JP3904581B2 - Defect inspection apparatus and method - Google Patents

Description

  The present invention detects and analyzes defects such as foreign matter generated in a manufacturing process in which a pattern is formed on a substrate to manufacture an object, such as a semiconductor manufacturing process, a liquid crystal display element manufacturing process, a printed circuit board manufacturing process, and the like. The present invention relates to a defect inspection apparatus and method for inspecting the occurrence state of defects such as foreign matters in a manufacturing process in which measures are taken.

In the conventional semiconductor manufacturing process, if foreign matter exists on the semiconductor substrate (wafer), it may cause defects such as wiring insulation failure or short circuit, and when the semiconductor element is miniaturized and fine foreign matter exists in the semiconductor substrate In addition, this foreign matter may cause defective insulation of the capacitor and destruction of the gate oxide film. These foreign substances are various due to various causes such as those generated from the moving part of the transfer device, those generated from the human body, those generated by reaction in the processing apparatus by the process gas, those mixed in chemicals and materials, etc. It is mixed in the state of.
Even in the same liquid crystal display element manufacturing process, if a foreign substance is mixed on a pattern or some kind of defect occurs, it cannot be used as a display element. The situation is the same in the manufacturing process of the printed circuit board, and the inclusion of foreign matter causes a short circuit of the pattern and a defective connection.
As one of the conventional techniques for detecting foreign matters on a semiconductor substrate of this type, as described in Japanese Patent Application Laid-Open No. 62-89336 (conventional technique 1), a semiconductor is irradiated with a laser to produce a semiconductor. By detecting scattered light from foreign matter generated when foreign matter is attached to the substrate and comparing it with the inspection result of the same type semiconductor substrate inspected immediately before, there is no false information due to the pattern, and high sensitivity and high reliability There are those that allow frequent foreign object and defect inspection. Further, as disclosed in Japanese Patent Application Laid-Open No. 63-135848 (Prior Art 2), the scattering from the foreign matter generated when the semiconductor substrate is irradiated with a laser and the foreign matter adheres to the semiconductor substrate. There is one that detects light and analyzes the detected foreign matter by an analysis technique such as laser photoluminescence or secondary X-ray analysis (XMR).

In addition, as a technique for inspecting the above foreign matter, there is known a method for emphasizing and detecting foreign matter and defects having no repeatability by irradiating a wafer with coherent light and removing light emitted from a repeated pattern on the wafer with a spatial filter. It has been.
In addition, the 0th-order diffracted light from the main line group is input into the aperture of the objective lens by irradiating the circuit pattern formed on the wafer from a direction inclined 45 degrees with respect to the main line group of the circuit pattern. A foreign matter inspection apparatus which is not allowed to be used is known in Japanese Patent Laid-Open No. 1-117024 (Prior Art 3). This prior art 3 also describes that other straight line groups that are not the main straight line group are shielded by a spatial filter.
Further, as prior art relating to a defect inspection apparatus and method for foreign matters, etc., Japanese Patent Laid-Open No. 1-250847 (prior art 4), Japanese Patent Laid-Open No. 6-258239 (prior art 5), and Japanese Patent Laid-Open No. 6-324003. (Prior Art 6), JP-A-8-210989 (Prior Art 7), and JP-A-8-271437 (Prior Art 8) are known.

JP-A-62-89336 JP-A 63-135848 Japanese Patent Laid-Open No. 1-117024 JP-A-1-250847 JP-A-6-258239 JP-A-6-324003 JP-A-8-210989 JP-A-8-271437

However, in the prior arts 1 to 8, it has not been possible to easily detect a defect such as a fine foreign matter on a substrate in which a repeated pattern and a non-repeated pattern are mixed with high sensitivity and high speed.
That is, in the prior arts 1 to 8, there is a problem that the detection sensitivity (minimum detection foreign matter size) is remarkably lowered in a portion other than a repeated portion such as a memory cell portion.
Moreover, in the said prior arts 1-8, there existed a subject that a sensitivity fell remarkably on the oxide film which permeate | transmits illumination light.
Moreover, the said prior arts 1-8 had the subject that defects, such as a fine foreign material, were not detectable.

Moreover, in the said prior art 1-8, the time of mass production start-up of a semiconductor manufacturing process and the mass production line are not distinguished, and the inspection apparatus used by mass production start-up work is applied as it is also in the mass production line. Therefore, it is necessary to quickly detect the occurrence of foreign matter and take countermeasures.
However, since the conventional defect inspection apparatus has a large apparatus size and must be installed independently, the semiconductor substrate, the liquid crystal display element substrate, and the printed circuit board processed on the production line are brought into the inspection apparatus to introduce foreign substances and It was to inspect for defects. Therefore, it takes time to transport these substrates and to inspect foreign objects and defects, and it is difficult to inspect all of the substrates, and it is difficult to obtain a sufficient inspection frequency even for sampling inspection.
In addition, such a configuration requires manpower.

In order to solve the above-described problems, the present invention can inspect a defect such as a minute foreign substance at high speed and with high accuracy for a substrate to be inspected in which a repeated pattern, a non-repeated pattern, and no pattern are mixed. An object of the present invention is to provide a defect inspection apparatus and a method thereof.
Another object of the present invention is to provide a defect inspection apparatus and method capable of realizing all inspections, sampling inspection with sufficient inspection frequency, and constructing a highly efficient substrate production line.
Further, another object of the present invention is to effectively use the light amount of a Gaussian beam emitted from a normal inexpensive light source, for example, a laser light source, and to have defects such as extremely small foreign matters of about 0.1 μm or less. Another object of the present invention is to provide a defect inspection apparatus and method capable of inspecting with high sensitivity and high speed.
Still another object of the present invention is to effectively use, for example, the light quantity of a Gaussian beam emitted from a laser light source, and on the inspection target substrate as the MTF decreases with increasing distance from the optical axis in the detection optical system. To provide a defect inspection apparatus and method capable of solving a shortage of illuminance in the periphery of a detection region and inspecting defects such as extremely small foreign matters of about 0.1 μm or less with high sensitivity and high speed. It is in.

In addition, another object of the present invention is to set a threshold level as a judgment criterion to an optimum sensitivity in accordance with various circuit pattern regions in a structure arranged on a substrate to be inspected, and to provide a false report. An object of the present invention is to provide a defect inspection apparatus capable of inspecting defects such as true foreign matters without significantly increasing the number of defects.
Still another object of the present invention is to set a threshold level as a determination criterion in accordance with a defect size such as a foreign object to be detected in various circuit pattern areas in a structure arranged on a substrate to be inspected. Another object of the present invention is to provide a defect inspection apparatus capable of inspecting defects such as foreign matters having a size desired to be detected.
Still another object of the present invention is to inspect defects such as foreign matter by enabling estimation of the size of defects such as foreign matters existing in various circuit pattern areas in the structure arranged on the substrate to be inspected. An object of the present invention is to provide a defect inspection apparatus capable of performing the above.

  Still another object of the present invention is to provide a method of manufacturing a semiconductor substrate that can manufacture the semiconductor substrate with high efficiency and high yield.

In order to achieve the above object, the present invention provides a slit beam having substantially parallel light in the longitudinal direction with a predetermined inclination from the normal direction of the substrate with respect to the substrate to be inspected on which the circuit pattern is formed. (Π / 2−α1), a stage having a predetermined inclination φ1 on a plane with respect to a main straight line group of the circuit pattern, and a longitudinal direction of the stage on which the substrate to be inspected is placed and traveled Illumination process that illuminates at almost right angles to the direction, and reflected and scattered light obtained from defects such as foreign matter existing on the substrate to be inspected illuminated in the illumination process is received by the image sensor and converted into a signal A defect inspection method comprising: a detection process for conversion and detection; and a defect determination process for extracting a signal indicating a defect such as a foreign substance based on a signal detected in the detection process.
In the detection process of the defect inspection method, the present invention is characterized in that a diffracted light pattern from at least a repetitive pattern of a circuit pattern existing on a substrate to be inspected is shielded by a spatial filter.
According to the present invention, in the defect determination process in the defect inspection method, a variation is calculated based on a signal obtained by the detection from a location where the same circuit pattern is originally formed or the vicinity thereof, and the calculated variation A signal indicating a defect such as a foreign substance is extracted from the detected signal based on a determination criterion (threshold value) set based on the above.
Further, the present invention provides a method for determining a foreign matter or the like based on a determination criterion (threshold value) set for each of various areas constituting the circuit pattern with respect to the detected signal in the defect determination process in the defect inspection method. A signal indicating a defect is extracted. Further, the present invention is characterized in that, in the illumination process in the defect inspection method, a plane inclination φ1 with respect to a main straight line group of the circuit pattern of the slit beam is about 45 degrees.

Further, the present invention provides a stage on which a substrate to be inspected on which a circuit pattern is formed is placed and traveled, and a beam emitted from the light source is formed into a slit-like beam composed of substantially parallel light in the longitudinal direction, and The substrate to be inspected has a predetermined inclination (π / 2−α1) from the normal direction of the substrate, has a predetermined inclination φ1 on a plane with respect to the main straight line group of the circuit pattern, It is obtained from an illumination optical system that illuminates so that the longitudinal direction is substantially perpendicular to the traveling direction of the stage, and defects such as foreign matter existing on the substrate to be inspected on which the slit beam is illuminated by the illumination optical system. Detection optical system that receives reflected and scattered light by an image sensor, converts it into a signal, and detects it, and image processing that extracts a signal indicating a defect such as a foreign substance based on the signal detected from the image sensor of the detection optical system Part and A defect inspection apparatus characterized by was e.
According to the present invention, in the image processing unit of the defect inspection apparatus, the variation is calculated based on a signal obtained by detection from the image sensor of the detection optical system from a position where the same circuit pattern is originally formed or its vicinity. And a detection criterion setting means for setting a determination criterion (threshold value) based on the calculated variation, and a detection criterion set by the determination criterion setting means, which is detected from the image sensor of the detection optical system. And extraction means for extracting a signal indicating a defect such as a foreign substance from the signal.
According to the present invention, in the image processing unit of the defect inspection apparatus, a signal indicating a defect such as a foreign substance based on a determination criterion set for each of various regions constituting the circuit pattern with respect to the detected signal. It has the means to extract, It is characterized by the above-mentioned. The present invention is also characterized in that, in the detection optical system in the defect inspection apparatus, the optical axis is substantially perpendicular to the inspection target substrate.

  According to the present invention, in the illumination optical system in the defect inspection apparatus, the light source is a laser light source. In the illumination optical system of the defect inspection apparatus according to the present invention, the illumination optical system includes an optical element that narrows a shape that approximates a conical surface. The illumination optical system in the defect inspection apparatus may further include an optical system that illuminates white light from a direction inclined with respect to the normal line. According to the present invention, the detection optical system in the defect inspection apparatus includes a spatial filter. According to the present invention, in the defect inspection apparatus, the image sensor in the detection optical system is a TDI sensor. In the invention, it is preferable that the TDI sensor is an anti-blooming type.

  In the detection optical system of the defect inspection apparatus, the present invention is characterized in that the optical axis is inclined with respect to the normal line of the substrate to be inspected. The present invention also provides a conical surface for illuminating light with a certain inclination from the normal direction with respect to the surface of the object to be inspected, and condensing the illumination light beam in at least one direction within the surface of the object to be inspected. An illumination optical system having optical elements having approximate shapes, a detection optical system that receives light reflected from the object to be inspected by an image sensor, converts it into a signal, and detects the signal, and a signal detected by the detection optical system A defect inspection apparatus comprising: an image processing unit for processing The present invention also provides an illumination optical system for illuminating the surface of the object to be inspected, means for accumulating the generated charge, and the accumulated charge exceeding the certain amount when the accumulated charge exceeds a certain amount. And an image sensor configured to read out the accumulated charge in a range up to a certain amount, and the image sensor receives light reflected from the object to be inspected. A defect inspection apparatus comprising: a detection optical system that converts a signal into a signal and detects the signal; and an image processing unit that processes the signal detected by the detection optical system.

  According to the present invention, in the detection optical system in the defect inspection apparatus, the reflected light beam from the object to be inspected is branched, and the intensity of the one reflected light beam branched is approximately 1 of the intensity of the other reflected light beam. / 100 branch optical system, and a plurality of image sensors that receive each reflected light beam branched by the branch optical system. In addition, the present invention is installed substantially parallel to an illumination optical system for illuminating light on the surface of the object to be inspected and light scattered from a pattern formed on the surface of the object to be inspected. A plurality of linear light shielding means, the optical axis has an inclination of a certain angle from the normal direction of the surface of the inspection object, and the light reflected from the inspection object is converted by the photoelectric conversion means. A defect inspection apparatus comprising: a detection optical system that receives light, converts it into a signal, and detects it; and an image processing unit that processes the signal detected by the detection optical system. The present invention also provides an illumination optical system for illuminating light on the surface of an object to be inspected on which a plurality of structures having substantially the same shape are arranged, and the reflected light from the object to be inspected is received by an image sensor. Based on the detection optical system that converts the image signal into a detection signal and the image signal detected from the detection optical system, the variation of the image signal is calculated for the corresponding pixel of the structure having the same shape or the neighboring pixel. A determination reference setting means for setting a determination reference (threshold value) of a signal level of a pixel for determining the presence of a defect such as a foreign substance according to the calculated variation, and a determination reference set by the determination reference setting means. A defect inspection apparatus comprising: a determination unit that determines presence of a defect with respect to an image signal detected from the detection optical system, and an image processing unit that processes the image signal. is there.

  In the image processing unit of the defect inspection apparatus according to the present invention, the image processing unit further includes setting means for setting a magnification of the determination reference with respect to variations in the image signal. The present invention also provides an illumination optical system for illuminating light on the surface of an object to be inspected on which a plurality of structures having substantially the same shape are arranged, and the reflected light from the object to be inspected is received by an image sensor. A detection optical system that converts the detection signal into an image signal and detects the difference, and a difference value calculation unit that calculates a difference value of the image signal for a corresponding pixel of a structure having the same shape based on the image signal detected from the detection optical system Variation of the difference value calculated by the difference value calculation means in a plurality of pixels adjacent to a pixel for determining the presence of a defect such as a foreign material and the presence of a defect such as a foreign material according to the calculated variation. A determination reference setting means for setting a determination reference for the signal level of the pixel to be determined, and the presence of a defect is determined for the image signal detected from the detection optical system based on the determination reference set by the determination reference setting means. With judgment means A defect inspection apparatus characterized by comprising an image processing unit for processing the image signal. Further, the present invention provides an output means for outputting an inspection result of the defect determined by the determination means and data corresponding to the determination criterion set by the determination criterion setting means to the image processing unit in the defect inspection apparatus. It is characterized by having.

  The present invention also provides an illumination optical system for illuminating light on the surface of an object to be inspected on which a plurality of structures having substantially the same shape are arranged, and the reflected light from the object to be inspected is received by an image sensor. A detection optical system that converts the image signal into an image signal and detects the image signal, a determination unit that determines the presence of a defect based on a determination criterion for the image signal detected from the detection optical system, and a determination criterion that is determined by the determination unit A defect inspection apparatus comprising an image processing unit having display means for displaying map information or an image with respect to a structure having the same shape. According to the present invention, the image processing unit in the defect inspection apparatus includes means capable of setting the determination criterion according to an area priority mode, a standard mode, and a sensitivity priority mode. The present invention also provides an illumination optical system for illuminating light on the surface of an object to be inspected on which a plurality of structures having substantially the same shape are arranged, and the reflected light from the object to be inspected is received by an image sensor. A detection optical system that converts the image signal into an image signal and detects the image signal, a determination unit that determines the presence of a defect based on a determination criterion for the image signal detected from the detection optical system, and a determination criterion that is determined by the determination unit And a display means for displaying a relationship between the index and the index related to the inspection area corresponding thereto.

  The present invention also provides an illumination optical system for illuminating light on the surface of an object to be inspected on which a plurality of structures having substantially the same shape are arranged, and the reflected light from the object to be inspected is received by an image sensor. A detection optical system that converts the image signal into an image signal and detects the image signal, a determination unit that determines the presence of a defect based on a determination criterion for the image signal detected from the detection optical system, and a determination criterion that is determined by the determination unit An image processing unit having display means for displaying sensitivity information for a structure having the same shape corresponding to the defect inspection apparatus. The present invention also provides an illumination optical system for illuminating light on the surface of an object to be inspected on which a plurality of structures having substantially the same shape are arranged, and the reflected light from the object to be inspected is received by an image sensor. A detection optical system that converts the image signal into a detection signal, and a determination reference that is set by a determination reference setting unit that changes and sets the determination reference corresponding to the ground state of the structure having the same shape. A determination means for determining the presence of a defect with respect to the image signal detected from the detection optical system based on a determination criterion, and an image processing unit for processing the image signal. It is a defect inspection apparatus. The present invention also provides an illumination optical system for illuminating light on the surface of an object to be inspected on which a plurality of structures having substantially the same shape are arranged, and the reflected light from the object to be inspected is received by an image sensor. A detection optical system that converts the image signal into a detection signal, a designation unit that designates a defect size, a judgment criterion setting unit that sets a judgment standard according to the defect size designated by the designation unit, and the judgment reference setting A determination unit that determines the presence of a defect with respect to the image signal detected from the detection optical system based on a determination criterion set by the unit, and an image processing unit that processes the image signal. A defect inspection apparatus including the defect inspection apparatus.

  The present invention also provides an illumination optical system for illuminating light on the surface of an object to be inspected on which a plurality of structures having substantially the same shape are arranged, and the reflected light from the object to be inspected is received by an image sensor. A detection optical system that converts the image signal into a detection signal, a designation unit that designates the size of the defect, and a control that controls the power of the illumination light illuminated by the illumination optical system in accordance with the size of the designated defect A defect inspection apparatus including an image processing unit that processes an image signal detected from the detection optical system. The present invention also provides an illumination optical system for illuminating light on the surface of an object to be inspected, which is placed on a stage and in which a plurality of structures having substantially the same shape are arranged, and reflection from the object to be inspected. An imaging optical system having a detection optical system that detects light by receiving light with an image sensor and converting it into an image signal, and a defect based on a criterion for the image signal detected from the detection optical system of the imaging optical system An image processing unit having a determination unit for determining the presence of the imaging object, and an optical observation microscope arranged in parallel with the imaging optical system for observing an optical image on the object to be inspected. It is a defect inspection device.

  Further, the present invention is characterized in that the optical observation microscope in the defect inspection apparatus is constituted by an ultraviolet optical observation microscope. The present invention also relates to an illumination optical system that illuminates light on the surface of the object to be inspected, and a detection optical system that receives light reflected from the object to be inspected by a photoelectric conversion means and converts it into a signal for detection. And an image processing unit having means for processing the signal detected by the detection optical system to perform defect inspection and outputting the defect inspection result including pattern information in which a defect exists. It is a defect inspection device. In the image processing unit of the defect inspection apparatus according to the present invention, the output pattern information is information obtained from design data of a structure. The present invention also relates to an illumination optical system that illuminates light on the surface of the object to be inspected, and a detection optical system that receives light reflected from the object to be inspected by a photoelectric conversion means and converts it into a signal for detection. And processing the signal detected by the detection optical system to extract the signal level of the defect, correcting the signal level of the extracted defect to indicate the size of the defect, and correcting the signal level of the corrected defect A defect inspection apparatus including an image processing unit having a means for outputting.

  Further, the present invention is characterized in that in the means in the defect inspection apparatus, the signal level of the defect is corrected based on data of illumination intensity or reflectance of the structure surface. According to the present invention, in the illumination optical system in the defect inspection apparatus, a beam emitted from the light source as the slit beam is emitted from the light in the detection region with respect to the detection region on the substrate to be inspected. It is characterized by having an optical system that is shaped so as to have an illuminance distribution that is a Gaussian distribution with the length from the axis to the peripheral portion being substantially standard deviation, and obtains a slit-shaped Gaussian beam. According to the present invention, in the illumination optical system in the defect inspection apparatus, a beam emitted from the light source as the slit-shaped beam is made to be the center of the detection region with respect to the detection region on the inspection target substrate. The diameter or the major axis length is set to the length between the peripheral portions around the optical axis of the detection region so that the ratio of the illuminance of the peripheral portion of the detection region to the illuminance of the portion is about 0.46 to 0.73. It has an optical system for obtaining a slit-shaped Gaussian beam by adapting and shaping. Further, the present invention is characterized in that the slit-shaped Gaussian beam beam illuminated by the illumination optical system in the defect inspection apparatus is a DUV beam beam. According to the present invention, in the defect inspection apparatus, the image sensor in the detection optical system is a TDI image sensor.

  As described above, according to the above configuration, a defect such as a minute foreign matter can be inspected at high speed and with high accuracy for a substrate to be inspected in which a repeated pattern, a non-repeated pattern, and no pattern are mixed. Can do. Further, according to the above-described configuration, the light amount of a Gaussian light beam emitted from a normal inexpensive light source, for example, a laser light source, is effectively used, and defects such as extremely small foreign matters of about 0.1 μm or less are also highly sensitive. And high-speed inspection. Further, according to the above configuration, for example, the amount of Gaussian beam emitted from the laser light source is effectively used, and as the MTF decreases with increasing distance from the optical axis in the detection optical system, the periphery of the detection region on the inspection target substrate It is an object of the present invention to provide a defect inspection apparatus and method capable of solving a shortage of illuminance in a portion and inspecting defects such as extremely small foreign matters of about 0.1 μm or less with high sensitivity and high speed. In addition, according to the above configuration, the threshold level that is a determination criterion is set to an optimum sensitivity in accordance with various circuit pattern regions in the structure arranged on the substrate to be inspected, and the false alarm is remarkably increased. And defects such as true foreign matter can be inspected.

  Further, according to the above configuration, it is desired to detect by setting a threshold level as a determination criterion in accordance with the defect size of a foreign substance or the like to be detected in various circuit pattern regions in the structure arranged on the inspection target substrate. It is possible to inspect defects such as size foreign matter. Moreover, according to the said structure, it can test | inspect defects, such as a foreign material, so that the size of defects, such as a foreign material which exists in the various circuit pattern area | regions in the structure arranged on the board | substrate to be inspected, can be estimated. . Further, according to the above configuration, it is possible to realize a total efficiency inspection, a sampling inspection with a sufficient inspection frequency, and build a highly efficient substrate manufacturing line.

According to the present invention, the efficiency of illumination can be improved, the diffracted light from the pattern in the substrate such as the LSI pattern can be reduced by the spatial filter and the illumination direction, and a threshold value is set for each position with different variations in the chip. Since it can be set low, it is possible to detect foreign matter and defects on a substrate such as an LSI wafer with high sensitivity and high throughput. In addition, according to the present invention, a high-sensitivity normal TDI sensor is used to detect, with high sensitivity and high speed, a minute foreign object or defect existing on a substrate to be inspected in which a repeated pattern and a non-repeated pattern are mixed. There is an effect that can be done.
In addition, according to the present invention, it is possible to effectively utilize the amount of light of a Gaussian light beam emitted from a normal inexpensive light source, for example, a laser light source, and to detect defects such as extremely small foreign matters of about 0.1 μm or less with high sensitivity. In addition, there is an effect that inspection can be performed at high speed. In addition, according to the present invention, for example, the amount of Gaussian beam emitted from a laser light source is effectively used, and as the MTF decreases with increasing distance from the optical axis in the detection optical system, the periphery of the detection region on the inspection target substrate Insufficient illuminance in the area is eliminated, and defects such as extremely small foreign matters of about 0.1 μm or less can be inspected with high sensitivity and at high speed.

Further, according to the present invention, the false alarm can be remarkably increased by setting the threshold level, which is a criterion, to the optimum sensitivity in accordance with various circuit pattern regions in the structure arranged on the substrate to be inspected. And there is an effect that it is possible to inspect for defects such as true foreign matter. In addition, according to the present invention, it is desired to detect by setting a threshold level as a determination criterion in accordance with the defect size of a foreign substance or the like to be detected in various circuit pattern areas in the structure arranged on the substrate to be inspected. There is an effect that it is possible to inspect defects such as foreign matters of size.
In addition, according to the present invention, it is possible to inspect defects such as foreign matters so that the size of the defects such as foreign matters existing in various circuit pattern regions in the structure arranged on the substrate to be inspected can be estimated. There is an effect.

  Embodiments according to the present invention will be described with reference to the drawings. First, an inspection object 1 for inspecting a defect such as a foreign object according to the present invention will be described with reference to FIGS. 1 and 2. As an inspection object 1 for inspecting a defect such as a foreign substance, there is a semiconductor wafer 1a in which chips 1aa made of a memory LSI are two-dimensionally arranged at a predetermined interval as shown in FIG. A chip 1aa made of a memory LSI is mainly formed with a memory cell area 1ab, a peripheral circuit area 1ac made up of a decoder, a control circuit, and the like, and another area 1ad. The memory cell region 1ab is formed by regularly (repetitively) arranging two-dimensional memory cell patterns having a minimum line width of, for example, about 0.1 to 0.3 μm. However, the peripheral circuit region 1ac is formed as a non-repetitive pattern in which a pattern having a minimum line width of, for example, about 0.2 to 0.4 μm is not regularly arranged two-dimensionally. Further, as other regions, for example, there is a bonding area region (minimum line width on the order of, for example, 10 μm and close to no pattern).

  As the inspection object 1 for inspecting a defect such as a foreign substance, there is a semiconductor wafer 1b in which chips 1ba made of LSI such as a microcomputer are two-dimensionally arranged at a predetermined interval as shown in FIG. A chip 1ba made of an LSI such as a microcomputer is mainly composed of a register group area 1bb, a memory area 1bc, a CPU core area 1bd, and an input / output area 1be. FIG. 2 conceptually shows the arrangement of the memory area 1bc, the CPU core area 1bd, and the input / output area 1be. The register group region 1bb and the memory portion region 1bc are formed by regularly arranging (repeating) a pattern having a minimum line width of about 0.1 to 0.3 μm in two dimensions. In the CPU core area 1bd and the input / output area 1be, a pattern having a minimum line width of about 0.1 to 0.3 μm is formed non-repetitively. As described above, the inspection object 1 for inspecting defects such as foreign matters has the chips arranged regularly even for the semiconductor wafer, but the minimum line width is different for each region in the chip, In addition, there are various forms such as repeated, non-repeated, and none patterns.

  The defect inspection apparatus and method for foreign matter or the like according to the present invention provides a zero-order diffracted light from a pattern (linear pattern) consisting of straight lines on a non-repeated pattern region in a chip in such an inspection object 1. As shown in FIG. 12 and FIG. 21, it is prevented from entering the entrance pupils 20a and 20c of the objective lens and receiving scattered light generated by defects such as foreign matters existing on the non-repeated pattern region. The signal can be detected, and the position coordinates of the defect can be calculated. In addition, the defect inspection apparatus and method for foreign matters according to the present invention can be applied to the inspected object 1 even if the background signal varies due to subtle differences in processes that do not become defects, noise during detection, and the like. By setting a threshold value for extracting a defect such as a foreign substance according to the variation, the detection sensitivity and throughput of the defect such as the foreign substance are improved.

  Next, a first embodiment of a defect inspection apparatus for foreign matter or the like according to the present invention will be described with reference to FIGS. A first embodiment of a defect inspection apparatus for foreign matter or the like includes a stage unit 300 including a substrate mounting table 304, xyz stages 301, 302, and 303 and a stage controller 305, a laser light source 101, a concave lens 102, and a convex lens 103. Three illumination optical system units 100 including a configured beam splitter, and an illumination lens 104 having a conical curved surface, a detection lens 201, a spatial filter 202, an imaging lens 203, an ND (Neutral Density) filter 207, and a beam splitter 204, a polarizing element 208, a detection optical system unit 200 including one-dimensional detectors (image sensors) 205 and 206 such as a TDI sensor, and an A / D conversion unit 401 such as a chip as shown in FIG. Since the pattern is repeated, the data memory is delayed by one chip. 402, a difference processing circuit 403 that takes a signal difference between chips, a memory 404 that temporarily stores a difference signal between chips, a maximum / minimum removal circuit 405 that removes an unusual maximum and minimum signal in the difference signal, and a signal level s Square calculation circuit 406, signal level s calculation circuit 407, number count circuit 408, square sum calculation circuit 409 that integrates the square of s, sum calculation circuit 410 that integrates s, and the number of samples for obtaining variation Count circuit 411 for calculating n, upper limit determination criterion (positive threshold) calculation circuit 412, lower limit determination criterion (negative threshold) calculation circuit 413, comparison circuits 414 and 415, defect detection results such as foreign matter and the like, and defect detection White illumination composed of an arithmetic processing unit 400 composed of output means 417 for outputting a result, a white light source 106 and an illumination lens 107 It constituted by a university system unit 500. In particular, as the TDI sensor, an anti-blooming type is desirable. As described above, when the anti-blooming type is used as the TDI sensor, it is possible to inspect defects such as foreign matters in the vicinity of the saturation region.

The arithmetic processing unit 400 will be described in detail later. The three illumination optical system units 100 pass the light emitted from the laser light source 101 through a beam splitter composed of a concave lens 102 and a convex lens 103, and an illumination lens 104 having a conical curved surface, as shown in FIG. 3 is planarly arranged in three directions 10, 11, and 12 such that the longitudinal direction of the slit-shaped beam 3 is directed to the chip arrangement direction with respect to the wafer (substrate to be inspected) 1 placed on the mounting table 304. Configured to illuminate. The reason why the slit beam 3 is used as the illumination light is that the inspection of defects such as foreign matters has been speeded up. That is, as shown in FIG. 5, the beam 3 illuminated on the wafer 1 in which the chips 2 are arranged in the x direction in the scanning direction of the x stage 301 and the y direction in the scanning direction of the y stage 302, Is illuminated with a slit beam that is narrow in the scanning direction y and wide in the vertical direction x (scanning direction of the x stage 301). The slit-shaped beam 3 is illuminated so as to be parallel light in the x direction so that an image of a light source is formed in the y direction. The illumination of the slit beam 3 from the three directions 10, 11, and 12 may be performed individually or from the two directions 10 and 12 at the same time.
Incidentally, the TDI sensor 205 is configured such that the longitudinal direction of the slit-shaped beam 3 is directed in the chip arrangement direction with respect to the wafer (substrate to be inspected) 1 and perpendicular to the scanning direction y of the y stage 302. , 206 and the stage traveling direction can be kept parallel to each other, so that a normal TDI sensor can be used as shown in FIG. In addition, the defect position coordinates can be easily calculated, and as a result, the speed of defect inspection for foreign matters can be increased. In particular, the illumination of the slit-shaped beam 3 from directions 10 and 12 is directed in the chip arrangement direction with respect to the wafer (substrate to be inspected) 1 and perpendicular to the scanning direction y of the y stage 302. In order to achieve this, an illumination lens 104 having a conical curved surface is required.

FIG. 6 shows a conical illumination lens 104. The illumination lens 104 is a lens that has a different focal length at a position in the longitudinal direction of the cylindrical lens and linearly changes the focal length. With this configuration, as shown in FIG. 6, even when illuminated obliquely (with both α1 and φ1 tilts compatible), it is possible to illuminate with the slit-shaped beam 3 that is narrowed down in the y direction and collimated in the x direction. That is, the illumination lens 104 can realize illumination having parallel light in the x direction as shown in FIG. 9A and near φ1 = 45 degrees. In particular, as shown in FIG. 9A, by making the slit-shaped beam 3 parallel light in the x direction, a diffracted light pattern is obtained from a circuit pattern in which main straight line groups are directed in the x direction and the y direction. The light can be shielded by the spatial filter 202.
Next, a manufacturing method of the illumination lens 104 having a conical curved surface will be described with reference to FIGS. The conical lens 104 can be made by polishing a cone 23 having a predetermined bottom area and height using glass or quartz as a material, and cutting out a lens on one side from a predetermined position. The curved surface of the lens shown in FIG. 6 required by the present invention should be a curved surface 24 as shown in FIG. However, since the solid shown in FIG. 8 is not a rotating body and is difficult to polish, it is approximated by the cone 23 shown in FIG. In reality, N.I. A. If the lens is about 0.02 to 0.2, there is no big problem.

  The shape of the surface of the cone 23 shown in FIG. 7 follows the following equation (1), where the origin is at the apex and the apex angle is θ1.

x ² + y ² = (ztanθ1) ² (Equation 1)
Further, the curved surface 24 shown in FIG. 8 follows the following equation (2), where the origin is at the apex and the apex angle is θ2.

(X-ztan θ2) ² + y ² = (z · tan θ2) ² (Equation 2)
Note that the method of creating the conical lens 104 is not limited to the production method shown here, for example, an injection molding method in which plastic or the like is poured into a die having a conical surface that has been created in advance, or a cone that has been created in advance. It can also be created by a method of placing a glass substrate on the surface and melting the substrate.

  In the present invention, this conical lens 104 is used to realize y-direction critical and x-direction collimated illumination. The configuration for this is shown in FIGS. 9 (a) and 9 (b). The light emitted from the laser light source 101 enters the conical lens 104 via a beam expander composed of a concave lens 102 and a convex lens 103. Since the conical lens 104 does not have a lens effect in the x direction, it is illuminated in a collimated form. Further, since the curvature is different at both ends of the conical lens 104, the focal position is different. At the same time, the light is condensed on the wafer 1 by the curvature of the conical lens 104 in the y direction.

FIG. 10 is a plan view showing three illumination optical system units 100 constituted by one laser light source 101 as the laser light source 101. The laser beam emitted from the laser light source 101 is branched into two optical paths by a branching optical element 110 such as a half mirror, one of which is reflected by mirrors 111 and 112 and incident downward on the concave lens 102 by a mirror 113. Can be obtained, and the other travels to a branching optical element 114 such as a half mirror. One of the beams branched by the branching optical element 114 is reflected by the mirror 115 and incident downward on the concave lens 102 by the mirror 117, whereby an illumination beam from 10 directions can be obtained. By making it enter the concave lens 102 downward, an illumination beam from 10 directions can be obtained.
By the way, when illuminating only from 11 directions, it can implement | achieve by switching from the branch optical element 110 to the mirror element 118. FIG. Further, in the case of illuminating only from the directions of 10 and 12, it can be realized by exiting the branching optical element 110 from the optical path or switching to a normal optical element. Further, in the case of illumination from only the 12 directions among the illuminations from the 10 and 12 directions, it can be realized by switching from the branch optical element 114 to the mirror element 119.

As the laser light source 101, it is preferable to use the second harmonic SHG of a high-power YAG laser and a wavelength of 532 nm because of the branching relationship, but the wavelength is not necessarily 532 nm. Further, the laser light source 101 does not need to be YAGSHHG. That is, the laser light source 101 may be another light source such as an Ar laser, a nitrogen laser, a He—Cd laser, or an excimer laser.
The detection optical system unit 200 detects light emitted from the wafer 1 as a detection lens (objective lens) 201, a spatial filter 202 that shields a Fourier transform image by reflected diffracted light from a repetitive pattern, an imaging lens 203, an ND filter (wavelength). The amount of light is adjusted regardless of the band.) It is configured to be detected by one-dimensional detectors 205 and 206 such as TDI sensors through 207, beam splitter 204, and polarizing element 208. The spatial filter 202 is placed at the spatial frequency region of the objective lens 201, that is, at the imaging position of the Fourier transform (corresponding to the exit pupil) so as to shield the Fourier transform image by the reflected diffracted light from the repetitive pattern. Further, when the illumination optical system unit 100 performs polarized illumination, the polarizing element 208 shields the polarized component due to the reflected scattered light generated from the edge of the circuit pattern and transmits a part of the polarized component due to the reflected scattered light generated from the defect such as a foreign object. Therefore, it is not always necessary in the present invention. Here, the illumination area 4 on the wafer 1 shown in FIG. 5 is imaged on the detectors 205 and 206 by the objective lens 201 and the imaging 203 constituting the relay lens. That is, 4 indicates a light receiving area of the one-dimensional detectors 205 and 206 such as a TDI sensor.

As described above, when the slit-shaped beam 3 is illuminated on the wafer (substrate) 1 on which various types of circuit patterns are formed, the reflected diffracted light (or scattered light) is reflected on the surface of the wafer, the circuit pattern. Injecting from defects such as foreign matter. The emitted light is received by the detectors 205 and 206 through the detection lens 201, the spatial filter 202, the imaging lens 203, the ND filter 207, the polarizing element 208, and the beam splitter 204, and is subjected to photoelectric conversion.
Here, the order of the ND filter 207, the polarizing element 208, and the beam splitter 204 does not need to be the order given here. In particular, when the ND filter 207 is disposed after the beam splitter 204, it has an effect that the intensity of light entering the two detectors 205 and 206 can be controlled independently.
Further, the transmission and reflectance of the beam splitter 204 need not be 50%. For example, when configured as 1% and 99%, light having an intensity of about 1/100 is incident on one detector, and is obtained from two detectors that receive light of different intensities. The apparent dynamic range of the detector can be improved by using the obtained signal. Therefore, the arithmetic processing unit 400 can obtain a detection signal from a defect such as a foreign substance having an improved dynamic range by using a signal obtained from the detector 205 and a signal obtained from the detector 206. In particular, the signal obtained when the detector receives light with high intensity will emphasize the component indicating a defect with high intensity, and the signal obtained when the detector receives light with low intensity is low in intensity. Components close to the background are emphasized. Accordingly, the dynamic range of a signal indicating a defect can be improved by taking a correlation such as a ratio of both signals.
However, the dynamic range can also be changed by controlling and changing the illuminance (power) of the light beam emitted from the illumination optical system such as the laser light source 101, and the beam splitter 204 and one detector 206 are eliminated. be able to.

Next, the relationship between the slit-shaped beam 3 that illuminates the wafer 1 with the illumination optical system unit 100 according to the present invention and the detection optical system unit 200 will be described more specifically. FIG. 5 shows a plan view of the direction of illumination by the slit-shaped beam 3 and detection by the one-dimensional detectors 205 and 206 such as a TDI sensor. The wafer 1 on which the pattern 2 is formed is illuminated with a slit beam 3. The image 4 by the detection optical system of the one-dimensional detectors 205 and 206 is shown. The slit-shaped beam 3 is illuminated from the planar directions 10, 11 and 12. FIG. 11A supplementarily explains FIG. 5 and shows an illumination direction 10, a detection direction 14 (showing a case perpendicular to the wafer surface), an x-axis, and a y-axis. Further, the spherical surface 17 is virtually assumed and is for considering the opening position of the objective lens 201 in the detection optical system unit 100 in FIG. The intersections of the spherical surface 17 with the illumination light 10 and the detection light 14 are 15 and 16, respectively. FIG. 11B shows how the diffracted light is emitted when illuminated from the direction 10. The intersection 18 between the emission direction 19 of the specularly reflected light and the phantom spherical surface 17 is zero-order light, and the cone having the pattern direction (x direction, y direction) as the center and the illumination point as the apex as shown in FIG. Because of the emission in the direction of the ridge, the miracle of the intersection with the virtual spherical surface 17 is on the circumference of the bottom surface of this cone.
Therefore, when this miracle is viewed from the normal direction, it becomes a straight line parallel to the x-axis and the y-axis.

  By the way, the opening of the objective lens 201 in the detection optical system unit 200 that is not tilted with β1 = 0 is the opening 20a shown in FIGS. Here, the angles φ1 and φ2 of the illumination directions 10 and 12 are set to about 45 degrees, for example. As shown in FIG. 3, when the optical axis of the detection optical system 200 is perpendicular to the surface of the wafer 1, that is, β1 = 0, the numerical aperture (NA) of the detection lens (objective lens) 201 and the illumination light As shown in FIG. 12, the relationship between the angle α1 (FIG. 3) and the zero-order diffracted light 21x and 21y in the y-direction and the y-direction from the circuit pattern in which the main straight line group faces in the x- and y-directions is detected. Based on the condition that the light does not enter the pupil of the lens 201, it should be set within a range determined from the following equation (3). That is, the angles φ1 and φ2 of the illumination directions 10 and 12 are set to about 45 degrees, and the relationship between the numerical aperture (NA) of the detection lens (objective lens) 201 and the angle α1 of the illumination light (FIG. 3) is obtained. By satisfying the following (Equation 3), even if it is a non-repetitive pattern, the 0th-order diffraction in the x and y directions from the circuit pattern in which the main straight line group is oriented in the x and y directions It becomes possible to eliminate the light 21x and 21y from entering the opening 20a of the objective lens 201.

N. A. <Cos α1 · sin φ1 and N.I. A. <Cos α1 · sin (π / 2−φ1) (Equation 3)
If α1 is set to 30 ° or less, the numerical aperture (NA) of the objective lens 201 may be set to about 0.4 or less. In particular, these conditions are as follows: a peripheral circuit area 1ac having a non-repeating pattern in the memory LSI 1aa, a CPU core area 1bd and an input / output part area 1be having a non-repeating pattern in the LSI 1ba such as a microcomputer, etc. This is effective for a logic LSI having a non-repeating pattern. In many cases, these LSI patterns are formed so as to be parallel to each other at right angles (major straight line groups are at right angles), so that the 0th-order diffracted light is emitted in a specific direction. Therefore, by preventing the emitted 0th-order diffracted light from entering the objective lens 201, diffracted light from many of these patterns is erased, and it becomes easy to detect only reflected diffracted light from defects such as foreign matter. . More specifically, the detection signal level from the circuit pattern is lowered, and the area capable of detecting defects such as foreign matters with high sensitivity is increased. Of course, in the case of a non-repetitive pattern, high-order (first-order, second-order, third-order,...) Diffracted light is incident on the aperture 20a of the objective lens 201. , Appear as a straight line group parallel to the 0th-order diffracted beams 21x and 21y shown in FIG. Therefore, it is possible to eliminate such high-order diffracted light by shielding it with a narrow-band spatial filter 202.

  In addition, it is necessary to inspect the inspection target substrate (wafer) 1 for foreign matters or defects that have entered the recesses between the wirings and the like, etching residues, and the like. However, there is a non-repeated pattern on the substrate 1 to be inspected, and in order to prevent 0th-order diffracted light from the non-repeated pattern from entering the objective lens 201, as described above, On the other hand, when the slit-shaped beam 3 having a longitudinal direction in the x direction from the directions 10 and 12 at an angle of approximately 45 degrees is illuminated on the substrate 1, the wirings which are convex portions obstruct the concave portions sufficiently. It becomes difficult to illuminate. Therefore, in many cases, since the wiring pattern is formed in a right angle and parallel direction, by illuminating the slit-shaped beam 3 on the substrate 1 from the direction 11 parallel to the y axis, the wiring pattern or the like It becomes possible to sufficiently illuminate the recess. In particular, the wiring pattern of the memory LSI is often a linear pattern having a length of several millimeters, and can often be inspected by illumination from this direction 11. Further, depending on the pattern, in the case of the 90-degree direction, the inspection can be performed by rotating the wafer by 90 degrees or changing the illumination direction to the x direction.

  However, when the slit-shaped beam 3 is illuminated from the direction 11, the 0th-order diffracted light 21 y ′ in the y-direction is incident on the opening 20 a of the objective lens 201 as shown in FIG. 13. Therefore, it is necessary to erase at least the 0th-order diffracted light 21y ′ by shielding it with the spatial filter 202. At this time, naturally, higher-order diffracted light can be eliminated by being shielded by the spatial filter 202. The method for erasing the 0th-order diffracted light from the non-repeated pattern in the case of the non-repeated pattern existing in the chip 2 on the substrate 1 to be inspected has been described above. The memory cell region in the memory LSI 1aa is included in the chip 2. 1ab and a register group region 1bb and a memory unit region 1bc in an LSI 1ba such as a microcomputer, a repetitive pattern exists, and diffracted light fringes (diffraction interference light fringes) from the repetitive pattern are shielded by the spatial filter 202. It is required to do. In short, in the chip 2, a repetitive pattern, a non-repetitive pattern, and a non-pattern are mixed, and the line widths are also different from each other. The light shielding pattern of the spatial filter 202 is set so as to be erased. As the spatial filter 202, as described in JP-A-5-218163 and JP-A-6-258239, the. If what can change the light shielding pattern is used, it may be changed according to the circuit pattern in the chip 2. Alternatively, a spatial filter 202 having a different light shielding pattern may be prepared and switched according to the circuit pattern in the chip 2.

  Next, detection sensitivity adjustment according to the defect size of a foreign substance or the like to be detected will be described. That is, if the detection pixel size on the inspection object 1 of the one-dimensional detectors (image sensors) 205 and 206 such as TDI sensors is reduced, the detection sensitivity can be improved although the throughput is lowered. Therefore, when detecting defects such as foreign matters of about 0.1 μm or less, it is preferable to switch to the detection optical system 200 for reducing the pixel size. Specifically, it is preferable to have three types of detection optical systems 200 such that the image size on the wafer 1 is 2 microns, 1 micron, and 0.5 microns for pixels such as TDI sensors. As a method for realizing this configuration, the entire optical system 200 may be switched, only the lens (lens group) 203 may be switched, or the lens (lens group) 201 may be switched. At this time, it is preferable to design the lens configuration so as not to change the optical path length from the wafer 1 to the one-dimensional detectors 205 and 206 such as TDI sensors. Of course, if such a design is difficult, a mechanism that can change the distance to the sensor in conjunction with the switching of the lens may be used. Moreover, you may switch what changed the pixel size of sensor itself. Next, a specific example of the relationship between the slit-shaped beam 3 from three directions and the TDI sensors 205 and 206 will be described with reference to FIG. FIG. 14 shows the relationship between the TDI image 4 on the wafer 1 and the slit beam 3-10 from the direction 10 and the slit beam 3-12 from the direction 12. As shown in FIG. 10, when the illumination beams obtained by branching from the same laser light source 101 are illuminated from directions 10 and 12, these beams interfere with each other, and the intensity varies within the illumination range. It will be out. Therefore, as shown in FIG. 14, the influence of interference can be eliminated by illuminating the beams 3-10 and 3-12 so that they do not intersect within the range of the TDI image 4. When the TDI sensors 205 and 206 are used, the detection output is integrated in the y direction in synchronization with the traveling of the y stage within the range of 4, so there is no problem even if the positions are shifted in this way. Similarly, when the slit-shaped beam 3-11 from the illumination direction 11 is used, the three beams may be illuminated so that they do not intersect within a range where overlapping does not cause a problem. Needless to say, the same applies when two of the beams 10, 11, and 12 are used.

  Although not shown here, even if the slit beams irradiated from directions 10 and 12 are simultaneously overlapped and illuminated at the same location, interference occurs, but the interference fringes are in the y direction. Therefore, variation due to interference in illumination intensity can be reduced by the integration effect of the TDI sensors 205 and 206. Therefore, it is not necessary to illuminate so that the beams 3-10 and 3-12 do not cross as shown in FIG. Next, a second embodiment of the defect inspection apparatus for foreign matter and the like according to the present invention will be described. In the second embodiment, as shown in FIG. 15, the optical axis of the detection optical system unit 200 is tilted by β1 from the vertical in order to increase the intensity of scattered light from defects such as foreign matter. Other configurations are the same as those of the first embodiment shown in FIG. By the way, the reason why the optical axis of the detection optical system unit 200 is inclined by β1 from the vertical is that, as shown in FIG. 16, the intensity of scattered light from the fine particles (foreign matter) to be detected is increased, and the detection sensitivity is improved. This is because particles (foreign matter) larger than a fraction of the illumination wavelength have a large forward scattered light 51, whereas scattered light 52 from a surface near 1/10 or less of the wavelength is almost me. This is caused by the fact that the scattered light from the fine particles becomes relatively large in the forward direction. As a result, even when there are a plurality of surface irregularities on the circuit pattern in the detection pixel, the total amount is indicated by the intensity 53. Therefore, by taking forward scattering, it is possible to detect fine particles or defects with respect to the rough surface.

  However, when TDI (Time Delay Integration) sensors are used for the detectors 205 and 206, the optical axis of the detection optical system unit 200 cannot be tilted due to the depth of focus. Therefore, in the case of the second embodiment, a one-dimensional sensor is used, or the detection optical systems 201 to 203 are set to the same magnification or several times, and the inclinations of the TDI sensors 205 and 206 are set as shown in FIG. Β2 is set according to the following equation (4). In this way, the magnification can be adjusted over the entire surface.

tanβ2 = M · tanβ1 (Equation 4)
However, M is the magnification of the detection optical system 201-203. Note that the inclination β2 is not necessary when a one-dimensional sensor is used.

Next, in the second embodiment, the non-repeated pattern and the diffracted light generated from the repetitive pattern are erased, and the scattered light from the defect such as a foreign substance is detected by the one-dimensional detectors 205 and 206 such as TDI sensors. This will be explained. Also in the second embodiment, the slit-shaped beam 3 is illuminated as shown in FIG. Then, when illuminated from the direction 10 as shown in FIG. 11 (a), the state of the diffracted light emitted from the substrate 1 is as shown in FIG. 11 (b), as in the first embodiment. That is, the intersection 18 between the emission direction 19 of the specularly reflected light and the virtual spherical surface is 0th order light, the pattern direction (x direction, y direction) is the center, and the illumination point is the vertex as shown in FIG. Since it emits in the direction of the edge of the cone, the miracle of the intersection with the virtual spherical surface 17 is on the circumference of the bottom surface of the cone. Therefore, in the case of a repetitive pattern, the miracle of the 0th-order diffracted light is a straight line parallel to the x-axis and y-axis as shown in FIG. 18 when viewed from the normal direction.
In particular, in the case of a repetitive pattern, the maximum of the 0th order diffracted light is located at the intersection 22 of this straight line group. Accordingly, the opening 20b of the objective lens 201 in the detection optical system 200 tilted by β1 is as shown in FIG. When the opening 20b is viewed from the direction 14 (optical axis direction), the 0th-order diffracted light 22 appears to be emitted at the intersection of a curve and a straight line as shown in FIG. Therefore, in the spatial filter 202, if these diffracted lights are shielded by the linear light shielding part 207 as shown in FIG. 19B, the signal from the pattern can be removed. When the pattern shape and pitch of the repetitive pattern on the wafer 1 are changed, the miracle pitch in the x and y directions is changed around the injection point 18 in FIG. Therefore, the pitch and phase of the diffracted light 22 change in the opening 20b. In order to shield these diffracted lights, the pitch and phase of the linear light shielding part 207 in the spatial filter 202 may be changed.
As described above, with respect to the repetitive pattern, the diffracted light generated by the spatial filter 202 can be shielded.

Next, the case of a non-repeating pattern will be described. Even in the non-repeating pattern, it is formed mainly from a linear pattern facing the x and y directions. Therefore, when the slit-shaped beam 3 is illuminated from the direction 10, as shown in FIG. 12, zero-order diffracted lights 21x and 21y in the x and y directions are generated as shown in FIG. However, if the optical axis of the detection optical system 200 is tilted by β1, the scattered light from the fine particles increases, but the 0th-order diffracted light 21x emitted in the y direction enters the opening 20b of the objective lens 201. become. Therefore, even in the case of a non-repetitive pattern, it is necessary to shield the 0th-order diffracted light 21x by the spatial filter 202.
As described above, since the diffracted light fringes generated in the case of the repetitive pattern are different from the 0th-order diffracted light pattern in the case of the non-repeated pattern, the spatial filter 202 needs to have both diffracted light patterns. However, if both diffracted light patterns are shielded by the spatial filter, the intensity of scattered light from a defect such as a foreign substance that passes through the spatial filter is attenuated and the sensitivity is lowered.
Therefore, as in the first embodiment described above, the aperture of the objective lens 201 is positioned at 20a with the optical axis of the detection optical system 200 vertical, so that a slit shape is formed from directions 10 and 12 with respect to the non-repeated pattern. Even when the beam 3 is illuminated, it is possible to prevent the 0th-order diffracted light patterns 21x and 21y from entering the opening 20a of the objective lens 201, and to detect defects such as foreign matters existing on the non-repeated pattern. It becomes possible.

  However, as described in the first embodiment, if a defect such as a foreign substance existing in a recess between wirings is detected, a slit-like shape is formed from the y-axis direction 11 as shown in FIG. It is necessary to illuminate the beam 3. However, as shown in FIG. 13, the 0th-order diffracted light 21y 'is incident on the opening 20a of the objective lens 201, and it is necessary to shield it with the spatial filter 202 or the like. However, the detection of defects such as foreign matters existing in the recesses between the wirings is not the main detection of defects such as foreign matters, but the pattern to be inspected is specified. Is possible. If the conditions described above are configured as described below, they can be satisfied. That is, the illumination from the directions 10 and 12 inclined by about 45 degrees with respect to the y-axis is stopped, the illumination is performed with the slit-shaped beam 3 from the y-axis direction 11, and the optical axis of the detection optical system 200 is changed from the vertical to the y and x axes. By tilting in the direction, the aperture of the objective lens 201 is placed at the position 20c shown in FIG. 21 to prevent the 0th-order diffracted beams 21x ′ and 21y ′ from entering the aperture 20c of the objective lens 201 in the case of a non-repetitive pattern. can do. In this way, the spatial filter 202 can be configured to shield only the diffracted light fringes resulting from the repetitive pattern, and it is possible to prevent a decrease in the intensity of scattered light from defects such as foreign matters that pass through the spatial filter. .

  However, in this case, the N.I. A. Must be reduced. The problem is the focal positions of the detectors 205 and 206, but as shown in FIG. 17, the detectors 205 and 206 can be focused over the entire imaging area by tilting the detectors 205 and 206. In this case, the detectors 205 and 206 need to be tilted not only in the direction of β2 but also in a direction perpendicular to both β2 and the direction 14. Furthermore, since the detection optical system 200 uses a telecentric optical system, the lateral magnification does not fluctuate at different focal positions.

  Next, a third embodiment of the defect inspection apparatus for foreign matter and the like according to the present invention will be described. This third embodiment is inferior to the first and second embodiments. In the third embodiment, as shown in FIG. 22, from the directions 10 and 12 of 45 degrees with respect to the circuit pattern, the conical lens 104 is not used and the cylindrical lens 104 ′ is used and the slit beam is simply used. A slit-shaped beam 3 ′ whose longitudinal direction is directed to the illumination directions 10 and 12 is illuminated on the wafer 1. That is, a beam having a shape parallel to the incident surface of illumination is illuminated on the wafer 1 from directions 10 and 12 inclined by an angle close to 45 from the arrangement direction of the chips formed on the wafer 1. Of course, the slit-shaped beam 3 'is formed by parallel light in the longitudinal direction and narrowed in the width direction. Note that the diffracted light from the repetitive pattern and non-repetitive pattern formed in the chip 2 on the wafer 1 is the same as in the first and second embodiments. In the third embodiment, in order to simplify chip comparison, the stage scanning direction y needs to be parallel or perpendicular to the chip. Further, in this embodiment, since the integration direction of the TDI sensor is not parallel to the stage scanning direction y, the TDI sensor cannot be used as the detectors 205 and 206. Therefore, it is necessary to use a one-dimensional linear sensor as the detectors 205 and 206. In the case of a linear sensor, since an optical signal from an area narrower than the beam width of the illumination is detected, the illumination beam 3 ′ may be narrowed to a width close to the sensor image 4 in order to efficiently use the illumination light. Specifically, for example, when the sensor pixel size is 13 microns and the optical system magnification is 6.5 times, the sensor image on the wafer has a pixel size of 2 microns. In this case, for example, when a laser having a wavelength of 532 nm is used, the numerical aperture N.I. in the direction perpendicular to the longitudinal direction of the sensor of the lens 104 'of the illumination system is used. A. Is about 0.5 in the following equation (5). Of course, this is to increase the efficiency of illumination. A. It doesn't matter.

d = 1.22 · λ / N. A. (Equation 5)
Here, d is the half width of the beam, and λ is the wavelength of illumination. Further, when TDI is used for the sensors 205 and 206 in the illumination method shown in FIG. 22, it is necessary to use a special TDI having a shape as shown in FIG. That is, the special TDI sensor has a pixel configuration in which the integration direction is inclined by φ1.

Next, a defect inspection for foreign matter existing on an insulating film such as an oxide film having no pattern will be described as an object to be inspected for defect inspection for foreign matter. FIG. 24 shows the state of light scattering in a transparent film such as an oxide film. For example, when there are sufficiently small particles (foreign matter) 34 on the surface of the oxide film 32 on the substrate 33 (a fraction of the illumination wavelength), the wavefront of light from here is emitted in a spherical shape. That is, it injects to the detector side simultaneously with the injection to the oxide film side. Here, the emitted wavefront is reflected at the interface between the oxide film 32 and the base 33. The reflected light and the light emitted toward the detector cause interference in the emission direction due to interference. As a result, for example, the detection output varies depending on the directions 36, 37, and 38. Depending on the thickness and refractive index of the oxide film, the intensity distribution changes. As a result, when detected from the same direction, the intensity of the detection light changes and the sensitivity changes. However, when this model is considered, the output of the detection light does not change depending on the direction of illumination. Further, it has been confirmed through experiments that the output of the detection light does not change even when the incident angle of the illumination light is changed.
However, when white illumination is used, light interference can be eliminated. For this reason, in the first and second embodiments, the installed white illumination optical system 500 is for detecting foreign matter in the form of the insulating film 32 such as an oxide film. Therefore, when detecting foreign matter on the insulating film 32, the white light source 106 may be turned on and the laser light source 101 may be turned off. Further, when detecting a defect such as a foreign substance on a normal circuit pattern, the laser light source 101 may be turned on and the white light source 106 may be turned off. Moreover, what is necessary is just to use white illumination with respect to the to-be-inspected object influenced by the wavelength of illumination light.

  In the case of white illumination, illumination is performed in a spot shape larger than the field of view of the TDI sensor. Further, when a laser light source is used as the illumination light, in order to stabilize the detection output on the oxide film 32, the objective lens 201 having a large numerical aperture capable of detecting most of the light emitted from the wafer surface. It is necessary to detect with. Further, when an objective lens having a small numerical aperture is used, a plurality of objective lenses may be used, and detection outputs by these may be integrated. Alternatively, a plurality of illumination light wavelengths may be used, and the detection results obtained by these may be integrated. Here, it may be considered that there is almost no absorption (attenuation) of scattered light from the foreign matter in the film. When there is no foreign matter, the injection direction is one direction, and the output in this direction varies due to interference. However, when there is a foreign substance and the injection direction is widened, interference occurs in the form of an intensity distribution according to the injection direction.

In FIG. 25, the structure of the Example in the case of detecting from several directions is shown. Light emitted in the directions 213, 214, and 215 is imaged by the detection lenses 210, 211, and 212, and detected by the detectors 213, 214, and 215, respectively. This result is analog-to-digital (A / D) converted by 451, 452, and 453, integrated by the integration means 454, and binarized by an appropriate threshold value to obtain a detection result. The number of the detection systems 210, 211, and 212 is not necessarily three, and may be two. In addition, the detection system here includes a case where a plurality of detection systems 200 shown in FIG. 3 (for example, β1 = 0 degrees, β1 = 45 degrees) are used.
FIG. 26 shows changes in the detection signal when the thickness of the oxide film is changed. (A) shows an intensity distribution 48 at a certain wavelength, and (b) shows an intensity change 48, 49, 50 at three different wavelengths in an overlapping manner. It can be seen from FIG. 26B that intensity changes as shown in FIG. 26A can be greatly reduced by integrating the detection results using a plurality of wavelengths. In this case, since it is known that the detection signal intensity does not depend on the incident angle of the illumination light, illumination with different wavelengths may be illuminated from the incident angle or the direction in which φ1 is changed. That is, it is possible to detect a signal indicating a foreign substance on an insulating film such as an oxide film by the same detection optical system 200 by making the wavelengths of the slit-shaped beams from directions 10, 11, and 12 different from each other. It becomes. In this way, by making the wavelengths of the slit-shaped beams from each direction different from each other, they do not interfere with each other, so that they can be detected by the same detection optical system 200, and the cost is increased by preparing a plurality of detection optical systems. Can be avoided. Further, since the detection optical system 200 can easily correct chromatic aberration (and focal length) with at least two wavelengths, there is no realization difficulty as long as two wavelengths are used.

  Next, a fourth embodiment of the defect inspection apparatus for foreign matter and the like according to the present invention will be described. Meanwhile, semiconductor devices are required to be further miniaturized and to further improve the yield. Accordingly, a semiconductor substrate such as a semiconductor wafer for manufacturing such a semiconductor element is present on the semiconductor substrate because a very fine circuit pattern of 0.3 to 0.2 μm or less is formed. Even if there are defects such as foreign matter that are extremely small molecules of about 0.1 μm or less or those close to the atomic level, this causes a malfunction of the semiconductor device. In such a situation, in the defect inspection apparatus for foreign matter and the like according to the present invention, a microminiaturized circuit pattern of about 0.3 to 0.2 μm or less exists on a semiconductor substrate such as a semiconductor wafer. It has been demanded that defects such as extremely small foreign matter can be inspected with high sensitivity and at high speed.

  FIG. 35 is a diagram showing a schematic configuration of a fourth embodiment of a defect inspection apparatus for foreign matter or the like according to the present invention. FIG. 36 is a diagram showing an example of the illumination optical system. That is, the defect inspection apparatus for foreign matter or the like is a stage 301 on which an inspection object 1 for inspecting defects such as foreign matter or the like on which an extremely fine circuit pattern is formed, such as the semiconductor wafer (semiconductor substrate). , 302, 303, an illumination light source 101 including a laser light source such as a semiconductor laser, an argon laser, a YAG-SHG laser, and an excimer laser, and high-intensity light emitted from the illumination light source (laser source) 101 As shown in FIG. 37, illumination optical systems 102 to 105 that illuminate the inspection target substrate 1 with a slit-shaped Gaussian beam (illumination region 3) 107 having a substantially Gaussian distribution as shown in FIG. Lens) 201, spatial filter 202, imaging lens 203, ND filter 207, beam splitter 204, etc. Detectors 205, 206 having a light receiving surface corresponding to the detection region 4, including a detection optical system that passes through and forms an image based on reflected diffracted light (or scattered light), a TDI image sensor, a CCD image sensor, and the like. And an arithmetic processing unit 400 that detects defects such as foreign matters based on image signals detected from the detectors 205 and 206. The defect inspection apparatus includes an automatic focus control system so that the surface of the inspection object 1 is imaged on the light receiving surfaces of the detectors 205 and 206.

The specific configuration of the illumination light source 101 and the illumination optical systems 102 to 105 includes, as shown in FIG. 36, a concave lens or convex lens 102 that enlarges the beam diameter of, for example, a laser beam 1006 emitted from the illumination light source 101, and a concave or convex lens. A collimating lens 103 for converting the beam expanded at 102 into a substantially parallel light beam, and the substantially parallel light beam converted by the collimating lens 103 are converged in the y-axis direction and shown on the inspection object 1 as shown in FIG. In this way, the illumination lens 104 (optical system having a focusing function in the y-axis direction) 104 is irradiated with a slit-shaped Gaussian beam (illumination region 3) 1007 having a substantially Gaussian distribution of illuminance.
The concave or convex lens 102 and the collimating lens 103 constitute a beam expander that expands the beam diameter. As the illumination optical systems 102 to 104, a beam expander composed of a collimator lens, a concave lens, and a receiver lens, and a substantially parallel light beam converted by the beam expander are focused in the y-axis direction on the object 1 to be inspected. 36, a conical lens (an optical system having a focusing function in the y-axis direction) 104 irradiated with a slit-shaped Gaussian beam (illumination region 2) 1007 having a substantially Gaussian distribution as illuminance, and the conical lens 104 In this case, the slit-shaped Gaussian beam 1007 can be reflected to irradiate the object 1 to be inspected from an oblique direction.

By the way, with this configuration, the distance b between the concave or convex lens 102 and the collimating lens 103 or the distance between the concave lens and the receiver lens is variably set, so that the illuminance in the x direction has a substantially Gaussian distribution. The illumination width can be set variably. That is, by adjusting the beam expander, it is possible to variably set the length Lx in the x direction of the illumination region (slit-shaped light beam 1007) 3 having a substantially Gaussian distribution as illuminance. Further, the width Ly in the y direction of the focused illumination region (slit-shaped Gaussian beam 1007) 3 can be set variably by changing the distance between the conical lens 104 and the object 1 to be inspected. .
A detection area 4 shown in FIG. 37 indicates a detection area on the inspection object 1 by a TDI image sensor or a CCD image sensor. For example, in the case of a TDI image sensor, each pixel size is, for example, 27 μm × 27 μm, and is configured by, for example, 64 rows in the time delay integration (TDI) direction, for example, 64 rows and 64 × 4096 CCD image sensors in the MUX direction operating in the TDI mode. The That is, as shown in FIG. 38, the TDI image sensors 205a and 206a have n (for example, 64) stages of line sensors. The line rate, which is the amount of information output from the sensor, is the same as that of the line sensor, but for each line rate rt, the accumulated charges are sequentially transferred to the lines 1, 2,. By synchronizing the feed rate of the y stage 302 that moves the inspection object 1 in the y-axis direction with the line rate, for example, the optical image 6 based on the scattered light or diffracted light from the minute foreign matter 5 reaches the line n. Therefore, even a defect such as a very small foreign object can be detected with high sensitivity. This image sensor basically detects the sum of scattered light or diffracted light intensity until an image of a defect such as a minute foreign substance reaches line 1 to line n. Scattered light or diffracted light from the same point on the target substrate is completely incoherent in time.

  As described above, the beam emitted from the illumination light source 101 is converted into a slit-shaped Gaussian beam beam 1007 by the illumination optical systems (irradiation optical systems) 102 to 104, and the stage is formed by using the converted slit-shaped beam 1007. Irradiation is performed, for example, from an oblique direction so that the illumination region 3 is formed on the surface of the substrate 1 to be inspected 301 to 303. The detectors 205a and 206a constituted by a TDI image sensor or the like have a line rate rt synchronized with the feed speed while moving the substrate 1 to be inspected in the y-axis direction by moving the y stage 302 in the y-axis direction. By sequentially transferring the charges accumulated in each pixel, the width of the detection region 4 is obtained while capturing a light image of the detection region 4 on the inspection target substrate 1 imaged by the detection optical systems 201-204. Scanning with H to detect each pixel (element), and processing the detected signal by the arithmetic processing unit 400, it is possible to detect defects such as minute foreign matters existing in the detection region 4 with high sensitivity and high speed. Can be inspected.

  As described above, by using the TDI image sensors 205a and 206a, it is possible to obtain the total sum of the illuminances of scattered light or diffracted light (light quantity = illuminance × time) generated from defects such as minute foreign matter, and the sensitivity can be improved. it can. Further, the TDI image sensor irradiates the irradiation region 3 with the slit-shaped light beam 1007 at a time and moves the substrate 1 to be inspected in the y-axis direction in synchronization with the line rate rt of the TDI image sensors 205a and 206a. By receiving light in the detection area 4, it is possible to inspect for defects such as minute foreign matter existing in the detection area 4 having a wide width H at high speed.

  Furthermore, a fourth embodiment according to the present invention for inspecting defects such as extremely small foreign matters of about 0.1 μm or less with high sensitivity and at high speed will be described. That is, if a defect such as a very small foreign substance of about 0.1 μm or less is to be detected with high sensitivity, the intensity of scattered light or diffracted light from the defect such as a very small foreign object received at each pixel of the TDI image sensor 302a. And the size of each pixel on the substrate 1 to be inspected must be about 1 μm × 1 μm or less. Thus, in order to make each pixel size on the substrate 1 to be inspected about 1 μm × 1 μm or less, when each pixel size of the TDI image sensor is 27 μm × 27 μm, for example, the detection optical systems 201 to 201 such as an objective lens are used. The imaging magnification M of 204 may be increased to about 27 times or more, and can be realized. When the TDI image sensors 205a and 206a are configured with 26 × 4096 CCD image sensors, the detection area 4 has W = 26 μm or less and H = 4096 μm or less. Further, the detection optical systems 201 to 204 for forming a light image by scattered light or diffracted light obtained from the surface of the substrate 1 to be inspected on the light receiving surfaces of the TDI image sensors 205a and 206a are constituted by an objective lens or the like. Then, MTF (Modulation Transfer Function) (a change in the contrast of the image of the sine wave pattern as a function of spatial frequency) as it goes to the periphery based on the lens aberration as compared to the center of the lens (optical axis 2001). It has the property of deteriorating. Therefore, the end (peripheral) pixels 205ae and 206ae that are the farthest away from the optical axis 2001 on the light receiving surfaces of the TDI image sensors 205a and 206a shown in FIG. 38A and have the lowest MTF, that is, the detection regions shown in FIG. It is necessary to increase the intensity of scattered light or diffracted light from a defect such as a very small foreign substance located at the end (periphery) that is farthest from the optical axis 2001 of No. 4 and has the lowest MTF.

  Incidentally, as shown in FIG. 37, the illuminance of the slit-shaped Gaussian beam 1007 irradiated on the surface of the substrate 1 to be inspected by the illumination light source 101 and the illumination optical systems 102 to 104 in the irradiation region 3 has a normal Gaussian distribution as shown in FIG. Therefore, illumination outside the detection area 4 is wasted, but it is necessary to illuminate the illumination area 3 wider than the detection area 4.

  Therefore, from such a state, the present invention effectively utilizes the light amount emitted from the illumination light source 101 without increasing the illuminance emitted from the illumination light source 101, and is farthest from the optical axis 2001 in the detection region 4. In other words, the illuminance located at the end (periphery) where the MTF is most decreased is increased to detect defects such as extremely small foreign matters of about 0.1 μm or less with high sensitivity. That is, an inexpensive illumination light source that emits the minimum necessary illuminance (laser light source such as semiconductor laser, argon laser, YAG-SHG laser, and excimer laser, discharge tube such as xenon lamp and mercury lamp, filament light source such as halogen lamp, etc. 101) is used to increase the illuminance located at the end (periphery) where the MTF is most lowered farthest from the optical axis 2001 of the detection region 4 by the illumination optical systems 102 to 104, and the efficiency is high. It is to realize lighting.

  That is, the present invention specifically detects when the illumination region 3 of the substrate 1 to be inspected 1 is irradiated with the slit light beam 1007 having a Gaussian illuminance by the illumination light source 101 and the illumination optical systems 102 to 104. The illumination optical systems 102 to 104 are adjusted (controlled) to determine the illumination width so that the illuminance at the periphery of the region 4 is maximized. Here, when the illuminance of the slit beam 1007 has a Gaussian distribution, the following (Equation 6) is used as shown in FIG. 37, and the illuminance is maximized at the outermost periphery of the illumination area. (Equation 7) shown below.

この場合、ＴＤＩイメージセンサ２０５ａ、２０６ａの受光面が対応する検出領域４のｘ軸方向の最外周（端部）での照度ｆ（ｘ０）は、中心部ｆ（０）の約６０．７％で最大となる。即ち、（数７）式において、ｘ０＝σ（σ＝１で、ｘ０＝１）のとき、最大値ｆ（ｘ０）＝０．６０７ｆ（０）となる。なお、上記（数６）式において、ｘ０＝０．８σ〜１．２σ（σ＝１で、ｘ０＝０．８〜１．２（ガウスビーム光束１００７について照明光学系１０２〜１０４による±２０％程度の整形誤差を許容する。））のとき、ｆ（ｘ０）＝０．４９ｆ（０）〜０．７３ｆ（０）となる。また、上記（数６）式において、０．８ｘ０〜１．２ｘ０＝σ（σ＝０．８〜１．２（ガウスビーム光束１００７について照明光学系１０２〜１０４による±２０％程度の整形誤差を許容する。）で、ｘ０＝１）のとき、ｆ（ｘ０）＝０．４６ｆ（０）〜０．７１ｆ（０）となる。従って、照明光学系１０２〜１０４によるガウスビーム光束１００７のｘ０＝σ（σ＝１で、ｘ０＝１）にする整形誤差として±２０％程度許容すると、検出領域４において中心部（光軸２００１）の照度ｆ（０）に対する周辺部（外周部）の照度ｆ（ｘ０）の比は、０．４６〜０．７３（ｆ（ｘ０）＝０．４６ｆ（０）〜０．７３ｆ（０））となる。なお、照明光学系１０２〜１０４によるガウスビーム光束１００７のｘ０＝σ（σ＝１で、ｘ０＝１）にする整形誤差として±１０％程度許容すると、検出領域４において中心部（光軸２００１）の照度ｆ（０）に対する周辺部（外周部）の照度ｆ（ｘ０）の比は、０．５４〜０．６７（ｆ（ｘ０）＝０．５４ｆ（０）〜０．６７ｆ（０））となる。

In this case, the illuminance f (x0) at the outermost periphery (end) in the x-axis direction of the detection region 4 corresponding to the light receiving surfaces of the TDI image sensors 205a and 206a is about 60.7% of the central portion f (0). Is the largest. That is, in the formula (7), when x0 = σ (σ = 1, x0 = 1), the maximum value f (x0) = 0.607f (0). In the above equation (6), x0 = 0.8σ to 1.2σ (σ = 1, x0 = 0.8 to 1.2 (with respect to the Gaussian beam 1007, ± 20% by the illumination optical systems 102 to 104) When a degree of shaping error is allowed))), f (x0) = 0.49f (0) to 0.73f (0). Further, in the above formula (6), 0.8 × 0 to 1.2 × 0 = σ (σ = 0.8 to 1.2 (with respect to the Gaussian beam 1007, a shaping error of about ± 20% by the illumination optical systems 102 to 104). In the case of x0 = 1), f (x0) = 0.46f (0) to 0.71f (0). Accordingly, when a shaping error of about ± 20% is allowed as a shaping error to make x0 = σ (σ = 1, x0 = 1) of the Gaussian beam 1007 by the illumination optical systems 102 to 104, the center portion (optical axis 2001) in the detection region 4 is allowed. The ratio of the illuminance f (x0) of the peripheral part (outer peripheral part) to the illuminance f (0) of 0.46 to 0.73 (f (x0) = 0.46f (0) to 0.73f (0)) It becomes. Note that if about ± 10% is allowed as a shaping error to make x0 = σ (σ = 1, x0 = 1) of the Gaussian beam 1007 by the illumination optical systems 102 to 104, the center portion (optical axis 2001) in the detection region 4 is allowed. The ratio of the illuminance f (x0) of the peripheral portion (outer peripheral portion) to the illuminance f (0) of 0.54 to 0.67 (f (x0) = 0.54f (0) to 0.67f (0)) It becomes.

  In any case, in the detection region 4, the ratio of the illuminance f (x0) of the peripheral portion (outer peripheral portion) to the illuminance f (0) of the central portion (optical axis 2001) is 0.46 to 0.73. By shaping the Gaussian beam flux 1007 by the illumination optical systems 102 to 104, it is possible to effectively utilize the beam emitted from the illumination light source 101 to bring the illuminance at the periphery of the detection region 4 close to the maximum. In the graph shown in FIG. 39, the x-axis direction of the detection region 4 when the illumination width in the x-axis direction, that is, the standard deviation σ is changed without changing the amount of light that is the sum of the illuminances emitted from the illumination light source 101 is shown. The change of the illuminance (light quantity per unit area) f (x0 = 1) at the outer peripheral portion (x0 = 1) of FIG.

In the graph shown in FIG. 40, the illumination width, that is, the standard deviation σ is set to σ = 0.5, σ = 1, σ = 2 without changing the amount of light that is the sum of the illuminances emitted from the illumination light source 101. The change in the illuminance (light quantity per unit area) f (x0) at the coordinate x0 in the x-axis direction of the detection region 4 when changed to is shown. As apparent from FIGS. 39 and 40, in order to substantially maximize the illuminance at the outer peripheral portion (x0 = 1) in the x-axis direction of the detection region 4, it is based on the Gaussian distribution by the illumination optical systems 102 to 104. It is sufficient to illuminate so that the width of illumination in the x-axis direction is approximately σ = 1 (standard deviation σ = x0). That is, as shown in FIG. 37, when the length from the optical axis center of the detection region 4 to the outer peripheral portion in the x-axis direction is x0, the illumination optical systems 102 to 104 substantially detect the standard deviation σ = x0 (detection). The illumination region 3 (with respect to the substrate 1 to be inspected) is shaped into a slit-shaped beam 1007 having a Gaussian distribution of illuminance, which is the length from the center of the optical axis of the region 4 to the outer periphery in the x-axis direction). Lx and Ly may be illuminated as a region where the illuminance f is 0.2 or more with f (0). Actually, when a TDI image sensor or a two-dimensional linear image sensor is used as the detectors 205 and 206, the pixel farthest from the optical axis 2001 and having the lowest MTF is the corner of the detection region 4 (the TDI image sensor). In this case, the pixels 205ac and 206ac located at the corners shown in FIG. 38 correspond to each other.) Therefore, x0 is set to √ ((H / 2) ² + (W / 2) ² ). Is desired. If W can be ignored, x0 = (H / 2). H and W indicate the width (length) in the x-axis direction and the width in the y-axis direction of the detection region 4 on the substrate to be inspected. The width in the x-axis direction of the light receiving area (imaging area) in the TDI image sensor or the two-dimensional linear image sensor is indicated by (H × M), and the width in the y-axis direction is indicated by (W × M). Note that M represents an imaging magnification by the imaging optical systems 201-204.

As described above, the outer periphery of the detection region 4 in the x-axis direction (when using a TDI image sensor or a two-dimensional linear image sensor, the pixel farthest from the optical axis 2001) is x0 (= √ ((H / 2) ² + (W / 2) ² ) or (H / 2)), the illumination optical systems 102 to 104 shape the beam into a slit-shaped beam 1007 having a Gaussian illuminance of approximately σ = x0. By illuminating the inspection target substrate 1 as an illumination area 3 (Lx and Ly are areas where the illuminance f is 0.2 or more with f (0)), a special illumination light source with high power is not used. , Inexpensive ordinary illumination light sources (laser light sources such as semiconductor lasers, argon lasers, YAG-SHG lasers, excimer lasers, discharge tubes such as xenon lamps and mercury lamps, filaments such as halogen lamps, etc. 101) can be used to realize efficient illumination, and as a result, minute foreign matter received by pixels in the periphery of the detectors 205 and 206 where the MTF decreases most by the detection optical systems 201 to 204. It is possible to increase the intensity of scattered light or diffracted light from defects such as 0.1 to 0.5 μm. And at high speed (with high throughput). It should be noted that the illuminance is different in the relationship of (f (x0) = 0.46f (0) to 0.73f (0)) between the central portion and the peripheral portion of the detection region 4, particularly in the x-axis direction. In the image processing unit 400, the image signal obtained from the same pixel column in the x-axis direction in the detection region detected by the detectors 205 and 206 such as the TDI image sensor by moving the inspection object 1 in the y-axis direction. Since the comrades will be compared, there will be almost no influence of the difference in illuminance between the central part and the peripheral part. Then, in the image processing unit 400, the inspected object 1 is moved in the y-axis direction, and each chip repeated with the same circuit pattern based on the image signal detected from the detectors 205 and 206 such as a TDI image sensor or the like. By extracting a difference image signal for each cell and determining the extracted difference image signal based on a desired determination criterion, it is possible to detect and inspect defects such as foreign matter.

  Here, what is important is that the illuminance (light quantity) in the peripheral portion of the detection region 4 is substantially maximized, and the means for that purpose is the illumination optical system 102 to 104 in the above embodiment. Although the width is changed, other means such as changing the shape of the secondary light source of the illumination by the illumination optical systems 102 to 104, or changing the size of the Fourier transform forming the secondary light source, etc. It may be a means. Further, since a DUV (Deep Ultra-violet) laser light source is used as the illumination light source 101, it is necessary to use image sensors 205 and 206 that are sensitive to DUV. However, when the surface irradiation type TDI image sensor shown in FIG. 41A is used as the image sensors 205 and 206, the incident light is transmitted through the cover glass 805, and the oxide film (SiO 2) located at the gate 801 between the metal films 802. ) Since it passes through 803 and enters the CCD formed on the Si substrate 804, short-wavelength incident light attenuates and has almost no sensitivity to wavelengths of 400 nm or less, and DUV light cannot be detected as it is. Therefore, in order to obtain DUV sensitivity with a front-illuminated image sensor, there is a method in which the oxide film 803 in the gate 801 is thinned to reduce the short wavelength attenuation. As another method, an organic thin film coating is applied to the cover glass 805 so that when DUV light is incident, visible light is emitted accordingly, so that DUV light can be emitted by an image sensor that is sensitive only to visible light. There is a way to detect.

On the other hand, as the image sensors 205 and 206, as shown in FIG. 41 (b), a back-illuminated TDI image sensor configured such that the thickness of the Si substrate 804 is thinned and light is incident from the thinned back side. By using light from the back side without a gate structure, the DVD quantum efficiency is set to about 10% or more, the quantum efficiency is high, the dynamic range is large, and the sensitivity is also obtained at a wavelength of 400 nm or less. it can. Further, the sensitivity can be increased by using TDI (Time Delay Integration) for the image sensors 205 and 206 as described above.
As described above, according to the fourth embodiment, the detector 205, 206 such as a TDI image sensor is adapted so that the MTF decreases as the distance from the optical axis 2001 in the detection optical system 201-204 increases. By increasing the illuminance at the periphery of the detection region 4 to be detected and improving the efficiency of illumination, an inexpensive laser light source or the like is used, and the substrate is about 0.1 to 0.5 μm on the inspection target substrate such as an LSI wafer. Even a very small foreign substance having a size of about 0.1 μm or less can be detected with high sensitivity and high throughput.
Further, according to the fourth embodiment, an optical image based on UVD (far ultraviolet) laser light such as excimer laser light obtained from the substrate to be inspected can be received by the TDI image sensor to 0.1 to 0.1. A minute foreign matter of about 0.5 μm can inspect even a very small foreign matter of about 0.1 μm or less.

Next, an example of the image processing unit 400 common to the first to fourth embodiments according to the present invention will be described.
In a device such as LSI that is the actual substrate 1 to be inspected, the detection signals obtained from the detectors 205 and 206 vary due to subtle differences in processes that do not cause defects, noise during detection, and the like. . That is, as shown in FIG. 27A, the signal levels of the corresponding pixels between the chips 71 and 72, for example, 73 and 74 are not the same, and variations occur. Specifically, as shown in FIG. 27B, the location of the detection signal depends on the location where the pattern structure is different (for example, in the case of a memory LSI, the memory cell area, the peripheral circuit area, other areas, etc.) 75, 76, 77, etc. Variation will be different. As a result, a small defect that causes a smaller signal change can be detected in a small variation portion, whereas only a large defect that causes a large signal change can be detected in a large variation portion.
Therefore, the image processing unit 400 according to the present invention is characterized by calculating variation (standard deviation) between corresponding chips for each pixel in the chip, and using the value for setting a threshold value. The small area has a small threshold value, and the large area has a large threshold value so that a defect such as a foreign object is determined and inspected. As a result, the threshold value at a small variation location (for example, a memory cell region in the case of a memory LSI) can be reduced without being affected by the large variation region. As a result, fine foreign matters of 0.1 μm or less can be removed. Can also be detected.

  FIG. 28 shows a first embodiment of the image processing unit 400. The first embodiment of the image processing unit 400 is stored for each column pixel obtained in synchronism with the movement of the inspected substrate 1 in the y-axis direction from the image sensors 205 and 206 including a TDI image sensor or the like. An A / D converter 401 for AD-converting an image signal indicated by a gray value, a start / stop command circuit 416 for taking a sampling timing, a data memory 404, a maximum / minimum removal circuit 405 for removing signals of maximum and minimum levels, and a signal A square calculation circuit 406 that calculates the square of level s, a calculation circuit 407 that calculates signal level s, a number counting circuit 408, a square sum calculation circuit 409 that integrates the square of s, and a sum calculation circuit that integrates s 410, a counting circuit 411 that integrates to obtain n, a positive threshold value (upper limit determination criterion) calculation circuit 412, and a negative threshold value (lower limit determination criterion) calculation circuit 41 The detection signal temporarily stored in the data memory 404 is compared with the positive threshold value calculated and set by the positive threshold value calculation circuit 412, and a comparison circuit 414 that outputs a signal indicating a defect such as a foreign object is output to the data memory 404. The temporarily stored detection signal is compared with the negative threshold value calculated and set by the negative threshold calculation circuit 413, and a signal indicating a defect such as a foreign object is output, and the comparison circuit 415 and the comparison circuits 414 and 415 output the signal. Output means 417 for adding a position coordinate in the coordinate system set for the substrate 1 to be inspected to a signal indicating a defect such as a foreign object, and outputting a detection result with information on the substrate 1 to be inspected. Consists of. The maximum / minimum removal circuit 405 is not necessarily required. When the maximum / minimum removal circuit 405 is not used, all the image data (including image data indicating foreign matter) detected in the calculation of the threshold level is used. And can be detected stably. On the other hand, the created threshold value makes it impossible to inspect the foreign matter in the area where the threshold value is created. Therefore, the threshold value of the region to be inspected must be created in the corresponding region of another chip row in the substrate 1 to be inspected. As a result, it is necessary to create the threshold value and the inspection of the foreign matter on separate lines, which slightly increases the throughput. In particular, when the number of chips is small, it is preferable to create a threshold value using image data over a plurality of lines. In this case, the data capture position is designated by the start / stop command means 416.

  Further, the output means 417 is provided with a CPU for controlling the entire defect inspection apparatus for foreign matter and the like according to the present invention. Reference numerals 406 to 411 are used to obtain a background signal variation σ for each predetermined area in the chip. Then, the positive side for extracting a signal indicating a defect such as a foreign substance by the positive side threshold calculation circuit 412 and the negative side threshold calculation circuit 413 based on the obtained variation σ of the background signal for each predetermined area in the chip. Negative thresholds Th (H) and Th (L) are set. These threshold values 406 to 413 are the threshold setting circuit 424. On the other hand, the data memory 404 is for temporarily storing the detected digital image signal until the threshold value is set by the threshold value setting circuit 424. The position coordinates in the coordinate system set for the substrate 1 to be inspected are those of the stage measured by a length measuring device (not shown) with the reference mark provided on the substrate 1 to be inspected as the origin. It is obtained based on the displacement and a readout signal (scanning signal) from the TDI sensor or the like. Reference numeral 421 denotes output means for displaying and outputting the positive threshold Th (H) indicating the variation (standard deviation σ) on, for example, a display means. By providing the display means 421, it is possible to determine whether or not the threshold value is appropriate for each region in the chip while looking at the defect extraction output of foreign matters and the like extracted from the comparison circuits 414 and 415.

Here, the detection result output means 417 records on a display such as a CRT, what is printed as a hard copy, hard disk, floppy disk, magneto-optical recording medium, optical recording medium, LSI memory, LSI memory card, etc. A network connected to a management system that manages other inspection apparatuses or inspection systems or manufacturing process apparatuses and manufacturing lines. In addition, the output means 417 is provided with a CPU that controls the entire defect inspection apparatus for foreign matter and the like according to the present invention.
Here, the A / D converter 401 converts signals output from the detectors 205 and 206 such as TDI sensors into pixel signals expressed as digital signals. The A / D converter 401 may be in the same substrate in the detection signal processing system 400 or in the vicinity of the detectors 205 and 206 such as TDI sensors in the detection optical system 200. In the case of being placed near the detectors 205 and 206, since it is digitized, there is an effect of reducing noise during transmission, but there is also a demerit that the number of signal transmission cables increases.

  Here, the contents of signal processing performed by the threshold setting circuit 424 will be described with reference to FIG. FIG. 27A shows an arrangement example of the chips 71 and 72 on the wafer 1. In many LSI manufacturing, these chips are manufactured repeatedly. Occasionally, multiple (2 to 4 etc.) chips may be manufactured simultaneously in a single exposure. Therefore, the same pattern is produced at the same position between these chips. Therefore, the detection signals of the corresponding positions of these chips are essentially the same. The signal of the pixel (i, j) in this chip (f, g) is assumed to be s (i, j, f, g). As described above, the signal levels should match in the corresponding pixels.

However, in actuality, variations in detection signals s of corresponding pixels between chips occur due to subtle differences in processes that do not cause defects, noise during detection, and the like. In addition, even within the chip, variations are different at different locations of the pattern structure. Therefore, the variation (standard deviation σ (s, f, g)) of the detection signal s ((i, j, f, g) between the corresponding positions of the chip is obtained based on the following equation (8). The threshold values Th (H) and Th (L) are set.
Th (H) = μ (s, f, g) + m1 · σ (s (i, j, f, g), f, g)
Th (L) = μ (s, f, g) −m1 · σ (s (i, j, f, g), f, g) (Equation 8)
Here, Th (H) is a threshold value calculated and set by the positive-side threshold value calculation circuit 412, and Th (L) is a threshold value calculated and set by the negative-side threshold value calculation circuit 413. μ (s, f, g) is an average value when the values of f and g of the signal s calculated based on the following equation (9) are changed.

μ (s, f, g) = Σs / n (Equation 9)
Σs (i, j, f, g) is calculated by a calculation circuit 407 that calculates a signal level s and an integration circuit 410 that integrates s, and n is calculated by a number counting circuit 408 and a counting circuit 411. σ (s, f, g) represents a standard deviation when the values of f and g of the signal s calculated based on the following equation (10) are changed. m1 is a magnification (coefficient).

σ (s, f, g) = √ (Σs ² / n−Σs / n) (Equation 10)
Σs (i, j, f, g) ² is calculated by a circuit 406 that calculates the square of the signal level s and a circuit 409 that integrates the square of s. In this way, the threshold value is drawn where the standard deviation σ (s, f, g) is multiplied several times. The magnification m1 is generally considered to be about 6. This is because the probability of occurrence of 6σ or more is about 1 × 10 (−11) power. This probability is, for example, that the number of images when the inside of a wafer of φ300 mm is detected with a pixel size of 2 × 2 microns is the power of 7 × 10. This is obtained from the fact that the whole area becomes less than one pixel. Of course, this value does not necessarily have to be 6, and it goes without saying that another value may be used in order to exhibit the effects of the present invention. Another magnification may be selected because the number of allowed false alarms is not necessarily less than one.

FIG. 4 shows a second embodiment of the image processing unit 400. The difference from the first embodiment is that the image signal of one chip is delayed by the data memory 402, and the difference Δs = {s (i, j, f, g) − s (i, j, f + 1, g)} is to be extracted. Therefore, in the comparison circuits 414 and 415, the difference signal Δs = {s (i, j, f, g) −s (i, j, f + 1, g)} is expressed by the following equation (11). The upper limit threshold Th (H) and the lower limit threshold Th (L) shown are compared to extract a signal indicating a defect such as a foreign object. Therefore, in the threshold setting circuits 406 to 413, the upper limit threshold Th (H) and the lower limit threshold Th (L) are set based on the following equation (11).
Th (H) = + m1 · σ (s (i, j, f, g) −s (i, j, f + 1, g), f, g)
Th (L) = − m1 · σ (s (i, j, f, g) −s (i, j, f + 1, g), f, g)
(Equation 11)
In this case, the standard deviation σ (Δs, f, g) of the difference image between adjacent chips is calculated based on the following equation (Equation 12). ΣΔs is calculated by a calculation circuit 407 that calculates a signal level Δs and an integration circuit 410 that integrates Δs, and n is calculated by a number counting circuit 408 and a counting circuit 411. ΣΔs ² is calculated by a square calculation circuit 406 that calculates the square of the signal level Δs and a square integration circuit 409 that integrates the square of Δs.

σ (Δs, f, g) = √ (ΣΔs ² / n−ΣΔs / n) (Equation 12)
As described above, by using the difference image Δs between adjacent chips, the standard deviation σ becomes small even when the detected image signal has a distribution in the chip, and a defect inspection such as a foreign substance with higher sensitivity can be performed.

Further, for example, when the process conditions differ in stages from the center toward the periphery in the wafer, the signal level in the wafer surface changes in stages from the center toward the periphery. As a result, the variation (standard deviation σ (s, f, g)) is greatly calculated in the calculation of the threshold value by the equation (8). In such a case, the variation in signal difference only between adjacent chips (standard deviation σ (Δs, f, g)) is actually not as great as in equation (8), but in reality it is detected with a smaller threshold. Is possible. Therefore, by calculating based on the difference value Δs as in the formulas (11) and (12), it is possible to draw a lower level threshold value. As an improvement method of this method, the threshold value may be calculated by the following equation (13).
Th = m1 · σ (| s (i, j, f, g) −s (i, j, f + 1, g) |, f, g) (Equation 13)
Here, | Δs | means the absolute value of the difference signal Δs. In this case, the difference processing circuit shown in FIGS. 29 and 30 is an absolute value difference processing circuit 403 that takes the absolute value | Δs | of the difference image of the adjacent chip. Further, the threshold value calculation circuit 423 calculates and sets the absolute value Th of the threshold value. Further, in the comparison circuit 414, the absolute value of the difference is compared with the threshold value Th, and a defect signal such as a foreign matter is extracted.

  In this case, the standard deviation σ (| Δs |, f, g) of the difference image of the adjacent chip is calculated based on the following equation (Equation 14). Σ | Δs | is calculated by a calculation circuit 407 that calculates a signal level | Δs | and an integration circuit 410 that integrates | Δs |.

σ (| Δs |, f, g) = √ (ΣΔs ² / n−Σ | Δs | / n) (Expression 14)
By the way, the fourth embodiment shown in FIG. 30 is obtained by adding a memory position controller 422 to the third embodiment shown in FIG. The memory position controller 422 designates coordinates on the wafer with respect to the detection signal s or the difference signal Δs. That is, pixels between chips for which the standard deviation σ is obtained can be arbitrarily designated based on the coordinates on the wafer. In addition, since the coordinates on the wafer can be arbitrarily designated, the standard deviation σ can be obtained from the surroundings of the pixel of interest between chips. In the third embodiment shown in FIG. 29, the position coordinates on the wafer are calculated from the signal count result. In this case, the chips for obtaining the standard deviation σ may be arranged in a horizontal row, but the standard deviation cannot be obtained from the corresponding points of the two rows of chips.

  Therefore, as in the third embodiment, the memory position controller 422 calculates the position coordinates of the detection signal s or the difference signal Δs flowing from the signal of the stage coordinate system or the like obtained from the stage controller 305, and By providing the calculated result to the square sum calculation circuit 409 having a memory function, the sum calculation circuit 410, and the counting circuit 411, the result is stored on the signal storage destination, that is, on the coordinates of the detection signal. With this configuration, the number of samples for standard deviation calculation can be increased even when the number of chips in a row is small around the wafer, and the threshold calculation circuit 423 can perform stable threshold calculation.

As described above, the absolute value of the difference in the difference processing circuit 403 has no sign, so that the capacity of the memory 404 and the like can be reduced. Further, from the result of calculating the absolute value, the calculated standard deviation σ is calculated to be smaller than the calculation result from the difference value, and in order to make the occurrence probability 1 × 10 (−11) power, that is, on the normal distribution In order to obtain 6σ, it is necessary to increase the magnification by about 10 times, which is a magnification approximately 1.66 times larger. It may be considered that σ is calculated to be 0.6 times smaller than the calculation from the difference value.
In addition, this method does not leave a threshold for the signal level s, which causes problems in process management and failure analysis. Therefore, as shown in FIGS. 29 and 30, a circuit for calculating the threshold level of the position (i, j) in the chip as a threshold map is provided. In this circuit, the threshold map includes the standard deviation σ × m1 (magnification) obtained based on the equation (14) in the threshold calculation circuit 423 and the absolute value of the difference signal calculated in the average value calculation circuit 425. Is calculated by obtaining a sum ((m1 × σ) + Σ | Δs | / n) of these values by the threshold value calculation circuit 418 for the detected image signal s or the difference signal Δs. The The result is a threshold map storage means having a memory corresponding to each pixel (i, j) in the entire chip according to the position data (i, j) calculated from the positions of the stages 301, 302 and the sensors 205, 206. It is stored in 419, and is displayed and output as required by the user by a threshold map output means (display means or the like) 421. Further, the display unit 421 can display the defect output such as foreign matter extracted from the comparison circuit 414 and the threshold map to determine whether the threshold is appropriate. Further, by providing the threshold map information to the output means 417, it is possible to output a defect output such as a foreign substance extracted from the comparison circuit 414 and the threshold map.

  Since this threshold level is related to the condition of the background, information such as whether the background is a repetitive pattern, a rough area, a thin film area, or an area with a small pattern size is used. And corresponding. Therefore, it is important to analyze to which threshold level the detected foreign matter is detected. Therefore, for example, it is meaningful to output the detected foreign matter data by displaying the threshold value of the position where the foreign matter is present on the display means 421 in addition to the foreign matter signal level. For this purpose, the threshold map calculated here is necessary.

  Incidentally, it is considered that the signal level s of the foreign matter is not (difference value + threshold) but the difference value Δs.

In addition, as background data of the position where the foreign object is detected, not only the threshold level described above, but also an area in the chip (memory area, logic circuit area, power supply area, It may be information on an area without wiring). For this purpose, an area map in the chip calculated from the design data may be created, and encoded from the coordinates in the chip when the foreign object is displayed, such as threshold data, or displayed as words and output. In addition, any of the above background data may be displayed and output in the form of a foreign matter map for each background data and the number of foreign matters for each background data.
As described above, the basic idea of the present invention is to obtain the magnitude of the signal variation and determine the threshold according to the obtained signal variation magnitude. The threshold value may be calculated by the threshold value setting circuit 424 for each pixel in the chip from the value. At this time, this calculation is performed in advance in the same process of the LSI of the same product type, and the result is read into the threshold memory in the threshold setting circuit 424 at the time of inspection and compared with the signal level sequentially input in the comparison circuits 414 and 415. It may be configured to do so. The threshold calculation data may be calculated once for each lot (13 to 25 wafers) or for each wafer.

In the present invention, as described above, the threshold level changes depending on the background state, and thus the threshold level indicates the background state. In other words, the CPU that is the output unit 417 classifies the signal indicating the defect such as the foreign substance based on the threshold level obtained from the threshold map storage unit 419, so that the defect such as the foreign substance adheres to what kind of background. You can know if it existed. For example, the background state can be classified into a region having no pattern, a region of a cell portion, a peripheral pattern portion, and the like. In addition, the CPU shown in FIG. 1 or FIG. 2 obtained based on the CAD information input from the CAD system or the like to the CPU as the output means 417 using the input means 426 constituted by a network, a storage medium, or the like. By using such in-chip region data, it is possible to directly know the state of the ground on which a defect such as a foreign substance exists.
Here, the method of analogizing the ground state from the ground signal level (threshold level) without using the region data has an effect that it is not necessary to set a region in the chip in advance. . In this case, the CPU 417 once obtains the threshold level of the entire chip from the threshold map stored in the threshold map storage means 419, and determines the background level based on the magnitude of the threshold level as an area (for example, Cell part etc.). Here, the region determination from the threshold level can be performed by using the difference Δs between adjacent chips or by calculating from the signal level itself. As described above, the CPU 417 can detect, output, and manage, for example, only the foreign matter or defect on the cell portion after knowing the background state.

  Next, a fifth embodiment of the arithmetic processing circuit 400 will be described with reference to FIG. That is, in the fifth embodiment, after calculating the difference value Δs of the data (detection signal s) of the adjacent chip, the variation (standard deviation σ (Δs, f, g)) around the target pixel is obtained. It is. The fifth embodiment includes delay memories 425 and 426 and a window cut-out circuit 427, and is configured using a so-called pipeline processing system. The peripheral pixel values (Δs (i + 1, j + 1, f, g)) and Δs (i + 1, j, f, g) excluding the center value (Δs (i, j, f, g)) of 406 to 413 , Δs (i + 1, j−1, f, g), Δs (i, j−1, f, g), Δs (i−1, j−1, f, g), Δs (i−1, j, f, g), Δs (i−1, j + 1, f, g), Δs (i, j + 1, f, g)), and variation σ (Δs, f, g) based on the following equation (15) And thresholds Th (H) and Th (L) are calculated and set based on the calculated σ.

σ (Δs, f, g) = √ (ΣΔs ² / 8−ΣΔs / 8) (Equation 15)
Then, the comparison means 414 and 415 compares the set threshold values Th (H) and Th (L) with the central value (Δs (i, j, f, g)) of the previous window to detect foreign matter or the like. Extract defects. The window size here does not necessarily have to be 3 × 3 as shown in the figure, and may be other sizes such as 4 × 4, 5 × 5, 7 × 7, or a plurality of window sizes. It may be configured to calculate. Further, the object to be inspected does not have to be a central value, and may be compared with any one of the windows or an average or sum of a plurality of pixels. The window size should be determined according to the size of the foreign object to be detected or the pattern shape of the background pattern.

  Next, a sixth embodiment in which a threshold value for the absolute sensitivity of the arithmetic processing unit 400 is provided will be described. By providing a threshold of absolute sensitivity based on the above equation (13), the management size of foreign matters or defects in the LSI manufacturing process can be made the same between the processes. Of the detection results (foreign matter coordinates, signal level (difference level)) detected by the detection signal processing circuit 400, the CPU 417 corrects which level is the signal level (desired value ss is desirable). . Specifically, the laser power Pl at the time of inspection, the value ND (%) of the ND filter, the presence or absence of a polarizing plate k (preferably about 1 if it is present, or about 10 if it is not present), the reflectance rb of the base, the oxide film The signal level ss ′ corrected by the following equation (16) may be used as the correction coefficient k (t) based on the thickness of: Since the laser power has a distribution (so-called shading) at the illumination position, it is better to use this distribution Pl (x).

ss ′ = ss / (Pl · ND · k · rb · k (t)) (Expression 16)
Using the correction signal level ss' calculated in this way, the foreign substance size d can be displayed by the display means 421 by using a correspondence function df (ss) between the signal level ss obtained in advance and the foreign substance / defect size d. .

d = df (ss') (Equation 17)
Here, when the foreign matter is particularly small, the relationship that the corrected signal level ss ′ is proportional to the −6th power of the foreign matter size d may be used by using the Mie scattering theory.

Next, a description will be given of an embodiment in which the arithmetic processing circuit 400 determines a defect with a high S / N ratio for not only a minute foreign matter but also a large foreign matter having a spread. By the way, the comparison circuits 414 and 415 for determining defects of the arithmetic processing circuit 400 need to detect not only a minute foreign substance but also a thin or thin film-like foreign substance spread over a range of several microns. However, since this large foreign matter does not necessarily increase the detection signal level, the detection signal in pixel units has a low S / N and may be overlooked.
Therefore, assuming that the average detection signal level of one pixel is S and the average variation is σ / n, the detection signal is obtained by performing extraction and convolution operation in units of n pixels × n pixels corresponding to the size of the large foreign matter. The level is n ² S, the variation is nσ, and the S / N ratio is nS / σ. On the other hand, when an attempt is made to detect large foreign matters in units of one pixel, the detection signal level is S, the variation is σ, and the S / N ratio is S / σ. Therefore, the S / N ratio can be improved by a factor of n by cutting out in units of n pixels × n pixels corresponding to the size of large foreign matter and performing a convolution operation. For a minute foreign substance of about one pixel unit, the detection signal level detected in one pixel unit is S, the variation is σ, and the S / N ratio is S / σ. If a minute foreign substance of about one pixel is cut out in units of n pixels × n pixels and subjected to a convolution operation, the detection signal level is S / n ² , the variation is nσ, and the S / N ratio is S / nσ. Become. Therefore, for a minute foreign substance of about one pixel unit, the signal of the pixel unit as it is can improve the S / N ratio.

As is apparent from the above description, as shown in FIG. 52, an image signal obtained from the image memory 404 is converted into a plurality of operators 520 (for example, one pixel unit) whose pixel unit size for defect determination is changed. Operator 521, operator 522 that cuts out in units of 3 × 3 pixels, operator 523 that cuts out in units of 4 × 4 pixels, operator 524 that cuts out in units of 5 × 5 pixels, and operator 525 that cuts out in units of n × n pixels The level of S, 9S, 16S, 25S, n ² S, where S is the detection signal level of the average of one pixel by performing convolution operations in each of the extracted operators. Are output from the central pixel. On the other hand, each of the multiplication circuits 541, 542, 543, and 544 multiplies the threshold value (m1 · σ) obtained from the threshold value circuit 423 of the threshold setting circuit 424 by 3 times, 4 times, 5 times, and n times. The approximate threshold coefficients 3, 4, 5, and n are estimated from the central limit theorem. In each of the comparison circuits 531, 532, 533, 534, and 535 constituting the comparison circuit 414 ′, 3 is applied to the gradation signal subjected to the convolution operation in each operator and the threshold value (m 1 · σ). Defect determination is performed by comparing with the doubled, 4-fold, 5-fold, and n-threshold values, and a signal indicating the foreign matter is output. That is, for the minute foreign matter of about 1 pixel unit, from the comparison circuit 531, for the foreign matter having a size of about 3 × 3 pixel unit, from the comparison circuit 532, for the foreign matter of about 4 × 4 pixel unit, the comparison circuit 533. Therefore, a foreign substance having a size of about 5 × 5 pixels is detected from the comparison circuit 534, and a foreign substance having a size of about n × n pixels is detected from the comparison circuit 535. Therefore, by taking the logical sum of the signals indicating the foreign substances detected from each of the comparison circuits 531 to 535 in the OR circuit 550, the signals indicating the foreign substances having various sizes are detected with a high S / N ratio. The detection rate can be improved even for a large foreign substance having a small detection signal level and a large spread.

  In addition, after the difference processing circuit 403 ′, an operator that can change the size of the pixel unit for defect determination as described above is provided, and the pixel signal is integrated and output every time the size of the pixel unit is changed. For example, the comparison circuit 414 detects a signal indicating a foreign substance having a size that matches the size of the changed pixel unit. However, in this case, it is necessary to inspect a plurality of times to change the size of the pixel unit, but an accurate value is set as the threshold value. Further, when a plurality of operators capable of changing the pixel unit size for defect determination are provided after the difference processing circuit 403 ′, the storage capacity of the image memory 404 needs to be multiple times. Also, a plurality of threshold setting circuits 424 may be provided, and the threshold value (m1 · σ) obtained from the threshold circuit 423 of the threshold setting circuit 424 is multiplied by an approximate threshold coefficient to generate a threshold value. You may ask for. As described above, in the comparison circuit 414 ′, the size of the pixel unit for defect determination that convolves or integrates the rectangular function is matched with the size of the foreign substance to be detected, so that the detection signal level is small and widened. It is possible to accurately capture a large foreign object that is held.

  Next, an example of a condition setting method in a defect inspection apparatus for foreign matter or the like according to the present invention will be described with reference to FIGS. That is, the defect inspection apparatus for foreign matter or the like according to the present invention has a condition setting sequence as shown in FIG. 42, and the inspection is executed according to the inspection conditions created by this sequence. That is, in step S41, the CPU 417 displays a screen for selecting various modes as shown in FIG. 43 on the display means 421, and uses the input means 426 such as a keyboard and a mouse to make a chip matrix (chip size). The chip start point coordinates and the chip arrangement data such as no chip) S411, conditional mode (a. Area priority, b. Standard, c. Sensitivity priority, d. Selection after sensitivity display) S412, threshold Value pre-selection (a.m1 = 6: false alarm occurrence probability OO%, b.m1 = 10: false alarm occurrence probability OO%, c.m1 = 15: false alarm occurrence probability OO%) S413 or the like mode is selected. In the conditional mode S411, a. Area priority is an inspection condition mode in which a relatively large foreign object can be inspected over a larger area than the standard mode by, for example, reducing the illumination light power. The area where the background level is saturated becomes a non-inspection area substantially, but area priority may be set such that the area of the non-inspection area is, for example, 5% or less. In FIG. 45, area priority indicates a case where foreign matter of about 2.5 μm can be inspected from all areas.

  b. Standard is an inspection condition mode in which foreign matter can be inspected with standard sensitivity. In FIG. 45, the standard indicates a standard mode in which foreign matter having a size of about 90% to about 1.0 μm can be inspected and foreign matter having a size of about 0.2 μm can be inspected.

  c. Sensitivity priority is a mode set to increase the sensitivity so that a minute foreign object can be detected than the standard mode, or an inspection condition mode set to ensure a specified detection sensitivity. In FIG. 45, the sensitivity priority indicates a mode in which foreign matter of about 75% to about 0.5 μm of the inspection area of the entire detection can be inspected, and foreign matter up to 0.1 μm can be inspected. Specifically, by increasing the illumination light power, inspection conditions that can detect a foreign object smaller than the foreign object specified by the detection size designation (in FIG. 45, about 0.1 μm), or the designated detection sensitivity ( In FIG. 45, the power of the illumination light is set to a level at which a foreign matter of about 0.5 μm can secure 75% or more in the inspection area.

d. Selection after sensitivity display refers to the inspection results in the above three modes, the threshold map in the chip, or the relationship between the size of foreign matter (sensitivity corresponding to the threshold) and the inspection area (threshold histogram) Is a mode for selecting an appropriate one.
Area priority weakens the power of illumination light to increase the dynamic range, and as the priority goes to standard and sensitivity priority, the power of illumination light is increased and the dynamic range is lowered. Therefore, in the threshold map, in the area priority mode, there are few non-inspection areas where foreign matters cannot be detected, but only foreign matters up to about 0.5 μm can be inspected. In the case of the standard mode, there are many non-inspection areas that are saturated and cannot detect foreign substances displayed in white in FIG. In the sensitivity priority mode, the saturated non-inspection area in which foreign matter displayed in white in FIG. 45 cannot be detected further increases, but foreign matter of about 0.1 μm can be inspected. The threshold histogram shows the area ratio 471 with respect to the sensitivity and its integrated value 472, and any value may be displayed.
The selection of the preset threshold value can be performed from the occurrence probability of the false information allowed by looking at the occurrence probability (occurrence frequency) OO% of the displayed false information. That is, as described above, since the threshold value is set from the level variation σ of the detected image, the false alarm occurrence probability OO% can be automatically calculated and displayed based on statistical theory in accordance with the magnification m1. It becomes possible. Thereby, the magnification m1, that is, the threshold value can be easily set according to the occurrence probability of the false alarm.

  Next, in step S42, the CPU 417 manually or automatically sets the spatial filter 202 corresponding to the circuit pattern structure in the selected wafer. In step S43, the image of the spatial filter 202 is displayed as shown in FIG. Is visually observed or automatically confirmed by an image formed by the imaging optical system 227 and the TV camera 228 focused on the filter position. If NO, the process returns to step S42 to set the spatial filter 202 again. If so, go to the next step. The spatial filter 202 is configured to change the phase and pitch of the light shielding pattern. As shown in FIG. 46, an arrow 230 indicates a spatial filter observation optical system composed of a half mirror 226, an imaging lens 227, and a TV camera 228, and a beam splitter 204, which are integrally formed. It is configured to be switchable as described above. That is, when detecting a normal foreign object, the beam splitter 204 is positioned on the detection optical axis, and switching is performed so that the half mirror 226 is positioned on the detection optical axis during spatial filter observation. In the case of automatic, as shown in FIG. 19B, the light shielding pattern and the diffracted light detected in the opening 20a are imaged by the TV camera 228, so that the phase and pitch of the light shielding pattern are diffracted light. Can be adjusted to block light. Further, by shifting the position of the TV camera 228 as indicated by an arrow 229, the image of the circuit pattern on the substrate to be inspected can also be observed to match the directionality of the light shielding pattern.

Next, in step S44, the CPU 417 sets the magnification (coefficient) m1 with respect to the standard deviation σ for setting the threshold value Th by using the input means 426 in the range of about 6-15. Next, in step S45, the CPU 417 inputs the detection size of the foreign matter using the input unit 426, thereby specifying the detection size S451, and calculates the laser power that can detect the foreign matter of the specified size. The laser light source 101 is controlled by the control signal 430 so that the calculated laser power is obtained.
Next, in step S46, the CPU 417 scans and inspects the wafer in order to create a threshold value for a part or the entire area of the chip, and uses the threshold value map calculated by the threshold value calculation means 418 as the threshold value. The threshold map (threshold image) shown in FIGS. 44 and 45, or the threshold histogram (sensitivity (for example, the horizontal axis) and the relationship between the inspection areas having the sensitivity) stored in the map storage means 419, Alternatively, an integrated display of this histogram (FIG. 45) is displayed on the display means 421, and the threshold value is set to a desired level (foreign object size to be detected) based on the displayed threshold map. If it is NO, the process returns to step S45 to specify the detection size again. If YES, the process proceeds to the next step.

Next, in step S47, the CPU 417 inspects the entire area of the wafer, and if there is an area where some false information appears, depending on the case, this area may be set as a non-inspection area (inhibit area) and the CAD information in the chip. Or set based on threshold map information. Thereafter, in step S48, based on a command from the CPU 417, the inspected substrate 1 is inspected for foreign matter or the like, and if the arithmetic processing circuit 400 determines that the foreign matter or the like is defective, the detection signal The level and the detected coordinates are stored in the storage device 427.
Next, in step S49, the actual substrate 1 to be inspected is optically observed by using an optical observation microscope 700 which is finally composed of a confocal microscope or an ultraviolet microscope arranged in parallel with the foreign substance inspection apparatus. And confirm whether it is a defect such as a foreign object or a false alarm. By this confirmation, it is possible to confirm for the first time whether or not the condition setting has been optimally set. In particular, in the chip on the substrate 1 to be inspected, a portion where a minute and complicated circuit pattern exists and a portion where color unevenness occurs are mixed, and final confirmation of conditions using the optical observation microscope 700 is performed. It is necessary to do. In the false alarm confirmation in step S49, if NO, in step S50, the threshold setting magnification (coefficient) m1 is increased or decreased depending on the case, and the process returns to step S45. change. If yes, the condition is complete.
Here, even if a part of the above procedure is omitted or the order is changed, the object can be achieved.

  As described above, it is possible to easily and quickly set the optimum condition for the desired foreign substance size (sensitivity) to be detected. Note that the optical observation in step S49 is performed by moving the stages 301 and 302 to detect the foreign matter (including false information) detected on the inspection object 1 at the position of the detection optical system 701 of the optical observation microscope 700 shown in FIG. Is moved and this image is observed. Since the detection system 200 of the present invention has a high-resolution imaging optical system, the coordinate accuracy during this movement is high (particularly, in the dark field illumination systems 102 to 105, the resolution of the detection optical system 200 is high). Because foreign objects smaller than the limit can be detected), it is often impossible to observe with a normal microscope. Therefore, the optical observation microscope 700 has an extremely high resolution, for example, a confocal optical system, or illumination with a short illumination wavelength (for example, 248 nm, 365 nm, 266 nm, or a wavelength in the vicinity thereof, ultraviolet rays or far ultraviolet rays). It is desirable to have an optical system. In other words, the optical observation optical system 700 can obtain an image close to an electron microscope image with an image having a wavelength of around 200 nm as described above. Etc. can also be classified. FIG. 46 shows the configuration of the optical observation microscope 700. A detection optical system 701 having a bright-field or dark-field ultraviolet irradiation optical system and a TDI sensor capable of detecting ultraviolet rays shown in FIG. 41, and A / D conversion is performed on an image detected from the TDI sensor of the detection optical system 701. An image processing system 702, and an image memory 704 for storing the image A / D converted by the image processing system 702 at an address based on the coordinate data of a foreign object (which is considered to be false information) detected from the arithmetic processing circuit 400; , And display means 703 for displaying an image. Accordingly, the stages 301 and 302 are controlled on the basis of the coordinate data of the foreign matter (what is assumed to be false information) detected from the arithmetic processing circuit 400, and an image that is assumed to be false information is displayed on the display means 703 and observed. The false information can be confirmed by the optical observation microscope 700. That is, the position of the detection coordinate stored in the storage device 427 is moved within the field of view of the optical observation microscope 700, the stages 301 and 302 are moved, the image within the field of view is detected by the optical observation microscope 700, and displayed on the display means 703. Alternatively, it is stored in the image memory 704 as numerical image data. This data can also be displayed again when needed. Further, the data stored in the image memory 704 can be provided to the CPU 417 of the arithmetic processing circuit 400, and can be observed later together with the image data transferred from another foreign substance inspection apparatus. In any case, as the optical observation microscope 700, the bright field microscope having the high resolution, the dark field microscope having the illumination optical system 100, the dark field microscope having incoherent illumination, the phase contrast microscope, A confocal microscope may be used.

Further, in the above condition setting, the condition reading can be completed only by inputting and setting the chip matrix in step S41 and the foreign substance size to be inspected in step S45. That is, the input setting of the chip matrix and the foreign matter size (sensitivity according to the foreign matter size may be used) is an essential setting condition in the condition setting.
Also, the filter confirmation in step S43, the setting of the magnification m1 in step S44, the sensitivity confirmation in step S46, the inhibition (non-inspection area) setting in step S47, and the false alarm confirmation in step S49 are optional setting conditions. Also, in setting the threshold value, it is possible to suppress the generation of false alarms by using the threshold value on the stable side (large threshold value). Conversely, by making the threshold value small, some false alarms are generated. However, it is possible to inspect highly sensitive foreign matter. The former is suitable for quality control of wafer processing equipment (finding anomalies), and the latter is used to analyze the occurrence of defective defects (classification of foreign object defects for investigation of the cause of defective occurrence). Turn to.

Next, foreign particle size estimation performed by the CPU 417 from the scattered light image detected from the image sensors 205 and 206, A / D converted by the A / D converter 401 and stored in the image memory 404 will be described with reference to FIG. explain. That is, the signal level of scattered light (difference value ss is desirable) corresponds to the size of particles or scratches that generate scattered light. Accordingly, the CPU 417 corrects the detection signal ss by multiplying the detection signal ss by a correction coefficient k (t) calculated according to conditions such as the laser power, the polarizing plate 208 at the time of inspection, the spatial filter 202, and the illumination angles φ1 and α1. The detected signal ss' can be associated with the foreign substance or defect size d. Therefore, the CPU 417 can use the size information of the foreign matter or defect thus obtained for the detection size designation in step S45 under the above conditions.
In addition, as shown in FIG. 47, the size of the image indicating the foreign matter (the number of pixels indicating the spread of the foreign matter image) and the size of the foreign matter in the detected image detected by the TDI image sensors 205a and 206a and stored in the image memory 404. Since there is a certain tendency, the CPU 417 can associate the particle size of the foreign object by counting the number of pixels indicating the foreign object from the detected image stored in the image memory 404. In particular, even when the size of the foreign matter is about 0.13 μm to 0.2 μm, it can be found that there is a correlation with the size of the image showing the foreign matter, and the size (particle size) of the foreign matter can be estimated. Become.
In addition, when the size of the foreign matter falls within one pixel and the signal level exceeds the dynamic range of the image sensors 205 and 206, the foreign matter size can be estimated by the following method. That is, even when entering one pixel, as shown in FIG. 48 (a), since the image is formed with a spread, the width of the rise and fall of this spread portion (the width of a certain threshold) The signal intensity exceeding the peak level, that is, the dynamic range can be estimated from W). In this case, as shown in FIG. 48B, the cover glass 220 of the image sensors 205 and 206 is made to have a specific surface roughness, thereby causing scattering on the surface of the cover glass 220 to forcibly spread. Thus, it is possible to further easily estimate the size of the foreign matter from the detected image.

  Next, a plurality of inspections by the foreign matter inspection apparatus according to the present invention will be described. That is, in order to increase the dynamic range, for example, a plurality of inspections are performed under the condition that the illumination light power is increased and the standard conditions are reduced, such as area priority, standard, and sensitivity priority. The surface is inspected, and the CPU 417 outputs the result as an inspection defect. The CPU 417 simply integrates the plurality of inspection results, and shows an inspection result map (inspection result map is a drawing in which defect marks are plotted at position coordinates where a defect such as a foreign object is detected in the inspection target substrate 1. .) Can be output. Further, the CPU 417 may not be a map, but may be a foreign matter coordinate list or a list or map showing the foreign matter detection signal level. In addition, as the plurality of inspections, not only to increase the dynamic range, but also, for example, the travel times of the stages 301 and 302 may be changed so that defects such as finer scratches and foreign matters can be detected. Furthermore, the illumination direction α1, φ1 (including zero) and φ2 (including zero) conditions of the illumination optical system 100, presence / absence of the polarizing plate 208, conditions for using the white illumination 500 and the laser illumination 100, and the like are changed. It may be.

  Further, the processing method in the CPU 417 also applies the detection signal level ss ′ corrected for defects inspected under a plurality of inspection conditions in each dimension (illumination light power, illumination direction, presence / absence of polarizing plate, white illumination and laser illumination). It may be a result of mapping into space and classifying (classifying) from distance in the space. For example, as shown in FIG. 49, the detection signal level ss ′ by the laser illumination system 100 is on the x axis, and another illumination system is on the y axis (the white illumination system 500 and the illumination system shown in FIG. 50). Plot the detection signal level ss' by. As a result, the positions of these plots can be classified into two regions according to a preset straight line y = βx. The classification result indicates the characteristics of the foreign matter. When the illumination system shown in FIG. 50 is used as another illumination system, a defect such as a foreign substance plotted in a region where y> βx is a flaw that does not shine by oblique illumination or a flat foreign substance 491, and y It has been confirmed by experiments that defects such as foreign matter plotted in the region of <βx are foreign matter 492 having a relatively high height. Here, the boundary line does not have to be the straight line described above, and may be an arbitrary curve, or a plurality of arbitrary straight lines or curves. The space for obtaining these distances may have a plurality of dimensions. Further, the plurality of inspections may be performed simultaneously by the detectors 205 and 206. 50, instead of obliquely illuminating the laser beams 10, 11, and 12 shown in FIG. 3, a linear fine mirror 240 is disposed between the objective lens 201 and the substrate 1 to be inspected. The laser light beam 10 is reflected by the linear fine mirror 240, and the beam light beam 3 is illuminated almost perpendicularly to the substrate 1 to be inspected. Therefore, the 0th-order diffracted light (regular reflected light) from the substrate 1 to be inspected is shielded by the linear fine mirror 240, and the 1st-order or higher-order diffracted light passes through the objective lens 201. Note that the linear fine mirror 240 is preferably a sufficiently thin linear band on the surface of the spatial filter 202 so that the function of the spatial filter 202 can be achieved.

Next, connection between the foreign matter inspection apparatus according to the present invention and an external apparatus will be described. That is, the CPU 471 controls the entire foreign matter inspection apparatus according to the present invention. Accordingly, the inspection result, the conditions for inspection (particularly the threshold map), and the like are stored in the storage device 427 connected to the CPU 417. Then, it is desired that the inspection results or inspection conditions stored in these storage devices 427 are connected to other computers via the local area network 428 or a modem. In particular, by connecting to the Internet, it is possible to exchange the improvement of the inspection conditions in the foreign substance inspection apparatus, the problem status of the foreign substance inspection apparatus, and the like between the user who uses the foreign substance inspection apparatus and the manufacturer of the foreign substance inspection apparatus. In the exchange of these data, confidentiality can be maintained by having the encryption key between the foreign substance inspection apparatus manufacturer and the user and encrypting the data. In addition, based on the inspection results of foreign matter inspected by the foreign matter inspection device, improvement of process conditions in the process processing device, status of problems of the process processing device, etc., the user of the process processing device and the manufacturer of the process processing device It is also possible to communicate between the two.
Also, by configuring the image processing apparatus 400 according to the present invention with a programmable system, the algorithms shown in FIGS. 4, 28, 29, 30, and 31 can be rewritten and executed. These algorithms are for dealing with partial fluctuations in signal intensity due to interference of oxide films on the wafer surface, and so-called color unevenness correspondence algorithms can be realized.

Next, a manufacturing line for semiconductors and the like using the defect inspection apparatus according to the present invention described above and a manufacturing method thereof will be described with reference to FIGS.
As shown in FIG. 32, a semiconductor production line using the defect inspection apparatus according to the present invention includes manufacturing processes 601 to 609, inspection apparatuses 610 to 612, a probe inspection process 614, and a data analysis system 613.
In particular, the manufacturing process includes processes 601, 605, 608, and 609 that have a large influence on the yield (which affects the yield), and these processes are constantly monitored by the inspection apparatus 612 such as the defect inspection apparatus according to the present invention. Is done. When an abnormality between processes is detected by this monitoring, for example, an abnormality is detected between the processes 601 and 606, the processes 602, 603, and 604 are monitored by the inspection device 610 to create an abnormality, or Identify the device. Also, a particularly important step 607 is exclusively monitored by the inspection device 611.

  By the way, in order to be able to inspect defects such as foreign matter only in a desired process or foreign matter adhering to the outermost surface with high identification accuracy, this process is performed before and after performing the process process of this process. It is preferable to perform a defect inspection of a foreign matter or the like by the foreign matter inspection apparatus 612 according to the invention and obtain a logical difference between the defect inspection result after the process processing and the defect inspection result before the process processing. Here, when making a determination based on this logical difference, a foreign matter generated before this process must not be erroneously determined as a defect generated in this process. Rather, this defect should be missed. This is because measures are taken so as not to cause defects based on erroneous determinations.

  However, due to the above logical difference, it is not always possible to detect only defects such as foreign matters generated in the process processing step. This is due to the following reasons. The reason is that, for example, a film is formed on the surface of a defect such as a foreign substance due to film formation or the like, and as a result, the defect size of the foreign substance or the like is increased, the inspection sensitivity is improved, and the defect existing before the film formation is This is because the film is inspected after the film formation. In other words, in fact, a defect that should have been attached before is not detected before the film forming process, but is found after the film forming process, and is mistakenly determined as occurring in the film forming process.

  Therefore, at the time of the inspection before the film forming process, for example, by reducing the magnification m1 and lowering the threshold value to increase the inspection sensitivity in advance, it becomes possible to detect a fine defect that has been attached before, which is wrong. Judgment can be eliminated. As described above, if the inspection sensitivity before the film forming process is increased, false detection (false report) increases. However, as shown in FIG. 51, the problem is caused by taking the logical difference (B−A) before and after the process. Must not. However, the state of the surface may change before and after the process for each region in the chip of the substrate 1 to be inspected. For this reason, even if the threshold value is lowered overall before processing, there is a region where the background level is high, resulting in a large threshold value and a practically non-inspection or low sensitivity state. From the area, it becomes impossible to detect a fine defect that has been attached.

Therefore, in the CPU 417 of the arithmetic processing circuit 400 of the defect inspection apparatus 612 according to the present invention, when Ib <Tb, it is detected as Ia> Tha, and is generated in the process step P only when Ia> κ · Thb. Judged as a defect. That is, in the inspection before the process processing step P, when a defect cannot be detected even if the inspection sensitivity is increased as much as possible, the inspection sensitivity is lowered in the inspection after the process processing step. Only when a defect is detected even if the threshold value is increased, it is determined that the defect has occurred in the process processing step P. Even if the inspection sensitivity is lowered in the inspection after the process processing step and the threshold is increased by κ, the defect is detected. If no error is detected, a process of overlooking this defect can be performed to eliminate erroneous determination. This is because the probability that it is determined that the defect has occurred in the process processing step decreases. Of course, when Ib ≧ Thb, it can be regarded as a defect that has occurred previously.
However, Ia indicates the detection signal level of the defect detected in the inspection after the process processing step, and Ib indicates the detection signal level of the defect detected in the inspection before the process processing step. Tha is an inspection threshold level obtained from the threshold map storage means 419 after the process processing step, and Thb is an inspection obtained from the threshold map storage means 419 lowered as much as possible before the process processing step. Indicates the threshold level. κ is a coefficient exceeding 1, and is determined according to Thb. In the comparison circuit 414 of the arithmetic processing circuit 400 of the defect inspection apparatus, Ia and Tha and Ib and Thb are compared.

Therefore, the defect determination processing performed by the CPU 417 is performed on the entire chip area before or after the process processing step (possibly after the process processing step) obtained from the threshold map storage means 419 and stored in the storage device 427. The threshold level (threshold image) of the conforming area and the defect detection signal before and after the process step obtained from the memory 404 and stored in the storage device 427 are required. What is important is that information on the threshold map at the time of the inspection before the process processing step is stored in the storage device 427, and the threshold at the time of the inspection after the process processing step is stored using this threshold map information. The purpose is to determine a coefficient κ that determines (κ · Tha). Naturally, Tha is calculated by the threshold value calculation means 418 during the inspection after the process processing step.
Further, a monitoring method that the defect inspection apparatus 612 monitors for the processes 602, 603, 604, etc. will be described. The first technique is a process monitoring technique using the same wafer that focuses on wafers in a lot and monitors the attachment state (change) of defects such as foreign matters on the wafer of interest every time a process is performed. The second technique is a technique for paying attention to a certain process apparatus or process, and monitoring the state of the process apparatus or process by monitoring the state before and after the wafer passing through the process. Both are common in that the state of a process is monitored, but the first method is for comparing processes and searching for a process with a bad state, and the second technique is for a certain process over time. The main purpose is to compare different changes. That is, the second method is intended to monitor anomalies such as sudden occurrence of foreign matter, or to evaluate the effect after implementing measures for reducing defects such as some foreign matter.

Here, in particular, the management that focuses on a specific process step or its apparatus by the inspection apparatus 612 can know how defects are increased or decreased in this process. Furthermore, in this management, in particular, by using the size of the foreign matter detected here and determining the criticality of the foreign matter in this process, the importance of the measure against the foreign matter can be known and the countermeasure is implemented. It becomes a motivation of time and becomes very effective. That is, by knowing the magnitude of the countermeasure effect of the defect such as a foreign object, the consciousness of the countermeasure is more strongly conscious and can be linked to the countermeasure action.
As described above, the monitored and captured data is captured by the data analysis system 613, and the occurrence of abnormality, the relationship with the yield from the relationship with the data from the probe inspection process 614, and the like are analyzed.
Furthermore, in addition to the defect inspection apparatus according to the present invention, inspection apparatuses such as bright bright field inspection and SEM inspection are used for the inspection apparatuses 610, 611, and 612. These inspection apparatuses have their respective features, and different foreign substances can be detected. Therefore, the reliability of inspection can be improved in total by using these inspection apparatuses in combination. In addition, these inspection apparatuses have a difference in inspection time (inspection throughput) from the detection principle. The high-throughput laser scattering method of the defect inspection method is suitable for inspection of fine particles, but the capture rate at the time of detection is low due to the coherence of the laser. The bright-field inspection has a high capture rate, but has a low throughput because it requires a high resolution at the time of sampling because it is a comparative inspection. The inspection using an electron beam is difficult to increase the inspection speed because the SN is low, but is suitable for inspection of poor conduction and the like, which enables high-resolution inspection.

In the LSI manufacturing process, it is necessary to systemize these inspection apparatuses in consideration of sensitivity, throughput, detectable objects, and the like.
As shown in FIG. 33, each inspection apparatus increases the total number of detections of the system by increasing defects such as foreign matter that can be detected from 24 to 27, from 25 to 28, and from 26 to 29. A system with high performance can be constructed.
FIG. 34 shows the yield transition 30 at the start of mass production. A transition 31 of the number of defects is also shown at the same time. As the yield increases, the number of defects decreases. However, even when the yield rises, the number of defects may suddenly increase and the yield may decrease. Therefore, it is necessary to know the occurrence of these defects quickly and temporarily stop the production of the defect generation process to take measures against the cause of the defect. Therefore, a defect inspection apparatus for foreign matter or the like according to the present invention is required.

It is a figure which shows the semiconductor wafer with which memory LSI which is one Example of the to-be-inspected board | substrate which concerns on this invention is arranged. It is a figure which shows the semiconductor wafer with which LSI, such as a microcomputer, which is another Example of the board | substrate to be tested which concerns on this invention is arranged. It is a schematic block diagram which shows 1st Embodiment of the defect inspection apparatus which concerns on this invention. It is a block block diagram which shows the 2nd Example of the image process part shown in FIG. It is a figure for demonstrating the method and the detection method of illuminating a slit-shaped beam on to-be-inspected board | substrates, such as a semiconductor wafer concerning this invention. It is a perspective view which shows the illumination light beam by the illumination lens which has a conical curved surface which concerns on this invention. It is a figure for demonstrating the 1st Example of the manufacturing method of the illumination lens which has a conical curved surface which concerns on this invention. It is a figure for demonstrating the 2nd Example of the manufacturing method of the illumination lens which has a conical curved surface which concerns on this invention. It is the side view seen from the y direction and x direction which shows the illumination optical system which concerns on this invention. It is a top view which shows the optical system for illuminating the to-be-inspected board | substrates, such as a semiconductor wafer, from 3 directions using one laser light source in the illumination optical system which concerns on this invention. It is a perspective view which shows the illumination direction and detection direction which concern on this invention, and the diffracted light from the pattern by an illumination direction. The state of generation of the 0th-order diffracted light pattern when the slit-shaped beam is illuminated from the 45 ° direction with respect to the main straight line group of the circuit pattern according to the present invention, and the objective lens of the detection optical system when the optical axis is vertical It is a figure which shows the relationship with opening. The state of generation of a 0th-order diffracted light pattern when a slit-shaped beam is illuminated from a direction parallel to the main straight line group of the circuit pattern according to the present invention and the objective lens of the detection optical system when the optical axis is vertical It is a figure which shows the relationship with opening. It is a figure which shows the relationship with the detection area which illuminates so that it may not cross when illuminating a slit-shaped beam from a different 45 degree direction with respect to the main straight line group of the circuit pattern which concerns on this invention, and is detected with a TDI sensor. It is a schematic block diagram which shows 2nd Embodiment of the defect inspection apparatus which concerns on this invention. It is a graph which shows the relationship between the injection angle from a foreign material, and detection signal strength. It is a figure which shows the Example at the time of inclining the light-receiving surface of a TDI sensor according to this inclination when the optical axis of a detection optical system is inclined. It is a projection view on the plane which shows the diffracted light fringe which arises from a repetitive pattern, when a slit-shaped beam is illuminated with respect to the main straight line group of the circuit pattern which concerns on this invention from a different 45 degree direction. It is a top view which shows the positional relationship of the diffracted light fringe from the repeating pattern in the Fourier-transform surface of the detection optical system which concerns on this invention, and this diffracted light fringe and a spatial filter. The generation state of the 0th-order diffracted light pattern when the slit-shaped beam is illuminated from the 45 degree direction with respect to the main straight line group of the circuit pattern according to the present invention, and when the optical axis is made vertical and tilted in the y direction It is a figure which shows the relationship with the opening of the objective lens of a detection optical system. The state of generation of the 0th-order diffracted light pattern when the slit-shaped beam is illuminated from a direction parallel to the main straight line group of the circuit pattern according to the present invention and the objective lens of the detection optical system in which the 0-diffracted light pattern does not enter It is a figure which shows the position of opening. It is a figure which shows the Example which illuminates a slit-shaped beam by making the longitudinal direction turn into an illumination direction from the 45 degree direction with respect to the main straight line group of the circuit pattern which concerns on this invention. It is a figure which shows the special TDI sensor required when the slit-shaped beam shown in FIG. 22 is illuminated. It is a side view which shows the interference model of the scattered light from the foreign material which exists on insulating films, such as an oxide film concerning this invention. It is a figure for demonstrating the Example which detects the scattered light from a foreign material from several detection directions, in order to detect the foreign material which exists on insulating films, such as an oxide film. Relationship between change in thickness of insulating film such as oxide film when irradiated with illumination light of certain wavelength and detection signal, and thickness of insulating film such as oxide film when irradiated with illumination light of three different wavelengths It is a figure which shows the relationship between a change and a detection signal. The figure which shows calculating and setting the determination standard (threshold value) for extracting defects, such as a foreign material, in the image processing part which concerns on this invention, The figure which shows the relationship between a wafer and a pixel, and a chip (it has various pattern area | regions) .) And a pixel. It is a block diagram which shows the 1st Example of the image process part which concerns on this invention. It is a block diagram which shows the 3rd Example of the image process part which concerns on this invention. It is a block diagram which shows the 4th Example of the image process part which concerns on this invention. It is a block diagram which shows the 5th Example of the image process part which concerns on this invention. It is a figure which shows schematic structure of the manufacturing line of the semiconductor in which defect inspection apparatuses, such as a foreign material, concerning this invention were installed. It is a figure for demonstrating that the system which has a high performance as a whole can be constructed | assembled by increasing the foreign material which each defect inspection apparatus can detect in the manufacturing line of a semiconductor. It is a figure which shows the transition of the yield and the number of defects at the time of mass production start-up. It is a figure which shows schematic structure of 4th Embodiment of the defect inspection apparatus which concerns on this invention. It is a figure which shows concretely the structure of the Example of the illumination optical system used for 4th Embodiment of the defect inspection apparatus shown in FIG. 35 from ay direction and ax direction. It is a figure for demonstrating the basic idea which shapes a slit-shaped Gaussian beam light beam with an illumination optical system, and improves illumination efficiency. It is a figure for demonstrating the method to receive and image the optical image of the detection area | region on a to-be-inspected board | substrate at the time of using a TDI image sensor as a detector. When varying the standard deviation σ a (corresponding to the width of the illumination) in the Gaussian beam flux is a diagram illustrating a change in the illuminance f (x ₀₎ in the peripheral portion of the detection area (x ₀ = 1). When irradiated with Gaussian beam flux when the standard deviation σ to 0.5, 1, 2, is a graph showing changes in intensity f (x ₀₎ to the length of the optical axis (x ₀₎ of the detection area . It is a figure for demonstrating the Example of the TDI image sensor which enabled it to receive DUV light. It is a figure which shows the Example of the sequence of condition determination in the defect inspection apparatus which concerns on this invention. It is a figure which shows the screen which performs condition setting mode selection and threshold value prior selection which were displayed on the display means. It is a figure which shows the screen which displayed the detection sensitivity and the detection area on the display means. It is a figure which shows the screen which displayed the threshold value map in the case of area priority, standard, and sensitivity priority, and the relationship of the inspection area with respect to a sensitivity on a display means. In the defect inspection apparatus which concerns on this invention, it is a figure which shows the embodiment which provided the optical system which observes the light-shielding pattern of a spatial filter in the detection optical system, and used the optical observation microscope. It is a figure which shows the relationship based on the experimental data of the standard particle diameter on the mirror-surface wafer which concerns on this invention, and an evaluation value (detection signal level of scattered light). It is a figure for demonstrating the Example which estimates the size of a foreign material from the detected image signal. It is a figure for demonstrating the Example which can classify | categorize the kind of defect from the detection signal level by a laser illumination system, and the detection signal level by another illumination system. In the defect inspection apparatus according to the present invention, it is a schematic configuration diagram showing an illumination optical system and a detection optical system when a light beam is bright-field illuminated using a linear fine mirror. It is a figure which performs a defect inspection with high sensitivity before processing with a certain process processing apparatus P, performs a defect inspection with optimal sensitivity after the processing, and shows the logical difference (B-A). It is a figure which shows the structure which can determine a defect with a high S / N ratio in the defect inspection apparatus which concerns on this invention from a large foreign material to a large foreign material with a breadth.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Test target substrate (wafer), 1a, 1b ... Semiconductor wafer, 1aa ... Memory LSI, 1ab ... Memory cell area, 1ac ... Peripheral circuit area, 1ad ... Other area, 1ba ... LSI such as microcomputer, 1bb ... Register Group area, 1bc ... Memory area, 1bd ... CPU core area, 1be ... I / O area, 2 ... Chip, 3 ... Slit beam (illumination area), 4 ... Detection area of image sensor such as TDI sensor, 100 ... Illumination optical system, 101 ... laser light source, 102 ... concave lens, 103 ... convex lens, 104 ... illumination lens having a conic surface, 200 ... detection optical system, 201 ... objective lens (detection lens), 202 ... spatial filter, 203 ... imaging Lens, 204 ... Beam splitter, 205, 206 ... One-dimensional detector such as TDI sensor, 207 ... ND filter, 208 Polarizing element, 226... Mirror, 227 .. Imaging optical system, 228... TV camera, 240... Linear fine mirror, 300 .. white illumination system, 400 .. arithmetic processing circuit (signal processing system), 401. , 402 ... data memory, 403 ... difference processing means (difference processing circuit), 403 '... absolute value difference processing circuit, 404 ... data memory, 405 ... maximum / minimum removal circuit, 406 ... square calculation circuit, 407 ... calculation circuit 408... Number counting circuit, 409... Sum of squares calculation circuit, 410... Sum calculation circuit, 411... Counting circuit, 412 ... Upper limit determination criterion (positive threshold) calculation circuit, 413 ... Lower limit determination criterion (negative threshold) calculation Circuits, 414, 415, 414 '... comparison circuit, 417 ... CPU (output means), 419 ... threshold map storage means, 421 ... output means (display means), 422 ... memory position controller Roller, 423 ... Threshold calculation circuit, 424 ... Threshold setting circuit, 425 ... Average value calculation circuit, 426 ... Input means, 427 ... Storage device, 428 ... Network, 520, 521 to 525 ... Operator, 531 to 535 ... Comparison circuit, 541 to 544 ... Multiplication circuit, 550 ... OR circuit, 600 ... Optical observation microscope, 601 ... Detection optical system, 602 ... Image processing system, 603 ... Display means, 604 ... Image memory, 610 to 612 ... Inspection apparatus, 613 ... Data analysis system, 1007.

Claims (6)

Stage means for placing the object to be inspected and moving in at least one axial direction;
A laser beam emitted from a laser light source is formed into a slit beam that is substantially parallel light in the longitudinal direction, and has a predetermined inclination with respect to the normal direction of the surface of the object to be inspected placed on the stage means. An illumination optical system that illuminates the surface of the object to be inspected from an illumination direction along the longitudinal direction of the slit beam;
A spatial filter for shielding a diffracted light pattern from at least a repetitive pattern of a circuit pattern present on the object to be inspected, and present on the object to be inspected by a slit beam illuminated by the illumination optical system; An imaging optical system that forms an image of reflected and scattered light obtained from a defect such as a foreign object;
An image sensor that receives an image formed by the imaging optical system and outputs a photoelectric conversion signal in synchronization with the movement of the stage;
A defect inspection apparatus comprising: an image processing unit that extracts a signal indicating a defect such as a foreign substance based on a signal output from the image sensor . The defect inspection apparatus according to claim 1, wherein the image sensor is a TDI sensor . 3. The defect inspection apparatus according to claim 1, wherein the slit beam is formed using a cylindrical lens in the illumination optical system . 3. The defect inspection apparatus according to claim 1, wherein in the illumination optical system, an illumination direction along the longitudinal direction of the slit beam is tilted with respect to a main straight line group of a planar circuit pattern . 5. The defect inspection apparatus according to claim 1, wherein in the illumination optical system, the slit beam has a shape parallel to an incident surface of the illumination . 6. 5. The defect inspection apparatus according to claim 1, wherein in the detection optical system, an optical axis is inclined with respect to a normal line of a substrate to be inspected .

Images