark 107. Grating 17 having a sine curve phase distribution formed on a transparent substrate
5 and a drive mechanism 176 are added. Here, in the method described above, the pattern removal processing using the repetition of the chip is realized by the electric processing system shown in FIG. As described above, this method has a problem that foreign substances in a pattern cannot be inspected unless they are particularly large. Further, as shown in FIG. 14, as described above, information in such a large background noise can be detected only by a method such as light processing that can eliminate the offset uniformly and reliably. Therefore,
The pattern removal processing using the repetition of the above chips is to be realized by an optical method, specifically, an interference method.

Here, the illumination shown in FIG. 29 is shown as transmitted light with respect to the substrate, but it may be transmitted light or reflected light. Here, the illumination is coherently illuminated in at least one axial direction. here,
When there are repeated patterns 37 and 38, the principal ray (0th-order diffracted light) of the light emitted from each travels along each optical axis and forms each image on the detection mark 107. However, since the diffraction grating 175 is provided on the Fourier transform surface, the light is diffracted and emits diffracted light in the ± first order directions. What is important here is that since the sine curve (the same cosine curve) is used as the diffraction grating 175, the zero-order diffracted light disappears (light intensity is 0), and ±
The point is that the light is emitted only to the primary light. Here, by adjusting the diffraction grating 175 in the optical axis direction by the driving mechanism, the +1 order light from the pattern 37 and the −1 order light from the pattern 38 can be superposed on the detector. Further, if a phase plate 178 that shifts the phase by π is placed on the illumination side or at an appropriate position on the optical axis, two light beams can be caused to interfere on the detection mark 107. As a result, if the phase change of the phase plate 178 can be finely adjusted so that the phase of the light beam in the middle can also be corrected, pattern removal processing using repetition of chips by interference can be realized.

Further, the diffraction grating 175 and the driving mechanism 176
If the distance between the gratings of the diffraction grating is appropriately changed by changing the wavelength of the ultrasonic wave using a refractive index variable mechanism based on a surface wave caused by an ultrasonic wave such as SAW, each + 1st-order light beam emitted from the diffraction grating 175 can be obtained. Primary light can be emitted in exactly the same direction and superimposed on the detector. This method has an effect that a sine curve can be automatically created on the diffraction grating 175. Further, it is not necessary to move the diffraction grating 175 in the optical axis direction. Of course, these may be used in combination.

In the above description, a diffraction grating having a sine curve is used. However, the present invention is not limited to this. When comparing the patterns 37 and 38 with a sufficiently large interval, the 0th-order diffracted light generated due to the non-sine curve is used. It is not necessary to have a sine curve because the influence can be suppressed. FIG. 29B shows a configuration in which, when illuminating a pattern, only spots separated by pitches of the pattern are illuminated with spots. In this configuration, the light source 17
Numeral 9 has a scanning device, and illuminates only a portion of the pattern separated by a pitch via a half mirror 180 and a mirror 181 with a spot. With such a configuration, an accurate phase shift π can be created, and the detection performance can be improved.

(Others) As described above, the inspection apparatus is for inspecting foreign substances or defects on the substrate being transported, so that the substrate may fall within the depth of focus of the detection optical system during transport. Need to be. Therefore, it is preferable that the depth of focus is deep in order to enable inspection even in a transport system that is not so high in accuracy. Therefore, in order to give priority to the depth of focus over the detection resolution, an aperture may be provided in the detection optical system to increase the depth of focus.

In the inspection apparatus described above, a linear filter is used. However, this filter is not necessarily required to be such a filter. A reversible light cut filter used may be used, or a complex conjugate type nonlinear element may be used.

In the above embodiments, the case where the illumination light is light from information, that is, reflected light has been shown. It is a configuration with the lighting of, no problem.

The foreign object defect information detected by the inspection apparatus of the present invention not only detects an abnormality by counting the number of foreign object defects but also determines the cause of the foreign object defect by knowing the occurrence distribution of the foreign object defects. It can be a clue to analogy. Further, the arrangement, the number of monitors, the sensitivity, and the like of the monitor of the present invention may be set based on the result of the high-precision inspection device.

Although the illumination system of the present invention is configured independently of the detection system as the illumination system 102, this illumination system can be omitted by using a part of the detection system. Specifically, this can be realized by mounting a semiconductor laser and a cylindrical lens having an appropriate focal length on the Fourier transform surface of the detection optical system. By doing so, the detection head can be made smaller, lighter, and inexpensive.

In the above configuration, a semiconductor laser is used. However, the present invention is not limited to this, and the effects of the present invention can be obtained with a gas laser, a solid laser, an almost short wavelength, and a white light from a point light source.

Although the small foreign matter detection head described above uses the spatial filter unit 106, the semiconductor laser 112 is used as polarized light without using the spatial filter unit 106, and the polarized light of this illumination is shielded at the position of the spatial filter 106. Almost the same foreign matter detection effect can be expected even if a polarizing plate in the direction is mounted. In this case, the polarization direction may be p-polarized light or s-polarized light, but s-polarized light can be expected to have higher detection performance. In addition, the polarizing plate does not necessarily need to be placed at the position of the spatial filter 106 and may be anywhere in the detection optical system 103. Also, the semiconductor laser 112 does not need to have a polarization output, A polarizing plate may be inserted in the device. Of course, in this case, the detection performance is slightly lower than in the case of using a spatial filter, as in other foreign matter inspection devices that use polarized light. Has an effect. Even when the polarizing plate is used, the scattered light from the non-repeated pattern around the repetitive pattern cannot be sufficiently eliminated as described above, so that the operator processing system 203 of the present invention is effective. At the same time, the effect of the telecentric optical system 103 'for improving the function of the operator processing system 203 is great. Therefore, a mechanism for adjusting the rotational displacement of the substrate 1 is also required. However, basically, the method using the polarizing plate is strong against the rotational deviation of the substrate, and the rotational deviation does not always need to be adjusted. Further, a means for erasing the pattern of the test element group is also required. These means, as explained above,
Not all of them are required at the same time, and a great effect can be expected even if they are used one by one or in an arbitrary plural combination. Therefore, they should be combined experimentally or in accordance with the detection conditions of a high-precision defect foreign matter inspection device, depending on the size of the foreign matter to be detected, the required detection performance, the conditions of the target substrate to be detected, and the like. .

When the polarizing plate and the spatial filter 106 are used in combination, the performance of detecting foreign matter defects is further improved. Also in this case, as described above, these means are not required all at the same time, and a great effect can be expected even if they are used one by one or in an arbitrary plural combination. Since the amount of detected light is slightly reduced, it is necessary to improve the illuminance of illumination. Again,
Specifically, it is necessary to reduce the beam width on the substrate and improve the precision of the stage by using an optical system having a large NA. In such a case, for example, a well-known automatic focusing mechanism such as a method of projecting a stripe pattern or detecting a displacement of a laser beam is required.

FIG. 31 shows an example of a configuration block diagram of a method and apparatus for starting up mass production in a semiconductor manufacturing process and inspecting foreign substances on a mass production line.

In FIG. 31, the mass production start-up in the semiconductor manufacturing process and the foreign substance inspection apparatus for the mass production line are performed by the exposure apparatus 5
11 etching apparatus 512, cleaning apparatus 513, ion implantation apparatus 514, sputtering apparatus 515, CVD apparatus 516
And a temperature sensor 52.
1, a foreign matter monitor 522 in the transport system, a pressure sensor 523, a foreign matter monitor 524 in the processing apparatus, and the like
0 and its sensing unit control system 525
A utility group 530 including a gas supply unit 531 and a water supply unit 532; a sampling unit 540 including a water quality sampling wafer 541, a gas sampling wafer 542, an in-apparatus sampling wafer 543, a device wafer 544, and an atmosphere sampling wafer 545; A detection unit 550 including a foreign matter detection unit 551 and a pattern defect detection unit 552; a scanning electron microscope (SEM); a secondary ion mass spectrometer (SIMS) 562; and a scanning tunneling microscope / spectrometer (STM / STS) 563; An analysis unit 560 including an infrared spectrometer 564, a foreign matter fatality determination system 571, and a minute foreign matter cause investigation system 57
2 and a pollution source countermeasure system 573. These components are divided into an on-line foreign substance inspection system 581 corresponding to the line and an off-line foreign substance inspection system 582 corresponding to the mass production start-up line, and the mass production start-up and mass production line foreign substance inspection system 580 in the semiconductor manufacturing process are combined. Make.

Therefore, as shown in the figure, by separating the mass production line from the mass production line, the foreign matter detection, analysis and evaluation device can be efficiently operated at the mass production start, and the mass production can be started quickly. In addition, the inspection and evaluation equipment for foreign substances used in the mass production line can be reduced to a necessary and simple monitoring device to reduce the weight of the mass production line.

Next, the online foreign matter inspection system 581
An example of the foreign matter monitor 522 in the transport system and the foreign matter monitor 524 in the processing apparatus, which are online monitors of the present invention, will be described. FIG. 32 shows an example in which a foreign substance monitor 3101 (schematic configuration is shown in FIG. 1) which is an online monitor is applied to a transport system of a single wafer type CVD apparatus 516 having a large number of defects among semiconductor manufacturing apparatuses 510. A loader 3102 having a foreign substance monitor 3101, a preliminary chamber 3103, and a reaction chamber 3104
, A heating unit 3105, a gas system 3106, a controller 3107, and a host CPU 3108.
The product wafer 3111 is transferred from the loader cassette 3109 placed in the loader unit 3102 to the preliminary chamber 3103, the gate valve 3112 is closed, and the preliminary chamber 3103 is evacuated.
Next, the gate valve 3113 is opened, and the product wafer 3111 (1) in the preliminary chamber 3103 and the reaction chamber 3104 is exchanged.
The gate valve 3113 is closed, and film formation is started in the reaction chamber 3104. During the film formation, the preliminary chamber 3103 is returned to the atmospheric pressure, the gate valve 3112 is opened, and the product wafer 3111 is opened.
During the transportation of the collected wafer to the unloader cassette 3110, the foreign matter on the product wafer 3111 (1) is measured by the foreign matter monitor 3101 (schematic configuration is shown in FIG. 1). A foreign substance monitor 3101 may be provided immediately before the gate valve 3112 to compare foreign substances before and after film formation.

Next, the configuration of the foreign substance monitor 3101 (the schematic configuration is shown in FIG. 1) will be described with reference to FIG. First, the product wafer 31 is detected by the wafer rotation direction detector 3121 (162) provided on the foreign substance inspection start side of the foreign substance monitor 3101.
The orientation of the orientation flat 11 (1) is detected, and the product wafer 31 is detected.
The rotation direction of 11 (1) is detected. After that, a foreign substance inspection on the product wafer 3111 (1) is performed on the entire surface by the foreign substance detection optical system 3122 (101). Next, the foreign substance monitor 310
1 is processed by the foreign matter information processing system 3123, and if there is an abnormality of the foreign matter, the foreign matter information is notified by an alarm or the like, or the apparatus main body 31
25 can be stopped. Also, the keyboard 312
6 and CRT 3127 (230) to display foreign matter. Further, it is linked with the foreign substance analysis system 3128, and can exchange data. For example, from the system 3128, the name, location,
By transmitting a command of desired data such as sampling, the data can be obtained from the foreign matter information processing system 3123. Here, in the foreign substance monitor 3101, the foreign substance detection optical system 3122 (101) is connected to the foreign substance information processing system 31.
23 (202, 203, 206-211, 212-21
3, 229, 230, 232)
Further, the apparatus does not have a stage system, and uses a transport system of the processing apparatus. However, of course, a configuration having a stage system is also possible. Therefore, the external dimensions of the foreign substance monitor 3101 are such that the width W, the depth L, and the height H are each within 1 m, or the width W of the foreign substance monitor 3101 is shorter than the width Ww of the wafer, thereby enabling downsizing. In addition, the foreign substance monitor 3101 has an automatic calibration function, and since the reflectance of the product wafer surface differs between manufacturing apparatuses and processes, the reflectance is automatically measured and fed back to the illumination light amount of the foreign substance detection optical system. And does not require cumbersome calibration. Further, the depth of focus d of the detection lens of the foreign object detection optical system 3122 (103) is calculated from the following equation, and is 0.1 to 0.5 mm, which does not require an automatic focus.

[0157]

D = 0.5λ / (NA) ² (Equation 16) where λ is the wavelength of light and N.A. A. Is the numerical aperture of the detection lens. Furthermore, because of its small size, the unit can be replaced, and the structure is easy to mount and set on the apparatus, so that maintenance is easy.

Referring to FIG. 34, wafer rotation direction detector 3121
The detection method (162) will be described. A product wafer 3111 is placed under an illumination system having several or more light emitting points 3131.
(1) passes along the wafer movement direction 3133 and 31
It moves from the position 32 to the position 3134. The figure shows a locus 3135 on the product wafer 3111 (1) of the illumination light emitted from the light emitting point of the illumination system of the wafer rotation direction detector 3121. In the case of the light emitting point A, the illumination light is
The time As corresponding to (1) and the time Ae deviating from the product wafer 3111 are measured, and this is also performed for the other light emitting points B to G. The orientation of the orientation flat of the product wafer 3111 is obtained from the above data and the moving time of the product wafer 3111,
The rotation direction of the product wafer 3111 is calculated. As a method of detecting the rotation direction of the product wafer 3111, there are scribe area detection, chip detection, and detection of a special mark such as an alignment mark.

Therefore, the foreign substance monitor 3101 (schematic configuration is shown in FIG. 1) is a wafer rotation direction detector 3121.
The rotation direction of the product wafer 3111 (1) obtained in (162) and, as shown in FIG.
The position of the detected foreign matter on the product wafer 3111 (1) is determined by the coordinates based on the orientation flat with the intersection of the axes as the virtual origin 3141 or the coordinates based on the circuit pattern 3142 with the intersection of the extension of the circuit pattern 3142 as the virtual origin 143. Foreign object coordinate management that can obtain information is possible.

In order to know the distribution of dust in the apparatus, FIG.
As shown in FIG. 6, the rotation direction of each product wafer 3111 (1) varies in various directions 3142, 3143, 3144, 314.
5, the transport direction 3
Device-based foreign object coordinate management that does not depend on the rotation direction of the product wafer 3111 (1) consisting of the x-axis at which the outer periphery of the product wafer 3111 (1) contacts the outer periphery of the product wafer 3111 (1) and the y-axis at right angles to the outer periphery of the product wafer 3111 (1). Also have. If there is dust in the apparatus, a regular distribution of foreign substances as shown by 3146 is shown.

Further, the wafer rotation direction detector 3121 (16) of the foreign matter monitor 3101 (schematic configuration is shown in FIG. 1).
In 2), since the transport speed of the product wafer 3111 can be obtained at the same time as detecting the rotation direction of the product wafer 3111 (1), scanning of a detector, for example, a CCD linear sensor, is synchronized with the transport speed of the product wafer 3111. The speed can be changed. Therefore, stable detection performance can be obtained regardless of the transfer speed of the product wafer 3111 (1).

FIG. 37 shows an embodiment of a configuration diagram of a foreign substance detection optical system 3122 (101) using a spatial filter which is capable of performing a high-speed foreign substance inspection on the product wafer 3111 (1) and having a small structure. Oblique illumination optical system 3151 and detection optical system 31
52. The oblique illumination optical system 3151 comprises one or more illumination arrays as shown in the figure. Detection optical system 31
Reference numeral 52 denotes a lens array 3153 as a detection lens, and a spatial filter 3154 (10
6), the detector 3155 (10
7).

FIG. 38 shows the configuration of the oblique illumination optical system 3151. Here, oblique illumination refers to the product wafer 3111
This means that illumination is performed from the direction 3164 inclined by θ from the normal 3163 of (1). A small, high-output semiconductor laser 3161 is used as an illumination light source, and high-luminance coherent light illumination is enabled by an anamorphic prism 3162. This is because by performing coherent light illumination on the product wafer 3111 (1), a sharp Fourier transform image of the pattern of the product wafer 3111 (1) can be obtained on the Fourier transform surface of the detection lens 3153. In addition, the anamorphic prism 3162 allows for large area illumination that is not affected by adjacent illumination components of the illumination array. When there is an influence of the adjacent illumination light, the Fourier transform image of the pattern by the adjacent illumination is shifted and overlapped on the Fourier transform surface of the detection lens 3153, so that the area of the Fourier transform image increases, and the area of the filter portion of the spatial filter also increases. This is because the amount of scattered light from the foreign matter that passes through the spatial filter 3154 (106) decreases, and the foreign matter detection performance decreases.

FIG. 39 shows the detection width of the detection optical system 3152. The detection width 3170 of the detection optical system 3152 is the same as the width of the product wafer 3111 (1),
The entire surface of the product wafer 3111 can be inspected collectively by only one scan 3156 of the feed 3156, and high-speed inspection can be performed.

FIG. 40 shows a case where a CCD linear sensor is used as the detector 3155 (107). Product wafer 3
Since the width of 111 (1) is detected at once, as shown in FIG.
The CD linear sensors 3171 are arranged in a zigzag pattern. In addition, the column B is deleted for the 3172 that is the overlapping portion of the sensors, and the data in the column A is made valid.

FIG. 41 shows a spatial filter 3154 (106).
FIG. Each spatial filter 318 is attached to each lens element 3181 of the lens array 3153 (107).
2 corresponds.

FIG. 42 is a detailed diagram of the spatial filter 3154. The diffracted light 3191 from the regularly repeated pattern of the product wafer 3111 becomes a regular image 3192 at the position of the spatial filter 3154 on the Fourier transform plane of the lens array 3153. Therefore, a regular repetitive pattern on the product wafer 3111 can be shielded by the spatial filter 3154 as shown in FIG.
It is not captured by the D linear sensor 3155.

As the spatial filter 3154, a spatial filter of a dry plate system which is created by printing a Fourier transform image of a repetitive pattern of the product wafer 3111 on a dry plate is used.
Therefore, the baked portion of the spatial filter 3154 does not pass light from the regularly repeated pattern of the product wafer 3111. Alternatively, there is a spatial filter of a liquid crystal system using a liquid crystal. First, the spatial filter 3 on the Fourier transform plane of the lens array 3153 of the diffracted light 3191 from the regular repeating pattern of the product wafer 3111
The regular image 3192 at the 154 position is detected by a TV monitor or the like, and the position of the liquid crystal element corresponding to the image 3192 is stored. Next, by applying a voltage to the memorized liquid crystal element portion, light hitting that portion can be shielded. Therefore, by storing and formatting the driving liquid crystal elements corresponding to the image of each product wafer in each step, it is possible to perform a spatial filter by turning on and off the liquid crystal of each product wafer in each step.

FIG. 43 shows a dry filter type spatial filter group 3201 corresponding to the product wafer 3111 in each step. A spatial filter corresponding to the product wafer 111 of each process is prepared by a dry plate method, replaced by a moving mechanism such as a linear guide stage as shown in the figure, and positioned with respect to the detection lens 3153, thereby obtaining a product filter of all processes. 3111.

FIG. 44 shows an AND space filter 3221 using a dry plate method. By taking the AND of the spatial filters of several processes, the number of spatial filters can be reduced, and one AND spatial filter 3222,3
At 223, light from the repetitive pattern of the product wafer 3111 in several steps can be blocked. Therefore, by using the AND spatial filter 3221, the number of spatial filters can be reduced even when there are many steps,
The device configuration can be simplified. This method can also be used for a liquid crystal spatial filter, and can reduce the number of formats to be stored. However, it is possible to take the AND of the spatial filters in all the steps and use one AND spatial filter. However, the amount of scattered light from the foreign matter passing through the AND spatial filter is reduced, and the foreign matter detection performance is reduced.

Next, FIG. 45 shows an embodiment of a configuration diagram of the optical system 3122 for detecting foreign matter by partial inspection. A micro lens group 3231 is used as a detection lens, and a spatial filter 3232 is provided on the Fourier transform surface of each micro lens 3231.
(106) is arranged, and a CCD linear sensor 3233 (107) is arranged as a detector. Therefore,
By using the microlens 3231 having high resolution, finer foreign substances can be detected than using the lens array 3153. However, in this method, the inspection can be performed even when a conventional lens is used instead of the micro lens 3231 as the detection lens. Micro lens group 3 as one example of partial inspection
The inspection area can be made effective by adjusting the pitch of 231 to the interval between the chips of the product wafer 3111.

However, as shown by the hatched portion in FIG.
1 is a partial inspection, and the function of monitoring foreign substances can be performed, but the entire surface of the product wafer 3111 cannot be inspected. Here, 3236 is the detection width of one microlens 3231. However, by scanning the product wafer 3111 several times, the entire surface of the product wafer 3111 can be inspected. Alternatively, as shown in FIG. 47, by arranging the micro lenses 3241 in two rows or several rows in a zigzag pattern, one scan 315 of the product wafer 3111 can be performed.
Only 6 allows a full inspection. In addition, the micro lens 3
241, a spatial filter 3242 is arranged on the Fourier transform plane, and a CCD linear sensor 324 is used as a detector.
3 are arranged.

In FIG. 45, as another embodiment, the product wafer 3111 is illuminated with a wide area and high illuminance by using a pulsed laser for the oblique illumination system 3151. Further, if a two-dimensional CCD sensor or a TV camera 3233 is used as a detector, detection can be performed in a wide area. Here, when pulse light emission is performed in the oblique illumination system 3151, the detector also detects the pulse light in synchronization with the pulse light emission.

In the above description, when a spatial filter is used, when each product wafer 3111 is conveyed in a constant rotation direction, for example, an orientation flat aligning mechanism is installed in the middle of the conveying system of the apparatus, and the direction of the spatial filter is set. By adjusting the direction of the product wafer 3111 to the above, the spatial filter can be detected. However, when the rotation direction of each product wafer 3111 is conveyed in various directions, the direction of the repetitive pattern of the product wafer 3111 also changes. Therefore, it is necessary to rotate the spatial filter in accordance with the rotation direction of the product wafer 3111. When the microlenses shown in FIGS. 45 and 47 are used, the adjacent spatial filters are independent, so that the individual spatial filters may be rotated in accordance with the rotation direction of the product wafer 3111. However, when a lens array is used, since the adjacent spatial filters are connected, the foreign matter detection optical system 3122 (3253) is set to 3254 in accordance with the rotation direction (rotation position of the orientation flat) 3251 of the product wafer 3111 as shown in FIG. Rotate like 32
It must be in the direction of 52. Of course, even when a micro lens is used, the rotation direction 32
The foreign matter detection optical system 3122 may be rotated in accordance with 51.
Here, the direction of 3251 and the direction of 3252 are the same. The rotation angle is 45 ° at the maximum, and in the case of FIG. 48, the detection width becomes longer by the rotation.

When a spatial filter is used, regular repetitive pattern portions on the product wafer 3111 can be inspected, but other portions cannot be inspected. Therefore, portions other than the regularly repeated pattern portion on the product wafer 3111 are set as invalid data or detection prohibited areas using software or the like. However, in this case, the product wafer 3
Although not monitoring all the points on the product 111 for foreign matter, it monitors the product wafer 3111 at a specific ratio. However, in the manufacture of a memory having a large number of repetitive patterns, only the repetitive portion of the memory is monitored. But the effect is great.

Next, FIG. 49 shows an embodiment of a configuration diagram of the optical system 3122 for detecting foreign matter by white light illumination. It comprises a lens array 3153 and a detector 3155 as an oblique illumination system 3261 using white light and a detection optical system 3262. When this method is used, the performance of detecting foreign substances is lower than in the spatial filter method. However, as shown in FIG. 50, the white light illumination detection 3271 has higher detection performance than the laser illumination detection 3272 not using a spatial filter, and is not limited to a regular repeating pattern portion on the product wafer 3111.
The entire surface can be inspected. Here, the detection output from the foreign substance uses the peak values of all the patterns on the product wafer 3111 as the reference 3273.

Next, FIG. 51 shows an embodiment of a configuration diagram of an optical system for detecting foreign matter by wafer comparison inspection. It comprises an oblique illumination optical system 3151 and a detection optical system 3152. The oblique illumination optical system 3151 comprises one or more illumination arrays as shown in the figure. The detection optical system 3152 includes a lens array 3153 or a micro lens group as a detection lens, and a spatial filter 315 on the Fourier transform surface of the detection lens 3153.
4. Detector 3155 at the image forming position of detection lens 3153,
Further, it comprises an image processing system 3280 that performs image processing on a detection signal from the detector. First, the product wafer 3111
Is detected and stored in the memory 3282 as an image. Next, the detected image 3281 that has detected the second product wafer 3111 and the first stored image 3282 are compared by the comparison circuit 3.
By making the comparison according to 283, the foreign matter is revealed. Third and subsequent product wafer 3111 detected images 3281
Is compared with the first or immediately preceding second stored image 3282. In this embodiment, pattern information is reduced by using a spatial filter 3154. Therefore, an orientation flat alignment mechanism or the like is installed before detection by the present foreign matter detection optical system, and the rotation direction of all the product wafers 3111 is adjusted to the rotation direction of the spatial filter.

FIG. 52 shows a semiconductor FA (Factory Automati) using a foreign substance monitor 3101 (a schematic configuration is shown in FIG. 1).
The system diagram of (on) is shown. An integrated processing station 3291 capable of integrated processing of the product wafer 3111 (1), various job stations 3292 corresponding to various kinds of special processing, an inspection station 3293, and an analysis station 3294 are provided. Are combined. In the integrated processing station 3291 and the various job stations 3292, a foreign substance monitor 3101 is mounted on a CVD apparatus, an etching apparatus, or the like, which is particularly likely to have a large number of defects, and monitors foreign substances in the apparatus. Also, as shown in 3296 and 3297, the foreign matter monitor 3101
To monitor foreign matter in the entire station.

It should be noted that the present invention can be applied to mass production start-up.
Naturally, it is effective for monitoring mass production lines.

Next, another specific embodiment of the small foreign matter monitor according to the present invention will be described with reference to FIGS.

Hereinafter, the structure of this embodiment will be described with reference to FIG. In the present embodiment, an illumination optical system 1110 including a semiconductor laser 1111, a collimator lens 1112, an x-diffusion lens 1113, a condenser lens 1114, a y-diffusion lens 1115, and a mirror 1116, and an imaging lens 121
1, 1221, spatial filters 1212, 1222, polarizers 1213, 1223, one-dimensional detectors 1214, 12
24, a stage system 1300 including a wafer transfer means 1301, an automatic focus detector 1312, and an automatic focus positioning mechanism 1313.
A / D converter 1411, threshold circuit 1412, two-dimensional image cutout circuit 1413, pattern foreign matter determination circuit 14
14, a signal processing system 1401 composed of pattern information memories 1418 and 1416 and foreign substance information memories 1417 and 1415, an FFT circuit 1511, a repetition part removing circuit 1512, a data memory 1513, a microcomputer 1515, a data display system 1516, and an abnormal display. And a data processing system 1501 including an alarm 1517.

In the illumination optical system 1110, the semiconductor laser 1
Light emitted from the light 111 becomes a plane wave by the collimator lens 1112 and is spread only in the x direction by the x diffusion lens 1113. The light emitted from the x-diffusion lens 1113 is condensed by the condenser lens 1114 into a parallel light beam, that is, a plane wave in the x-direction, and is condensed in the y-direction. Then y diffusion lens 11
By 15, only the y direction is diffused to a parallel light beam. As a result, a parallel beam in both the x and y directions, that is, a plane wave, becomes a linear beam long in the y direction, and the wafer (semiconductor substrate) 1
001 is illuminated.

FIG. 54 shows a configuration of the illumination optical system 1110 viewed from the x direction, and FIG. 55 shows a configuration of the illumination optical system 1110 viewed from the y direction.
In the y direction, the illumination area on the wafer (semiconductor substrate) 1001 is sufficiently widened as much as possible to be illuminated, and the illuminance is narrowed down in the x direction so as to have a sufficient illuminance. However, the illumination is a plane wave, that is, a light beam parallel to both the x and y directions.

Here, in the present embodiment, illumination is performed with a parallel light flux, that is, a plane wave in both the x and y directions. However, any optical system that approximates a plane wave may be used. Here, linearly polarized light is applied so that the magnetic field vector is perpendicular to the plane of incidence of the illumination. Thereby, there is an effect that scattered light from the foreign substance is relatively improved with respect to scattered light from the pattern.
However, it is not always necessary to use s-polarized light, and other linearly polarized light or elliptical or circularly polarized light may be used to achieve the object of the present invention.

In the detection optical system 1210, the wafer 1001
The light beam emitted from the upper inspection position 1002 is
211, 1221, the spatial filter 1212, 1
The image is formed on the one-dimensional detectors 1214 and 1224 through the polarizing plates 1213 and 1223. Polarizing plate 121
No. 3,1223 shields light (S-polarized light) whose magnetic field vector is perpendicular to the plane of incidence of the illumination. This polarizing plate has an effect of improving scattered light from a foreign substance relatively to scattered light from a pattern. However, it is not always necessary, and even if it is omitted, there is no problem in achieving the object of the present invention.

The imaging lens 1211 of the detection optical system
May use a normal lens as shown in FIG. 56, or may use a lens array of a refractive index change type as shown in FIG. In any case, the illumination optical system 11
When an optical system capable of illuminating a plane wave as shown in FIG. 54 and FIG.

FIG. 58 is a plan view of the illumination optical system and the detection optical system. Detection optical systems 1210, 1220, 1230,
1240, 1250, 1260 and one-dimensional detector 12
A plurality of wafers 14, 12, 24, 1234, 1244, 1264 are arranged to cover the entire diameter L of the wafer. In addition, each illumination optical system 1110, 1120, 113
0, 1140, 1150, and 1160 are arranged to illuminate the detection areas of the one-dimensional detectors 1214 to 1264, respectively. With this configuration, the entire area of the wafer can be illuminated with a parallel light beam, that is, a plane wave. In this configuration, one inspection area is illuminated from one illumination direction. With this configuration, the effect of the spatial filter is sufficiently exhibited.
If one lighting area is illuminated from multiple directions,
Diffraction patterns by these multiple illuminations on a spatial filter
In this case, it is necessary to enlarge the light shielding area by the spatial filter. When the light-shielding region is made large in this way, an optical signal to be detected is also shielded by the light-shielding region. Illumination from one direction can prevent this.

As shown in FIG. 53, the stage system 1300
Then, after placing the wafer 1001 on the wafer transfer means 1301, the wafer transfer means 1301 moves in the x direction. Here, the wafer transfer means 1301 is a transfer system included in another processing apparatus, specifically, a semiconductor manufacturing inspection apparatus such as a film forming apparatus, an etching apparatus, and an exposure apparatus. Of course, the foreign matter inspection apparatus of the present invention may have this transporting means. Further, the distance between the wafer 1001 and the apparatus according to the present invention is measured by the automatic focus detection system 1312, and based on the measurement result, the automatic focus control system 1313 controls the distance between the wafer 1001 and the apparatus according to the present invention to be optimal. Is done. It is sufficient that this control is performed only once before the start of the inspection, but depending on the accuracy of the wafer transfer means 1301, it may be necessary to perform control in real time during the inspection.

In the signal processing system 1410, the one-dimensional detector 1
The detection signal from 214 passes through an A / D converter 1411 and a threshold circuit 1412, and a binarized 1-bit signal
The pattern and the foreign matter are sent to the × 5 two-dimensional image cutout circuit 1413 and are determined by the pattern foreign matter determination circuit 1414 based on the logical formula shown in FIG. That is, assuming that the logical value of the center point is P (0,0), when the following equation (Equation 17) is satisfied, the signal of p (0,0) is determined to be a foreign substance, and the following equation (Equation 18) Is determined as a pattern when.

[0190]

[Equation 17]

[0191]

(Equation 18)

The result of the judgment is the one-dimensional detector 1214
Are stored in the pattern memory 1415 and the foreign substance memory 1416 according to the coordinate signal obtained from the basic clock. Here, the threshold circuit 1412 outputs the foreign object memory 1416
A plurality of circuits such as three systems are prepared, and the threshold value of the threshold circuit 1412 is changed stepwise. With such a circuit configuration, there is an effect that the circuit scale is reduced while having necessary and sufficient functions.

Here, this signal processing system corresponds to each detection optical system 1
In order to process the signals 210 to 1260, the signal processing system 1
410 to 1460 are provided.

In data processing system 1501, foreign object memory 1
The foreign matter map data is Fourier-transformed from the data 416 by the FFT circuit 1511,
12 removes the repeated portion between the chips. The thus obtained foreign substance data stores coordinates and a threshold value in the foreign substance memory 1513. When the number of foreign substances is larger than the allowable range, an alarm signal is issued from the alarm 1517.
When this alarm signal is issued, the operator stops the operation of the line, pursues the cause of the foreign matter, and takes measures. In addition, when instructed by the microcomputer 1515, map data, coordinate data, and the like of the foreign matter are output to the display system 1516. In the present invention, pattern data is also stored in the memory 1416. This data means that no foreign substance inspection has been performed on this pattern portion. Therefore, the ratio of the pattern data to the entire area means the inspection area ratio. If the inspection area ratio is smaller than a predetermined value, there is a possibility that an error of the inspection apparatus or an error of the wafer process occurs. Therefore, also in this case, an alarm is issued from the alarm 1517.

FIG. 81 shows a foreign matter pattern judging system which also has the functions of the signal processing system 1410 and the data processing system 1501. In the data processing system 1501, non-repeated patterns around the chips are discriminated and removed by using the repeatability of the chips in the wafer. This function is also achieved by the circuit shown in FIG.

In this embodiment, the two-dimensional image cutout circuit 1
An image cutout circuit 1420 is provided in place of 413. Image cutout circuit 1420 (21 of operator processing system 203)
6, 217) are cutout portions 1421 (219), 142
2 (231) and the determined part 1423 (218). The cut-out portions 1421 (219) and 1422
(231) is arranged so that an image at a position separated by a chip pitch p on the sample can be cut out with respect to the to-be-determined portion 1423 (218). Here, the rotation error Δ
α, a chip transfer error, an imaging magnification error, and a chip interval error Δp due to an error due to binarization, so that the image cutout units 1421 and 1422 have a margin of approximately ± Δα and ± Δp with respect to the determined unit 1423. have. This value is a value that may be designed experimentally or based on the manufacturing accuracy of the apparatus. In this embodiment, the pixel size is 7 μm.
m, Δp is 1.5 pixels, Δα is 0.5 degrees, the pitch is about 10 mm, and Δw (= Δα · p) is 1
It has 2.5 pixels. This image cutout circuit 1420
Are processed according to the signal processing system shown in FIG. In the two-dimensional cutout circuit 1413, a logical product is obtained in a square shape in accordance with Equation 1 on the periphery of the cut square, and a logical product is obtained over the entire area of the cutout portions 1421 and 1422. That is, in the two-dimensional extraction circuit 1413, the periphery of the extracted square is set to P (i, j), whereas the extraction unit 142
In 1,1422, the whole area cut out is set to P (i, j). The determination of the pattern is represented by Expression 1 only by the difference in the shape of P (i, j), and the determination of the foreign matter is represented by Expression (17).

In this configuration, the FFT circuit 1511 and the repetitive portion removing circuit 1512 can be omitted.

The operation will be described below with reference to FIGS. 53 to 62.

In the present invention, attention is paid to the repeatability of a pattern in order to inspect a foreign substance on an VLSI having an ultrafine pattern formed thereon at high speed with high accuracy and with a small apparatus. In a conventional apparatus, a high-performance large-sized apparatus has been used to inspect the entire area of a wafer at high speed and with high accuracy. However, in order to improve the yield of semiconductor production, it has been found that it is better to conduct a whole wafer inspection at the expense of a whole area inspection rather than a foreign substance inspection for the whole area. did. As long as conventional equipment is used, wafers must be sampled and inspected at an appropriate frequency,
In this inspection method, once a defect occurs, a large amount of defects may be produced. When such an entire wafer inspection is performed, defects such as device dust and process dust can be found without inspecting the entire area of the wafer.

Therefore, attention has been paid to the fact that repeated patterns are present at a large ratio in LSIs represented by memories. It is 80% or more for DRAMs and SRAMs, and 30% or more for microcomputers and custom LSIs in many cases. With such a ratio, it is sufficient to inspect only this repetitive portion. For inspection of defects and foreign matter in a repeated portion, an enhancement detection technology for a non-repeated portion using optical filtering is effective. Therefore,
Suitable for this technique. The problem with this method is how to create a spatial filter.

Basic pattern 101 as shown in FIG.
When light is illuminated on the repetition pattern of 0 by the device shown in FIG. 53, a regular diffraction pattern 1011 as shown in FIG. 60 is observed by the spatial filters 1212 and 1222. This diffraction pattern 1011 is due to diffraction from the pattern shown in FIG. Here, when the foreign matter 1012 exists on FIG. 29, the diffracted light from the foreign matter 1012 has an irregular shape different from the regular diffraction pattern 1011. For example, the pattern 10 shown in FIG.
Observed as 13. Therefore, if a filter that blocks the diffraction pattern 1011 is provided on the spatial filters 1212 and 1222, the information of the pattern 1014 is deleted, and on the one-dimensional detectors 1214 and 1224,
Only the information on the foreign material 1012 is observed as shown in FIG.
That is, according to the present invention, only the foreign matter 1012 is selectively detected.

Here, the relationship between the pitch p of the pattern 1014 and the pitch θ of the diffraction pattern 1011 (indicated by the angle of the diffraction pattern incident on the imaging lenses 1211 and 1221 from the observation point 2) is shown in the illumination optical system. The wavelength (λ) of the light emitted by 1110 is shown by the following (Equation 19).

[0203]

Equation 19 sin θ = λ / p (Equation 19) Accordingly, θ decreases as p decreases. That is, L
As the SI becomes finer and the p becomes smaller, the θ of the diffraction pattern becomes larger, the diffraction pattern incident on the imaging lens 3211 decreases, and there is an advantage that the shape of the spatial filter becomes simpler.

In the case of the same product, the basic pattern 10
Even if the shape of 10 changes, the basic pitch of the diffraction pattern does not change because the position pitch does not change. In other words, as long as the same product is inspected, the shape of the diffraction pattern hardly changes, and therefore, the shape of the spatial filter that blocks the light does not substantially change. Utilizing this feature, even if a spatial filter that measures the shape of the diffraction pattern in each process for each product and blocks all the diffraction patterns is created, the filter blocks the entire aperture of the imaging lens. We noticed that there is no such thing. As described above, by using a filter that shields all the diffraction patterns in each step, replacement of the spatial filter can be omitted. In particular, the memory manufacturing line is effective because the number of products is small and the number of products is small.

Here, in the present invention, the imaging lens 121
If a refractive index change type lens array is used for 2,1222, the apparatus can be further miniaturized. The refractive index variable lens array can be configured as a facsimile,
Used in electronic copiers and the like. In order to achieve the purpose of reducing the size of the optical system, this lens array of the refractive index change type is effective. However, the present invention requires the use of a spatial filter. Conventionally, attention has not been paid to the fact that a refractive index change type lens array also has a Fourier transform surface and a spatial filter can be used. In the present invention, focusing on the fact that a spatial filter can be used for the refractive index variable lens array, a small foreign substance monitor using the refractive index variable lens array has been realized. The configuration and operation of the spatial filter are the same as those described above, and the spatial filter may be provided for each lens. In addition, the position of the spatial filter of the lens array of the refractive index change type is located at the end face on the exit side of the lens as shown in FIG.

FIG. 62 shows the shape of the spatial filter. In particular, in order to provide the simplest and effective effect for an arbitrary pattern, a straight line shown in FIG. Also, in order to obtain the discrimination performance from the pattern from the spatial filter on the straight line, a shape as shown in FIG. 62B is required. Further, FIG. 62 (c) shows an example of a shape that can be used in all steps in the product.

FIG. 63 shows an example of detection of foreign matter.

Here, in realizing a high-speed and compact foreign matter inspection apparatus, the method using this spatial filter is more suitable than the polarization detection method shown in the prior art (Japanese Patent Laid-Open No. 62-89336). The reason will be described with reference to FIGS.

In the method of illuminating a sample with light and detecting scattered light from a foreign substance, scattered light from a pattern formed on the sample surface becomes noise. This noise increases as the size of the pixel (the minimum unit detected as one signal) of the detector 2006 increases, as shown in FIG. Since the pattern serving as a noise source is formed on almost the entire surface of the sample, the noise increases in proportion to the pixel size.

On the other hand, the inspection time increases as the number of pixels increases, so that it is necessary to increase the pixel size in order to realize high-speed inspection. Therefore, it is necessary to increase the pixel size and reduce the noise level. As a method of reducing this noise level, Koizumi et al., "Analysis of Light Reflected from LSI Wafer Pattern", Transactions of the Society of Instrument and Control Engineers, 17-2, 77/82 (1981), analyze the method using polarization. Have been. According to this, scattered light (noise) from a pattern is obtained by using polarized light.
Can be attenuated. However, the attenuation rate of the scattered light by this method depends on the direction of the detector as analyzed in the above-mentioned paper. For this reason, when condensing light emitted in various directions as in the case of using an imaging optical system, the respective attenuation rates are integrated to obtain an attenuation rate of 0.1% to 0.01%.
About.

On the other hand, in the method using the spatial filter according to the present invention, the attenuation rate is reduced from 0.001% to 0.0001.
%. The reason will be described with reference to FIGS. The wafer 2001 on which the repetitive pattern is formed is illuminated with the illumination light 2002, and the illuminated area is set in the lens system 200.
3 and 2005, an image is formed on the detector 2006. Here, FIG. 66 shows the intensity distribution of light emitted from the pattern on the Fourier transform surface on which the spatial filter 2004 is mounted.
Light emitted from the repetitive pattern is concentrated at a position corresponding to the pitch of the pattern. As an example of calculating this concentration ratio, the diffracted light intensity distribution in the case of a double slit is written by Hiroshi Kubota,
This is described in "Applied Optics" (Iwanami). According to this, when the number of slits (in the present application, the number of repetitive patterns that are simultaneously illuminated) increases, the concentration ratio increases. This ratio can also be calculated using the Fourier transform F []. Assuming that the shape of the illuminated pattern is a (x, y), the light intensity distribution at the position of the spatial filter is F [a (x, y).
y)]. Assuming that the shape of the spatial filter is p (u, v), p (u, v) * F [a (x, y)] is light passing through the spatial filter. If the shape of the figure complementary to the spatial filter is ￣p (u, v), then ￣p
(U, v) * F [a (x, y)] is a light component that is shielded by the spatial filter. The ratio of these two components becomes the previous attenuation rate. When this attenuation rate is calculated when the number of pattern repetitions is 3, it is about 0.001%. When the number of repetitions is 5, it becomes about 0.0001%, and when the number of repetitions is further increased, the attenuation rate decreases. Therefore, the attenuation rate can be reduced as compared with the case of using polarized light, and the pattern noise can be reduced.

The above calculations are for the case where the pattern shape and other conditions are ideal, and may not always agree with actual experimental results. However, experimental results have shown that the attenuation rate is reduced by one to three orders of magnitude compared to the polarization method, and pattern noise can be reduced.

Next, another embodiment of the small foreign matter monitor of the present invention will be described with reference to FIGS. FIG. 64 shows the relationship between the detection pixel size of the foreign object detector and the noise level. High-speed and miniaturization are issues for small foreign matter monitors. FIG. 1A shows a foreign matter detection optical system. The pattern on the wafer 2001 and the scattered light from the foreign matter are detected by the detector 2006 through the detection lens 2003. The detection signal from the detector 2006 is output for each pixel of the detector 2006. FIG. 2B shows a case where the size of a pixel corresponding to one pixel of the detector 2006 on the wafer is a small pixel and a case where the size is a large pixel. The detection time T is expressed by the following (Equation 20) as the area S of the wafer, the data acquisition time t of the detector, the pixel size w of the detector, and the number n of pixels of the detector.

[0214]

T = (St) / (w · n) (Equation 20) From the expression (Equation 20), to realize high speed and small size, w
It is most effective to increase n and to increase n to perform parallel processing. However, as shown in FIG.
When w is increased, the noise level from the pattern on the wafer 2001 increases in proportion to w. Therefore, w
It is necessary to reduce the noise level from the pattern in order to increase the value and maintain the foreign matter detection performance.

Then, the effect of the noise reduction by the spatial filter method for reducing the noise level from the pattern will now be described. FIG. 65 shows a configuration diagram of a foreign matter detection optical system using a spatial filter. Detection lens 2003
Is provided with a spatial filter 2004 on the Fourier transform plane. Diffractive light 2007 from the repeatable memory pattern on the wafer 2001, which is noise,
After passing through 03, the light is shielded by the spatial filter 2004. Further, the scattered light 2008 from the foreign matter on the wafer 2001 passes through the detection lens 2003, the spatial filter 2004, and the imaging lens 2005, and is detected by the detector 2006.

Here, FIG. 66 shows the light intensity distribution in the x direction of the pattern diffraction light 2007 on the surface of the spatial filter 2004 in FIG. In the figure, a spatial filter 2004
The ratio of the light intensity of the pattern diffracted light 2007 corresponding to the transmission part (A) and the light shielding part (B), that is, A: B is 1:10 ⁵
By installing the spatial filter 2004,
Pattern noise can be reduced to 1/10 ⁵ .
In the polarization filter method used in the conventional foreign matter inspection device,
Since the pattern noise reduction is 1/10 ² , the noise reduction level is 10 ³ if the pixel size of the detector is the same.
This improves the foreign matter detection sensitivity. Therefore, by setting the target setting of the foreign substance detection sensitivity to be equal to or less than the performance of the conventional foreign substance inspection apparatus, it is possible to increase the pixel size of the detector, and it is possible to make the foreign substance detection optical system faster and smaller. .

The pattern that can be shielded from light by the spatial filter is a repetitive memory pattern, and the area other than the memory pattern portion is set as invalid data or a detection prohibited area by software or the like.

FIG. 67 shows the discrimination ratio in the foreign matter detection optical system when the spatial filter method is applied. Here, since the detection lens 2003 in the optical system of the foreign matter detection system also serves as an imaging lens, no imaging lens is required. Detector 2
From the detection signal distribution from 006, the detection signal of the foreign substance is S,
If the pattern noise is N, the discrimination ratio is represented by S / N. Next, FIG. 68 shows the relationship between the pixel size of the detector and the discrimination ratio. Here, an example of 2 μm standard particles is shown as the foreign matter.
In order to stably discriminate foreign matter from a pattern, a discrimination ratio of 1 or more is required. Accordingly, it can be seen from the figure that the pixel size of the detector should be 20 μm or less in order to detect the 2 μm standard particles from the pattern and detect them.

Next, FIG. 69 shows an illumination area and a detection area. The inspection time T is expressed by the following equation (Equation 21) as an inspection width Lx, Ly, a detector pixel size w, and a detector read clock frequency f.

[0220]

T = (Lx · Ly / w ² ) · (1 / f) (Equation 21) The effective illumination light intensity P is the illumination power P ₀ , the illumination width W
x and Wy are represented by the following (Equation 22).

[0221]

P = P ₀ · (w · Lx) / (Wx · Wy) (Equation 22) Here, since Wx ≒ Lx, the expression (Equation 22) becomes
3) It is shown by the equation.

[0222]

P = P ₀ · (w / Wy) (Equation 23) The total illumination light amount Pt is obtained from the equations (21) and (23).
(Equation 24) is shown by the formula.

[0223]

Equation 24] _{Pt = T · P = P 0} · (Lx · Ly / Wy) · (1 / (w · f)) = K 1 · (1 / (w · f)) ( Equation 24) Therefore, the detection The signal strength I is expressed by the following equation (25) based on the foreign substance signal coefficient K ₂ and the equation (24).

[0224]

I = K ₂ · Pt = K ₁ · K ₂ · (1 / (w · f)) (Equation 25) From the equation (Equation 25), I is a function of w · f.

FIG. 70 shows a performance chart for determining the device specifications based on the above results. The device specification is determined based on three figures, that is, the relationship between the pixel size and the inspection time, the relationship between the pixel size and the discrimination ratio, and the relationship between the pixel size and the clock frequency of the detector and the detection signal. For example, if the clock frequency of the detector is set to 2 MHz from the relationship between the pixel size and the inspection time in order to realize an inspection time of 20 seconds, the pixel size of the detector may be 13 μm. At this time, from the relationship between the pixel size and the discrimination ratio, the discrimination ratio from the pattern of the 2 μm foreign matter is 2, and it is possible to discriminate from the pattern. Finally, from the relationship between the pixel size and the clock frequency of the detector and the detection signal, the detection signal of a 2 μm foreign substance is determined by the pixel size and the clock frequency, is 60 mV, and can be detected by the detector.
As described above, it is possible to arbitrarily determine the specification of the detected foreign matter size and the inspection time of the apparatus from the three performance diagrams.

FIG. 71 is a diagram showing a device configuration of a foreign matter detecting optical system using the spatial filter method. The foreign object detection optical system
Product wafer 2 by uniaxial scanning 2010 of product wafer 2001
001 is configured to be inspectable. for that reason,
The foreign object detection optical system includes an illumination optical system 2011 and a detection optical system 20.
13 and each has a unit configuration. The case where the wafer to be inspected has a diameter of 200 mm will be described below. For example, in order to inspect the entire width of the wafer 2001 with eight units, the illumination area and the detection area 201 of one unit are required.
2 may be 25 mm. Therefore, when the wafer to be inspected has a diameter of 150 mm, six of the eight units may be used. One unit of detection optical system 2013
Comprises a detection lens 2014, a spatial filter 2015 installed on the Fourier transform surface of the detection lens 2014, and a linear sensor 2016 as a detector. When the outer dimensions of the detection lens 2014 are larger than the detection width, the entire width of the wafer 2001 can be secured by disposing the detection lenses 2014 in a zigzag shape as shown in FIG. When the outer dimensions of the detection lens 2014 are equal to or smaller than the detection width, or when the inspection on the wafer is limited, that is, in the case of the partial inspection, the detection lenses 2014 can be linearly arranged. The spatial filter 2015 used here uses two sets of four-unit configurations when the detection optical system 2013 is in a zigzag shape.
When 13 is linear, one set of 8 units is used.
The detection signal from the linear sensor 2016 is processed in a foreign matter detection process (separate body) 2017 and is output as foreign matter data.

When the detection optical system 2013 is in the shape of a dot, two sets of filters need to be exchanged. One type of wafer can be handled by one type of spatial filter 2015 without depending on the process.

Next, an example of the specifications of this embodiment will be described.
The illumination optical system has a wavelength of 780 nm and an output of 2 as an illumination light source.
The area of 26 × 1 mm ² on the wafer is illuminated at a light incident angle of 60 ° from above using a semiconductor laser of 00 mW. The detection optical system includes a projection lens (50
mmF2.8, detection magnification 1 × (detection NA = 0.
Detect in 1). The detector includes a CCD linear sensor having a pixel size of 13 μm, a number of pixels of 2048, and a driving frequency of 4 MHz, or a pixel size of 7 μm having a high foreign matter discrimination performance.
A CCD linear sensor having 4096 pixels and a driving frequency of 4 MHz is used.

FIG. 72 is an example showing the effect of the pattern noise light on the wafer rotation angle. Wafer 200
When 1 rotates, the diffracted light from the pattern of the wafer 2001 also rotates according to the wafer 2001. Therefore, when the wafer 2001 is rotated with respect to the foreign matter detection optical system 2021, diffracted light from the pattern of the wafer 2001 leaks from the light blocking part of the spatial filter of the foreign matter detection optical system 2021. Therefore, the leakage light of the diffracted light from the pattern, that is, the pattern noise light, is a function of the light blocking width of the spatial filter and the rotation angle of the wafer. Here, the rotation angle θ of the wafer represents the angle between the center line 2020 of the foreign matter detection optical system 2021 and the center line 2000 of the wafer 2001. However, if the light shielding width of the spatial filter is increased, the scattered light from the foreign matter is also reduced, so it is necessary to find the optimum width. Therefore,
In the conventional pre-alignment apparatus, the rotation angle of the wafer is ±
Since it can be suppressed within 2 °, foreign matter detection performance, for example, an example of a change in pattern noise light due to rotation of the wafer when the light shielding width of a spatial filter capable of detecting a 2 μm foreign matter from a pattern is set to an optimum width is the same. Shown in the figure.

In order to have a function as a monitor for foreign substance inspection, a foreign substance detection system having a depth of focus as deep as possible is required.

The value of the depth of focus can be larger than the depth of focus calculated from the NA of the detection lens depending on the size of the detection pixel.

If the size of the detection pixel is sufficiently smaller than the size of the foreign matter to be detected, the depth of focus d depends on the numerical aperture of the detection lens, and is expressed by the following equation (26) as the wavelength λ of light and the numerical aperture NA of the detection lens. It is.

[0233]

D = 0.5 · λ / (NA) ² (Expression 26) In the expression (11), for example, λ = 780 nm, NA = 0.1
In this case, d = 39 μm. If the detection pixel size is sufficiently larger than the detected foreign matter size, the depth of focus depends on the detection pixel size. In this case, assuming that the detection pixel size is the equivalent resolution a ′, the relationship with the equivalent numerical aperture NA ′ is expressed by the following (Equation 27).

[0234]

A ′ = 0.61 · λ / NA ′ (Expression 27) Further, when NA ′ in Expression (27) is substituted for NA in Expression (26), the actual depth of focus d can be obtained. For example, if a ′ = 13 μm, NA ′ = 0.037, and d
= 285 μm.

Therefore, there is an effect of increasing the depth of focus of the foreign matter detection system by increasing the number of pixels of the detector.

FIG. 73 is an example diagram showing a change in the foreign matter detection output depending on the height of the wafer stage. λ = 780nm, NA
= 0.1, the change in the detection output of a 5 μm foreign substance when a detection pixel size of 13 μm is used. As shown in the figure, the depth of focus is ± 70 μm. This value is a value between the value obtained from the numerical aperture of the detection lens (39 μm) and the value obtained from the detection pixel size (285 μm). Therefore, 13μ
Although the detection pixel size of m is not large enough for a 5 μm foreign substance, the depth of focus is increased.

As described above, by reducing the numerical aperture of the detection lens and increasing the size of the detection pixels, the depth of focus can be increased, and the position control of the wafer transfer system in the height direction can be roughly performed. It is possible to

Next, the structure of one unit of the illumination optical system used in the small foreign matter monitoring device will be described. On the wafer, one side is widened so that the inspection area can be sufficiently illuminated, and one side is narrowed down to have sufficient illuminance, so that linear illumination can be performed. If the illumination light source is a point light source, a plane wave, that is, a parallel light beam is generated on both sides. Here, if the illumination light is parallel light, the image at the position of the spatial filter of the detection optical system can be sharpened, and the light shielding performance of the pattern by the spatial filter and the foreign matter detection performance can be enhanced. However, for example, when a small semiconductor laser is used as the illumination light source, the length of one side of the light emitting point becomes longer as the output becomes higher. Therefore, one side cannot generate a plane wave, that is, a parallel light beam. Accordingly, two types of embodiments of the illumination optical system corresponding thereto will be described. However, in the linear illumination on the wafer, the long direction of the beam is the y direction, and the short direction of the beam is the x direction.

FIG. 74 shows the configuration of the first system, FIG. 74A shows the configuration viewed from the x direction, and FIG. 74B shows the configuration viewed from the y direction. Here, the semiconductor laser 2101
The long direction of the light emitting point 2100 is the x direction, and the light emitting point 2100
Is short (close to the point light source) is the y direction. However,
In the case of P-polarized illumination on the wafer, a λ / 2 plate is inserted so as to be S-polarized illumination.

The x direction in FIG.
The light emitted from the lens No. 01 is a lens 2102 to a lens 2106
To illuminate the wafer 2001 by focusing the light beam. In the y direction of FIG. 13B, the light emitted from the semiconductor laser 2101 is expanded into a parallel light by using the lenses 2102 to 2106. According to this method, the luminous flux in the x direction can be easily narrowed down, so that the illuminance of the illumination can be increased. In this method, since the light beam in the x direction is narrowed down by a certain angle instead of parallel light, the diffraction pattern in the x direction on the spatial filter surface of the detection optical system becomes long, but a linear spatial filter as shown in FIG. 62 is used. Thereby, the diffracted light from the pattern can be shielded. FIG.
5 shows an example of a plan view of a diffraction pattern on a wafer on the spatial filter surface of the detection optical system when the illumination optical system of FIG. 74 is used. The size of one point of the diffraction pattern from the pattern on the wafer depends on the numerical aperture of illumination in the x direction, x1 = several mm, and y1 = several μm since the y direction is parallel light.
And sharp light is obtained only in the y direction. Depending on the direction of the wafer, the pitch py in the y direction is x as shown in FIG.
If the pitch is smaller than the pitch px in the direction, the light blocking ratio of the spatial filter is increased, and the detection output from foreign matter is also reduced. Then, by rotating the wafer by 90 °, the diffraction pattern from the pattern on the wafer becomes as shown in FIG. 2B, and the pitch in the y direction is the same as px in FIG. Can improve the light shielding performance. As described above, by setting the orientation of the wafer in advance so that the longer pitch of the diffraction pattern comes in the y-direction, the detection output from foreign substances can be further improved.
The optimum orientation of the wafer can be input in advance as data. Further, the direction of the diffraction pattern is once checked to detect the optimal direction of the wafer, and thereafter, the upper part of the optimal direction is used.

FIG. 76 shows the configuration of the second method, FIG. 76A shows the configuration viewed from the x direction, and FIG. 76B shows the configuration viewed from the y direction. Here, the semiconductor laser 2101
The shorter (closer to the point light source) direction of the light emitting point 2100 is the x direction, and the longer direction of the light emitting point 2100 is the y direction. However, in the case of P-polarized illumination on the wafer, a λ / 2 plate is inserted so as to be S-polarized illumination.

The x direction in FIG.
The light emitted from the lens No. 01 is a lens 2202 to a lens 2207.
And squeezes the light beam into parallel light. The light emitted from the semiconductor laser 2101 in the y direction in FIG.
202 to spread the light beam using the lens 2207
01 is illuminated. However, since the length of the light emitting point 2100 in the x direction is as long as several tens of μm, the light cannot be made into parallel light. Here, the light source 2100 includes a lens 2202 and a lens 2
207, spatial filter 20 through imaging lens 2014
An image is formed at position 15. Since the total imaging magnification is optimally several tens μm or less from the light shielding performance of the spatial filter, it is set to be about 1 ×.

FIG. 77 shows an example of a plan view of a diffraction pattern from a pattern on a wafer on the spatial filter surface of the detection optical system when the illumination optical system of FIG. 76 is used. The size of one point of the diffraction pattern from the pattern on the wafer on the spatial filter surface is x since the x direction is parallel light.
2 = about 100 μm, and the y direction is proportional to the size of the illumination light source, so y2 = several tens of μm, and relatively sharp light can be obtained in both the x and y directions regardless of the orientation of the wafer.
The light shielding performance of the spatial filter can be improved.

Next, another embodiment of the optical system for detecting foreign matter by the polarization detecting method of the small foreign matter monitor of the present invention will be described with reference to FIGS. 78 to 79.
This will be described with reference to FIG.

The polarization detection method is not limited to the memory pattern, and it is possible to detect foreign substances by discriminating from all patterns on the entire surface of the wafer.

FIG. 78 shows a configuration diagram of a foreign matter detection optical system using a refractive index change type lens array as a detection lens.
It comprises an oblique illumination optical system 3002 and a detection optical system 3003. The oblique illumination optical system 3002 comprises one or more illumination arrays as shown in the figure. The detection optical system 3003 includes a refractive index variable lens array 3004 as a detection lens, a polarizing plate 3005 as a polarizing element, and a detector 3006 at an image forming position of the refractive index variable lens array 3004. The illumination array is linearly illuminated to illuminate the entire width of the wafer to detect the full width of the wafer. Therefore, the wafer 300
The entire surface of the wafer 3001 can be inspected by one uniaxial scan 3010. The illumination angle of the illumination array 3002 is set to be several degrees above the horizontal direction, and the wafer 3001 is irradiated with linearly polarized light (S-polarized light) such that the magnetic field vector is perpendicular to the plane of incidence of the illumination. Further, the scattered light from the pattern and the foreign matter on the wafer 3001 passes through the lens array 3004 of the refractive index change type, and then only the P-polarized light (the linearly polarized light whose magnetic field vector is parallel to the incident surface of the illumination) by the polarizing plate 3005. , The scattered light from the pattern is reduced and the scattered light from the foreign matter is emphasized, and is detected by the detector 3006.

FIG. 79 is a view showing the arrangement of a foreign matter detection optical system using a normal lens as a detection lens. The foreign matter detection optical system is configured so that the entire surface of the product wafer 3001 can be inspected by uniaxial scanning 3110 of the product wafer 3001. Therefore, the foreign matter detection optical system is divided into an illumination optical system 3111 and a detection optical system 3113, and each has a unit configuration. The case where the wafer to be inspected has a diameter of 200 mm will be described below. For example, in order to inspect the entire width of the wafer 3001 with eight units, the illumination area and the detection area 3112 of one unit may be set to 25 mm. Therefore, when the wafer to be inspected has a diameter of 150 mm, six of the eight units may be used. The detection optical system 3113 of one unit includes a detection lens 3114, a polarizing plate 3115, and a linear sensor 3116 as a detector. When the outer dimensions of the detection lens 3114 are larger than the detection width, the entire width of the wafer 3001 can be secured by disposing the detection lenses 3114 in a zigzag shape as shown in FIG.
Further, when the outer dimensions of the detection lens 3114 are smaller than the detection width, or when the inspection on the wafer is limited, that is, in the case of the partial inspection, the detection lenses 3114 can be linearly arranged. The illumination unit 3111 illuminates the wafer 3001 with linearly polarized light (S-polarized light) such that the magnetic field vector is perpendicular to the plane of incidence of the illumination. Further, the scattered light from the pattern on the wafer 3001 and the foreign matter passes through the detection lens 3114 and is then reflected by the polarizing plate 3115.
The linear sensor 3116 detects only the polarized light (linearly polarized light whose magnetic field vector is parallel to the incident surface of the illumination), reduces the scattered light from the pattern, and emphasizes the scattered light from the foreign matter. The detection signal from the linear sensor 3116 is processed in a foreign matter detection process (separate body) 3117 and is output as foreign matter data.

FIG. 80 shows the position and function of the present invention.
One of the main tasks in the mass production start-up of LSIs is to investigate the cause of foreign matter occurrence and take countermeasures. It will be a great clue. On the other hand, in the mass production line,
It is necessary to detect these foreign substances promptly and take countermeasures.
When the time from the generation of a foreign object to the detection of the foreign object has elapsed, the number of occurrences of defects increases and the yield decreases. Therefore,
In order to maintain a high yield, it is indispensable to reduce the elapsed time from the generation of foreign matter to its detection. In addition, a strict experiment for detecting the number of foreign particles on a wafer has revealed that the number of foreign particles does not gradually increase or decrease but suddenly increases or decreases. FIG. 7A shows a time transition of the number of foreign substances on a product wafer generated in a processing apparatus such as a CVD. FIG. 1B shows a conventional system. The conventional apparatus is of a stand-alone type, and is a sampling inspection in which a wafer processed on a mass production line is brought to a location of an inspection apparatus to inspect for foreign matter.
Therefore, since it takes time to transport the wafer and inspect the foreign matter, the frequency of the inspection, ie, sampling, is one lot, several lots, or one lot as shown in FIG.
There was one sheet per day, and there was a limit to the number of sheets inspected. In such a sampling, a sudden increase in foreign matter is overlooked or detected after a while with the increase, and a considerable number of defects (spot defects) occur. That is, in such sampling, it cannot be said that the generation of foreign matter was detected sufficiently early. Therefore, as shown in FIG. 4C, a small foreign substance monitor, which is a compact foreign substance monitoring device, is placed in the input / output port of the processing apparatus or in the transport system between the processing apparatuses, and the foreign substance data from the small foreign substance monitor is read. By taking in the foreign substance management system, the foreign substance management can be performed on a single wafer. Therefore, by using the small foreign substance monitor, as shown in FIG. 3A, the sampling time of the monitor can be shortened, real-time sampling of a sheet can be performed, and the effect of the foreign substance inspection can be maximized. it can.

The functions of the present invention include the following five items. A size that can be attached to the transport system of the processing apparatus, that is, a small size, enables high-speed inspection capable of single-wafer inspection of wafers, and can be an option of the processing apparatus so that foreign substances can be managed for each processing apparatus. Inexpensive price. Also,
The monitor is easy to set up and maintenance-free. Hereinafter, an embodiment of the spatial filter will be described with reference to FIGS. This spatial filter may be configured using a liquid crystal display element, but in the case of a liquid crystal element, only light having a specific polarization direction can be used. Further, there is a problem that the diffracted light from the pattern cannot be sufficiently shielded because the light attenuation rate is small. Therefore, it is preferable that the spatial filter is mechanically configured using a metal plate or the like.

The spatial filter is composed of a set of linear patterns as described with reference to FIGS. (Of course, the spatial filter is desirably a set of points that are slightly larger so as to block the set of points as shown in FIG. 75 (a), but a set of straight lines as shown here is sufficient. There is also an effect that the function is fulfilled and the configuration is simple.) The pitch and phase of this linear pattern may be matched. FIG. 82 shows an embodiment of a variable pitch spatial filter 1270 using this metal plate.

In this embodiment, the illumination optical system 1110, the detection optical system 1210, the stage system 1300, and the signal processing system 14
01 and the data processing system 1501
This is the same as the embodiment shown in FIG. Here, the semiconductor laser 1
When the exit 1021 of 111 is vertically arranged as shown in FIG. 82, the linear direction of the spatial filter is parallel to the incident surface of the illumination light flux as shown in FIG. 82 by using the illumination system of FIG. Become. In this case, it is only necessary to adjust the pitch of the linear pattern with reference to the linear pattern at the center of the spatial filter as the alignment of the spatial filter. In this case, the pitch variable mechanism of the spatial filter can be simply configured.

The variable pitch spatial filter 127 shown in FIG.
83 is shown in FIG. Pitch variable spatial filter 1
Reference numeral 270 denotes a plurality of linear patterns 1271, spring-like supports 1272, and supports 127 formed of a material having a high light-shielding rate, such as metal, metal oxide, or plastic.
3, fixing means 1274, screw 1275, screw driving means 1
276. Here, the screw 1235 has
A right screw is formed at 1277 and a left screw is formed at 1278. Here, the screw 127 is operated by the screw driving means 1276.
By rotating 5, the pitch between the linear patterns 1271 can be changed. This screw driving means 12
The drive of 76 is controlled according to the calculated value of the pitch between the linear patterns 1271 by receiving the cell pitch d in the chip at the same time as the chip pitch p on the wafer when the wafer is loaded. Here, the spring-like support 1272
May be rubber.

Here, this spatial filter 1270
Is difficult to change over a wide dynamic range. For example, if the pitch is reduced to 1/10, the screw 127
No. 5 is necessary because the length required for the spatial filter is ten times as long. Therefore, the spatial filter 127
In the case where a plurality of zeros are installed in a superposed manner and the pitch is to be changed small, the pitch can be changed by the variable mechanism while being superimposed. . Of course, if necessary, it can be shifted from the variable mechanism at the same time.

Here, it is desirable that the linear pattern 1279 at the center of the spatial filter 1270 is configured to be thicker than other linear patterns. This is because the diffracted light at the center, that is, the 0th-order diffracted light has a high light intensity and a wide width of the intensity distribution of the diffracted light, and therefore requires a wide linear pattern to sufficiently shield the diffracted light. is there.

Although one embodiment of the driving mechanism has been described here, the present invention is not limited to this embodiment, and the linear pattern 1 having a high light-shielding property is not limited to this embodiment.
Other drive mechanisms may be used as long as they drive the 271. Specifically, a configuration as shown in FIG. 84 may be used. In this embodiment, the linear pattern 1271 is supported by the link 1291 and the link driving mechanism 1292
Thus, the pitch is changed by changing the inclination of the link 1291.

It is more preferable to set the orientation of the wafer in a direction in which the pitch of the spatial filter can be increased, that is, in a direction in which the pitch d of the pattern on the wafer is small.

As shown in FIGS. 85 and 86, the illumination optical system 1
When the optical system shown in FIG. 75 is used as 110, a slightly larger linear spatial filter 1279 is arranged in the center of the spatial filter in parallel to the incident surface of the illumination, and a linear pattern is arranged perpendicular to this. There is a need. in this case,
It is necessary to adjust the pitch and phase for the spatial filter alignment. If the incident angle of the illumination is α, the emission angle of the linear diffraction pattern is θn, the wavelength of the illumination light is λ, and the basic pitch of the pattern on the wafer is d, the following equation (28) holds.

[0258]

Sin (α−θn) = n · λ / d (Equation 28) Therefore, it is necessary to construct a variable mechanism that satisfies the expression (Equation 28). Specifically, the pitch variable spatial filter 1270 shown in FIG. 85 is arranged in a direction rotated by 90 degrees, and in addition to pitch adjustment, the entire pitch variable spatial filter 1270 is moved in a direction perpendicular to the linear pattern 1271. This adjusts the phase. This phase adjustment is performed by the phase adjusting means 1281. In this configuration, a light shielding plate that is slightly thicker, specifically, about 1 to 3 times the size of the linear pattern, may be arranged at the center position of the spatial filter in parallel with the light incident surface.

The thickness of the linear pattern is preferably determined experimentally, but is designed to be 10% to 20% larger than the size of the image on the spatial filter of the light source 1111 of the illumination system. Should. However, when considering the accuracy of the adjustment mechanism of the spatial filter, it is necessary to provide a larger margin. Also, although the configuration of FIG. 53 is described using six channels in parallel, the configuration is not limited to six channels and is determined by specifications such as the size of the wafer and the inspection time.
Here, the spatial filter mechanism in the case where the illumination optical system is configured by the optical systems shown in FIGS. 74 and 75 has been described. However, even if another illumination system (not described) is used, the mechanical By using the filter, the light blocking ratio can be improved, so that the diffracted light from the pattern can be efficiently blocked, and the detection sensitivity of foreign substances can be improved.

If the pitch and width of the spatial filter are to be further reduced, the precision is insufficient with the mechanical configuration shown here. In this case, a variable spatial filter can be made using the method introduced as "micromechanism".

In the above configuration, the pitch of the spatial filter can be automatically changed by receiving the data such as the pitch between the chips and the cell pitch of the product. Having. A spatial filter may be created for each product, and this spatial filter may be automatically replaced. An example is shown in FIG. This method uses the detector 1 shown in FIG.
The three filters 254, 1234, and 1214 are installed on one substrate and are replaced.

[0262]

According to the present invention, since repetitive information can be omitted at a high speed, a defect such as a foreign substance existing as non-repeated information can be detected at high speed from a repetitive pattern. , All the wafers passing through the line can be inspected, and an increase in foreign substances can be detected in real time. As a result, the production of a large number of defective products due to the generation of foreign matters can be prevented beforehand, and the yield can be improved. That is, according to the present invention, by monitoring foreign substances only with a simple and compact monitoring device in a mass production line of a semiconductor manufacturing process, the production line can be reduced in weight and the production cost can be reduced, and the monitoring device can reduce foreign substances. Since the inspection can be performed in real time, the production of defects can be minimized, which can greatly contribute to improving the yield of products. Further, according to the present invention, by separating a foreign substance inspection system at the time of mass production start-up in a semiconductor manufacturing process and a mass production line, and by using a high-precision foreign substance inspection apparatus, foreign substances required at the time of mass production start-up can be obtained. Since the functions of detection, analysis, and evaluation can be maximized, the feedback to the mass production line can be smoothly performed, and the mass production start-up period can be shortened. High frequency sampling can be realized, and high quality and high yield production of products can be realized.

[Brief description of the drawings]

FIG. 1 is a structural block diagram showing an embodiment of the device according to the present invention.

FIG. 2 is a configuration block diagram illustrating an example of a detection head illustrated in FIG. 1;

FIG. 3 is a perspective view showing a spatial filter mechanism shown in FIG. 1;

FIG. 4 is a block diagram illustrating a configuration of an operator processing unit illustrated in FIG. 1;

FIG. 5 is a configuration block diagram showing a parameter detection system shown in FIG. 1;

FIG. 6 is a perspective view showing the rotation adjusting mechanism shown in FIG. 1;

FIG. 7 is a configuration block diagram showing a conventional method.

FIG. 8 is a configuration block diagram showing a basic concept of the present invention.

FIG. 9 is a configuration block diagram showing a pattern removing method of the present invention.

FIG. 10 is a configuration block diagram showing another embodiment of the parameter detection system according to the present invention.

FIG. 11 is a configuration block diagram of a detection lens according to the present invention.

FIG. 12 is a block diagram illustrating a configuration of a rotation deviation detection system according to the present invention.

FIG. 13 is a block diagram showing a configuration of another embodiment of the rotation deviation detection system according to the present invention.

FIG. 14 is a schematic view showing the effect of a comparative inspection using the optical filtering according to the present invention.

FIG. 15 is a perspective view showing a state when light is shielded by the spatial filter according to the present invention.

FIG. 16 is a schematic diagram illustrating an operator process according to the present invention.

FIG. 17 is a schematic diagram illustrating the shape of a spatial filter according to the present invention.

FIG. 18 is a schematic diagram illustrating conditions for pattern erasure according to the present invention.

FIG. 19 is a perspective view illustrating a method of manufacturing the spatial filter mechanism according to the present invention.

FIG. 20 is a perspective view illustrating a method of adjusting a coil spring in the spatial filter mechanism according to the present invention.

FIG. 21 is a structural block diagram for explaining a method of removing the influence of diffraction in the spatial filter mechanism according to the present invention.

FIG. 22 is a schematic diagram illustrating a scanning direction of a sensor according to the present invention.

FIG. 23 is a configuration block diagram of a configuration for realizing vertical illumination according to the present invention.

FIG. 24 is a perspective view showing how to use the detection head according to the present invention.

FIG. 25 is a perspective view showing how to use the detection head according to the present invention.

FIG. 26 is a perspective view showing how to use the detection head according to the present invention.

FIG. 27 is a schematic view illustrating a stage scanning method according to the present invention.

FIG. 28 is a block diagram showing a configuration of a pattern pitch measuring unit according to the present invention.

FIG. 29 is a block diagram showing a configuration of a pattern removing method using interference according to the present invention.

FIG. 30 is a diagram illustrating a signal processing method according to the present invention.

FIG. 31 is a block diagram showing a configuration of an apparatus for starting up mass production in a semiconductor manufacturing process and inspecting foreign substances on a mass production line, and an apparatus therefor, according to an embodiment of the present invention.

FIG. 32 is a plan view of a single-wafer CVD apparatus equipped with a foreign substance monitor according to an embodiment of the present invention.

FIG. 33 is a configuration diagram of a foreign matter monitor according to the present invention.

FIG. 34 is a diagram showing a detection method of a wafer rotation direction detector according to the present invention.

FIG. 35 is a diagram showing product wafer reference coordinates for foreign object coordinate management according to the present invention.

FIG. 36 is a diagram showing device-based coordinates for foreign object coordinate management according to the present invention.

FIG. 37 is a configuration diagram of a foreign matter detection optical system according to the present invention.

FIG. 38 is a configuration diagram of an oblique illumination optical system according to the present invention.

FIG. 39 is a diagram showing a detection width of the detection optical system according to the present invention.

FIG. 40 is a configuration diagram of a detector according to the present invention.

FIG. 41 is a configuration diagram of a spatial filter according to the present invention.

FIG. 42 is a detailed view of a spatial filter according to the present invention.

FIG. 43 is a configuration diagram of a spatial filter group by a dry plate method corresponding to a product wafer in each step according to the present invention.

FIG. 44 is a configuration diagram of an AND spatial filter using a dry plate method according to the present invention.

FIG. 45 is a configuration diagram of an optical system for detecting foreign matter by partial inspection according to the present invention.

FIG. 46 is a diagram showing a detection area of a foreign object detection optical system by a partial inspection according to the present invention.

FIG. 47 is a configuration diagram of a microlens-based foreign matter detection optical system arranged in two rows according to the present invention.

FIG. 48 is a diagram showing a spatial filter detection method based on wafer rotation when the lens array according to the present invention is used.

FIG. 49 is a configuration diagram of a foreign matter detection optical system using white light illumination according to the present invention.

FIG. 50 is a diagram showing foreign matter detection performance by white light illumination according to the present invention.

FIG. 51 is a configuration diagram of an optical system for detecting foreign matter by wafer comparison inspection according to the present invention.

FIG. 52 shows a semiconductor FA using the foreign substance monitor according to the present invention.
FIG.

FIG. 53 is a block diagram showing one embodiment of a foreign matter inspection device according to the present invention.

FIG. 54 is a side view of the illumination optical system viewed from the x direction in FIG. 53;

FIG. 55 is a side view of the illumination optical system viewed from the y direction in FIG. 53;

FIG. 56 is a diagram showing one embodiment of the imaging lens in FIG. 53;

FIG. 57 is a diagram showing one embodiment of the imaging lens in FIG. 53;

FIG. 58 is a plan view showing an arrangement of the optical system shown in FIG. 53;

FIG. 59 is a plan view showing an on-wafer pattern according to the present invention.

FIG. 60 is a plan view showing a diffraction pattern according to the present invention.

FIG. 61 is a diagram showing a refractive index variable lens according to the present invention.

FIG. 62 is a plan view showing a spatial filter according to the present invention.

FIG. 63 is a diagram showing an example of detecting a foreign substance according to the present invention.

FIG. 64 is a diagram illustrating a relationship between a detection pixel size and a noise level according to the present invention.

FIG. 65 is a configuration diagram of a foreign matter detection optical system using a spatial filter according to the present invention.

FIG. 66 is a diagram showing a light intensity distribution on a spatial filter surface according to the present invention.

FIG. 67 is a diagram showing a discrimination ratio in the foreign matter detection optical system according to the present invention.

FIG. 68 is a diagram showing a relationship between a detection pixel size and a discrimination ratio according to the present invention.

FIG. 69 is a diagram showing an illumination area and a detection area according to the present invention.

FIG. 70 is a performance diagram for determining device specifications according to the present invention.

FIG. 71 is a diagram showing a device configuration of a foreign matter detection optical system according to the present invention.

FIG. 72 is an example showing the effect of the pattern noise light according to the present invention on the wafer rotation angle.

FIG. 73 is an example diagram showing a change in foreign matter detection output according to the height of the wafer stage according to the present invention.

FIG. 74 is a side view of a lighting unit according to the present invention.

FIG. 75 is a plan view of a diffraction pattern on a spatial filter surface according to the present invention.

FIG. 76 is a side view of the lighting unit according to the present invention.

FIG. 77 is a plan view of a diffraction pattern on a spatial filter surface according to the present invention.

FIG. 78 is a configuration diagram of an optical system for detecting foreign matter by polarization detection according to the present invention.

FIG. 79 is a diagram illustrating the configuration of an optical system for detecting foreign matter by polarization detection according to the present invention.

FIG. 80 is a diagram showing the positioning and functions of the present invention.

FIG. 81 is a block diagram illustrating an embodiment of a signal processing system according to the present invention.

FIG. 82 is a configuration diagram showing one embodiment of the present invention using the variable spatial filter according to the present invention.

FIG. 83 is a specific configuration diagram of the variable spatial filter in the case shown in FIG. 82;

FIG. 84 is another specific configuration diagram of the variable spatial filter in the case shown in FIG. 82.

FIG. 85 is a configuration diagram showing another embodiment using the variable spatial filter according to the present invention.

86 is a specific configuration diagram of the variable spatial filter in the case shown in FIG. 85.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Substrate (wafer) 101 ... Detection head (detection optical system) 102 ... Illumination means 103 ... Detection optical system
105: rotation adjustment mechanism 106: spatial filter unit 107: detector 108: rotation detection means
110, 111: Fourier transform lens, 112: semiconductor laser, 113: collimator lens, 114: concave lens, 115: receiver lens, 116: cylindrical lens, 118: mirror, 119, 120 ...
Coil spring support, 121, 122 ... Coil spring, 1
23, 124: detector for detecting rotation, 125: guide, 126: screw, 127: right-hand thread, 128: left-hand thread, 129, 130: worm gear, 140 ...
Motor, 141 ... multiple linear spatial filters, 15
DESCRIPTION OF SYMBOLS 1 ... Rotation guide, 152 ... Rotating bar, 153 ... Spring, 154 ... Piezo element, 155 ... Piezo element controller, 156 ... Base, 201 ... Op amp,
202: A / D converter, 212: pitch detection means,
203: operator processing system, 204: cut-out means, 206: foreign matter data memory, 208: pattern memory, 209: parameter transmission means, 210: software processing system, 211: foreign matter memory, 214: 4-pixel adding means, 215 ... 8 values 216: Multiple line memories, 217: Buffer memory, 218: Judgment pixel extraction means, 219, 231: Operator extraction means, 220: Foreign matter comparison circuit, 221: Threshold setting circuit, 224: OR circuit, 226: AND circuit,
212: pitch detection means, 241: operator pitch calculation means, 242: FFT circuit, 243: spatial filter control system, 244: filter pitch calculation means,
232 ... Coordinate data creation means 229 ... Microcomputer 230 ... Display means 510 ... Semiconductor manufacturing equipment group 520 ... Sensing section 524 ... Vacuum foreign substance monitor 530 ... Utility group 540 ... Sampling section 550 ... Detection section , 560 ... Analysis unit, 56
3 ... STM / STS, 570 ... Compatible system, 58
0: Mass production start-up of semiconductor manufacturing process and mass production line foreign matter inspection system 581: Online foreign matter inspection system 582 Offline foreign matter inspection system 31
01: Foreign matter monitor, 3111: Product wafer, 312
DESCRIPTION OF SYMBOLS 1 ... Wafer rotation direction detector 3122 ... Foreign matter detection optical system 3123 ... Foreign matter information processing system 3124 ... Device stop function 3128 ... Foreign matter analysis system 3151 ...
Oblique illumination optical system, 3152 ... detection optical system, 3153
... Lens array, 3154 ... Spatial filter, 315
5 Detector 3201 Spatial filter group 3221
... And spatial filter, 3231 ... Micro lens group, 3280 ... Image processing system, 1110 ... Illumination optical system, 1210 ... Detection optical system, 1410 ... Signal processing system, 1211,1221 ... Imaging lens, 1212,
1222 ... Spatial filter

──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Yukio Mibo 292 Yoshida-cho, Totsuka-ku, Yokohama-shi, Kanagawa Prefecture Inside the Hitachi, Ltd. Production Technology Research Laboratory (72) Inventor Yoshimasa Oshima 292 Yoshida-cho, Totsuka-ku, Yokohama-shi, Kanagawa Address: Hitachi, Ltd., Production Technology Research Laboratories (72) Inventor Kazuhiko Matsuoka 111, Nishi-Yokote-machi, Takasaki-shi, Gunma Co., Ltd. Inside Hitachi, Ltd. Takasaki Plant (72) Inventor Yoshiharu 111, Nishi-Yokote-cho, Takasaki, Gunma, Japan (56) References JP 5-218163 (JP, A) JP 4-238255 (JP, A) JP 5-45862 (JP, A) JP 1-180067 (JP, A) JP-A-2-38951 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) G01N 21/84-21/958

Claims (17)

(57) [Claims] An illumination system for linearly irradiating a substrate having a repeating pattern with a different pitch with plane wave light, and an imaging optics for imaging a reflected light image from the substrate irradiated by the illumination system. System, a spatial filter installed in the middle of the imaging optical system so as to block the diffracted light from the small-pitch repetitive pattern on the substrate, and obtained through the spatial filter and imaged by the imaging optical system. And a detector for detecting the detected light image and an erasure for comparing and erasing signals generated based on a repetitive pattern having a large pitch on a substrate obtained through the spatial filter among signals detected by the detector. Means for detecting a defect on a substrate based on a signal obtained from the erasing means. 2. The defect detecting apparatus according to claim 1, wherein said imaging optical system is constituted by a lens array of a refractive index change type. 3. The defect detecting apparatus according to claim 1, wherein said image forming optical system comprises a plurality of Fourier transform lens groups, and at least an image side is telecentric. 4. The method according to claim 1, wherein the erasing means determines whether a signal of one pixel on the detector is defective or not.
The signal level of one pixel is compared with the signal level of a pixel at a corresponding location of an adjacent repetition pattern captured by the detector and the signal levels of a plurality of pixels adjacent to the corresponding location, and the corresponding When there is a pixel having a value equal to the signal level of the one pixel in the signal level of the location or a close location, the signal detected by one pixel on the detector is a signal from a repetitive pattern. 2. The defect detection apparatus according to claim 1, further comprising processing means for judging that a defect has occurred. 5. The pitch of the linear pattern of the spatial filter, the repetition pitch of the repetitive pattern light-shielded by the spatial filter being larger than the width of the corresponding pixel or the width of the region where adjacent pixels are combined. 5. The defect detection apparatus according to claim 4, wherein the defect detection apparatus is set as follows. 6. A Fourier transform means for performing a Fourier transform on the signal detected by the detector, and a pitch calculation for calculating a pitch of a Fourier transform image by a repetitive pattern formed on the sample from a calculation result by the Fourier transform means. 6. The defect detection apparatus according to claim 5, further comprising: means for changing a pitch of the spatial filter based on a result calculated by the pitch calculation means. 7. A pitch calculating means for calculating a pitch of a spatial filter in which a signal obtained from the detector takes a minimum,
7. The defect detection device according to claim 6, further comprising: a pitch changing unit that changes a pitch of the spatial filter so that the pitch is calculated by the pitch calculation unit. 8. A transport means for transporting a substrate having a repetitive pattern, an illumination system for linearly irradiating the substrate transported by the transport means with plane wave light, and a substrate illuminated by the illumination system. An imaging optical system that forms an image of the reflected light from the imaging optical system, and is provided at an imaging position in the middle of the imaging optical system so as to block the diffracted light from the repetitive pattern on the substrate, and the distance between each other is variable. A spatial filter composed of a plurality of linear light-shielding objects, a detector for detecting a light image obtained through the spatial filter and formed by the imaging optical system, and a signal detected by the detector. And a defect detecting means for detecting a defect on the substrate. 9. The defect detecting apparatus according to claim 8, wherein said image forming optical system comprises a plurality of Fourier transform lens groups, and at least an image side is telecentric. 10. A lens system according to claim 1, wherein said plurality of Fourier transform lens groups are composed of lens groups having different numerical apertures, and one of the Fourier transform lens groups is configured to be interchangeable with a lens group having a different numerical aperture. 9. The defect detection device according to item 9. 11. A measuring means for measuring the direction of the repeating pattern of the substrate, and a control for controlling the direction of the linear light-shielding member of the spatial filter according to the direction of the repeating pattern of the substrate measured by the measuring means. 9. The defect detection apparatus according to claim 8, further comprising: means. 12. The defect detecting apparatus according to claim 8, wherein said detector is constituted by a linear detector. 13. A defect detection device comprising an illumination system, a telecentric detection lens, and a one-dimensionally compressed spatial filter. 14. An illumination system having a linear spatial filter, an incident surface parallel to the linear spatial filter, and means for matching the length of the illumination surface with the length of the linear spatial filter and the repetition direction of the pattern on the substrate. A defect detection device comprising: 15. A defect detection apparatus characterized in that the zero-order diffracted light from each point on the object passes through the center of the pupil plane (the plane on which the spatial filter is installed) of the detection optical system. 16. A substrate having repetitive patterns having different pitches is irradiated with a plane wave of light in a straight line by an illumination system, and a reflected light image from the irradiated substrate is formed by an imaging optical system. A diffracted light from a small-pitch repetitive pattern on the substrate is shielded by a spatial filter in the middle of the imaging optical system, and a light image obtained through the spatial filter and imaged by the imaging optical system is detected by a detector. The detected signals are compared with each other, and the signals generated based on the repetitive pattern having a large pitch on the substrate obtained through the spatial filter are compared with each other by the erasing means, and are erased. And detecting a defect on the substrate based on the detected defect. 17. A substrate having a repetitive pattern is conveyed, and a plane wave light is radiated to the conveyed substrate in a straight line by an illumination system, and a reflected light image from the irradiated substrate is formed into an image forming optical system. A spatial filter formed by a plurality of linear light-shielding objects having a variable distance from each other and diffracted light from the repetitive pattern on the substrate is formed in the middle of the imaging optical system while passing through the spatial optical filter. And detecting a light image formed by the imaging optical system with a detector, and detecting a defect on the substrate based on the detected signal.