JP2010243283A - Defect inspection method and defect inspection device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To improve an operation rate of an inspection device by setting an inspection condition, in a state where an actual inspection object does not exist, in defect inspection that uses a substrate with a pattern as the inspection object. <P>SOLUTION: Pattern data acquired from pattern design data 111 are decomposed into fundamental patterns (121), and each detection output of the fundamental patterns is calculated, by adding process condition data 112, material constant data 113 and a device parameter 115 (122), and each detection output of the fundamental patterns is integrated to generate an estimation image (124). Meanwhile, the output data of a defect are generated by using fundamental defect data 114, the material constant data 113 and the device parameter 115 (123A). The inspection condition of the inspection device is set, by using the estimation image 124 and defect data 123A. Consequently, the inspection condition can be set off-line, and the operation rate of the device is improved. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to an inspection technique for detecting a defect on an inspection target having a pattern formed on a surface based on image information obtained by using light or an electron beam, and more particularly, to a pattern with a pattern such as a semiconductor wafer or a photomask. The present invention relates to an inspection technique for detecting fine defects on a substrate.

  In the defect inspection of a substrate with a pattern, the defect detection sensitivity greatly depends on how the defect detection signal can be distinguished from the pattern detection signal (pattern noise). Particularly in the inspection of semiconductor wafers, with the progress of pattern miniaturization, detection of finer defects is required, and it is necessary for inspection technology to extract weak detection signals from fine defects separately from pattern noise. It has become a big issue.

  Here, with reference to FIGS. 2 to 3, a patterned substrate to be inspected will be described by taking a semiconductor wafer as an example. FIG. 2 is an explanatory view of a semiconductor wafer and a semiconductor chip to be inspected. FIG. 2A is a semiconductor wafer, and a plurality of semiconductor chips 201 are formed on the semiconductor wafer 200.

  FIG. 2 (2) shows a configuration in the semiconductor chip 201 taking a logic product as an example, and includes a plurality of logic functions (Logic 1 and 2) blocks 211 and 212 and a memory function (SRAM) block 213. Each block is configured with a different circuit pattern according to a required function, and pattern noise at the time of wafer inspection is also different for each block.

  FIG. 3 is an explanatory diagram of the vertical structure of the semiconductor device. The vertical structure of the semiconductor device will be described with reference to 30 to 35 and 301 to 351, and the defect to be detected will be described with reference to 361 to 364.

  An element isolation layer 30 is a structure (302) for isolating a transistor element formed on a wafer by digging a groove in a silicon (Si) substrate 301 and then filling it with silicon oxide (SiO2) as an insulator. is there. 31 is a gate and contact layer, and 311 is a gate electrode portion made of polysilicon (poly-Si). This part has a great influence on the transistor performance, and is emphasized also in the inspection.

  Reference numeral 312 denotes a contact portion, which is a portion in which a metal (tungsten: W, etc.) is buried in a hole opened by etching in an insulating film (silicon oxide: SiO 2) to connect the transistor portion and the upper wiring layer. 32 to 35 are wiring layers, and a circuit is formed by the plurality of wiring layers. Each layer is filled with an insulating film (silicon oxide: SiO2).

  32 is a first wiring portion, the first wiring portion 321 performs wiring in the plane direction, and the first via portion 322 is metal in a hole opened by etching in an insulating film (silicon oxide: SiO2). This is the part that connects the upper wiring layers. Similarly, 33 is a second wiring layer, 331 is a second wiring portion, 332 is a second via portion, 34 is a third wiring layer, 341 is a third wiring portion, and 342 is a third wiring layer. The via portion 35 is a fourth wiring layer, and 351 is a fifth wiring portion.

  In each wiring layer, the material of the wiring part is made of a metal such as aluminum (Al) or copper (Cu). The metal embedded in the via portion is made of tungsten (W) or copper (Cu). Further, the defect to be detected includes a scratch 361, a pattern defect short 362 and disconnection 363, a foreign object 364, and the like.

  FIG. 4 is an explanatory diagram of processes, materials, and representative defects of each layer of the semiconductor device shown in FIG. Although details are omitted, the film formation process of the material that forms each layer, the formation of a resist pattern by a lithography process, the etching process that processes the formed film to follow the formed resist pattern, and the CMP ( Each layer of the semiconductor device is formed by a chemical mechanical polishing (chemical mechanical polishing) process or the like.

  The materials used in each layer and each process vary, and the defects to be detected are foreign matter in the film formation process, foreign matter and pattern defects in the lithography and etching processes for pattern formation, and foreign matter and scratches in the CMP process for polishing. And so on.

  As described with reference to FIGS. 2 to 4, in the inspection of semiconductor wafers, there are a wide variety of pattern shapes and materials, and the shapes and materials of defects to be detected. Exists. Each inspection apparatus is configured so that a plurality of detection condition parameters can be set so as to optimize the defect detection sensitivity. Specific examples thereof will be described with reference to FIGS.

  FIG. 5 is an explanatory diagram of a dark field optical inspection apparatus. 510 is an illumination optical system, and light emitted from the light source 511 illuminates the substrate 500 with an illumination light beam 5110 formed through a condenser lens 512. As the light source 511, a laser light source with good directivity is generally used.

  In the illumination optical system 510, the polarization of the illumination light is controlled by the wave plate 513, and the amount of illumination light is controlled by the light control filter 514. Reference numeral 520 denotes a detection optical system, and the objective lens 521 detects reflected / scattered light from defects and patterns existing on the substrate. In the dark field optical system, the regular reflection light 5111 from the substrate is configured not to enter the objective lens.

  An imaging lens 525 forms an image of the substrate on the sensor 531. In the detection optical system 520, reference numeral 522 denotes a polarizing plate that controls the polarization of the detection light. An aperture stop 523 controls the detection solid angle of the reflected scattered light by controlling the aperture 5230 (numerical aperture: NA = sin θ) of the detection optical system. Reference numeral 524 denotes a spatial filter that blocks regular diffracted light generated from a pattern regularly formed on the substrate, thereby reducing pattern noise during detection and improving defect detection sensitivity.

  The sensor 531 converts the detected optical image into an electrical signal by photoelectric conversion. The image processing unit 532 converts the electrical signal into image data, and performs defect detection by performing image processing on the image data. An overall control unit 550 controls the illumination optical system, the detection optical system, the image processing unit, and the stage 540. The stage 540 mounts the substrate 500 and inspects the entire surface of the substrate by moving in the x-y direction.

  FIG. 6 is an explanatory view of a bright-field optical inspection apparatus. Reference numeral 610 denotes an illumination optical system. The light emitted from the light source 611 passes through the condenser lens 612, and is then folded back by the half mirror 621 of the detection optical system to illuminate the substrate 600 through the objective lens 622. In the bright field optical system, the light source 611 is generally a lamp having a wide wavelength range.

  In the illumination optical system 610, the illumination aperture 613 controls the illumination light aperture 613, the polarization filter 614 controls the illumination light polarization, and the wavelength filter 615 controls the illumination light wavelength range. Reference numeral 620 denotes a detection optical system, and the objective lens 622 detects reflected light from the substrate and defects and patterns existing on the substrate.

  An imaging lens 625 forms an image of the substrate on the sensor 631. In the detection optical system 620, reference numeral 623 denotes an aperture stop, which controls the detection solid angle by controlling the aperture 6230 (numerical aperture: NA = sin θ) of the detection optical system. A polarizing plate 624 controls the polarization of the detection light.

  The sensor 631 converts the detected optical image of the substrate into an electrical signal by photoelectric conversion. The image processing unit 632 converts the electrical signal into image data, and performs image processing on the image data to detect defects. To do. An overall control unit 650 controls the illumination optical system, the detection optical system, the image processing unit, and the stage 640. The stage 640 mounts the substrate 600 and inspects the entire surface of the substrate by moving in the x-y direction.

  FIG. 7 is an explanatory diagram of an electron beam type inspection apparatus. Reference numeral 710 denotes an electron optical system barrel. The electron beam emitted from the electron gun 711 forms a beam spot by the condenser lens 712 and the objective lens 714 and is irradiated onto the substrate 700. As the beam condition, the acceleration voltage (electron beam energy) is controlled by the electron gun 711, the beam current is controlled by the condenser lens 712, and the spot size is controlled by the objective lens 714 as parameters of the detection condition.

  Further, the deflection of the beam is controlled by the deflector 713 so that the beam spot scans the substrate 700. A sensor 715 for detecting secondary electrons 7111 generated from the substrate converts the secondary electron intensity into an electric signal by photoelectric conversion. The image processing unit 730 converts the electric signal into image data, and the image data is converted into an image data. Defect detection is performed by processing. An overall control unit 750 controls the electron gun, the condenser lens, the objective lens and the image processing unit, and the stage 740. The stage 740 is for inspecting the entire surface of the substrate by placing the substrate 700 and moving in the x-y direction.

  As described above with reference to FIGS. 5 to 7, various inspection apparatuses exist in order to cope with various patterns of layers of the semiconductor wafer to be inspected and defect types to be detected. There are various control parameters for the detection conditions of the inspection apparatus. However, the detection condition for one inspection object needs to be set as one that optimizes the defect detection sensitivity.

  The task of selecting the optimal condition from the multiple detection conditions that can be set by the inspection device and setting it in the inspection recipe (conditioning operation) involves trial and error and requires a lot of work and skill. Therefore, there is a problem that when the apparatus is occupied in the condition setting operation, the apparatus operation rate is lowered, and a technique for solving this problem is demanded.

As one of the means, a technique for determining conditions by calculation is disclosed. According to these technologies, it is possible to improve the operation rate of the equipment by setting the conditions in offline work that does not occupy the actual equipment, and to obtain the advantages that the conditions can be theoretically calculated, so that it is easy to provide guidelines for optimizing the inspection conditions. I can do it.
For example, Patent Document 1 discloses a technique for creating an inspection recipe by determining the reflectance of a surface to be inspected based on film thickness information and design information (layout information) in an apparatus for optically inspecting a semiconductor wafer. Is disclosed.

Patent Document 2 and Patent Document 3 disclose techniques for creating an inspection recipe by calculating a sensitivity parameter of an inspection apparatus by simulation from design information to be inspected in inspection / measurement of a semiconductor wafer. .
Further, Patent Document 4 discloses a technique for determining a noise intensity from a periodic pattern by simulation when optically inspecting a semiconductor wafer on which a pattern is formed, and setting conditions for a detection solid angle and a partial light shielding filter. Is disclosed.

  Further, in Patent Document 5, in a light scattering inspection system (dark field optical inspection apparatus), scattered light generated from an inspection object is obtained by simulation, and the signal-to-noise ratio is increased accordingly. A technique for determining the configuration of is disclosed.

Japanese Patent Laid-Open No. 2004-340773 JP 2007-311418 Special Table 2005-514774 Special Table 2004-501371 JP 2007-279040

  However, the above Patent Document 1 does not disclose a specific calculation method for determining the reflectance of the inspection object from the design information.

In the above-mentioned Patent Document 2, it is disclosed that optical simulation by the finite element method is performed based on the Maxwell equation as a means for calculating the sensitivity parameter of the inspection apparatus by simulation. The size of the object that can be calculated practically is limited to the micrometer order, and it is difficult to perform sensitivity simulation over the entire inspection area of the inspection object.
In Patent Document 3, a specific calculation method for determining measurement sensitivity from design data by sensitivity simulation is not disclosed.

  In the above-mentioned patent document 4, the target of simulation is limited to a periodic pattern, and a calculation method for a general pattern is not disclosed.

  In the above-mentioned Patent Document 5, a substrate on which a pattern is not formed is an inspection target, and a technique corresponding to the substrate on which a pattern is formed is not disclosed.

  The present invention has the following configuration in order to solve the above-described problems and achieve the object of carrying out the calculation of the inspection conditions based on the calculation in the inspection of the defect on the inspection object having the pattern formed on the surface. .

  (1) In a defect inspection method for an inspection object having a pattern formed on the surface, one or more basic patterns are extracted from pattern design data formed on the inspection object, and a detection output of the basic pattern is calculated by physical simulation Then, a pattern estimation image corresponding to the pattern design data is created by integrating the detection output of the basic pattern, an inspection condition is defined based on the pattern estimation image, and the inspection condition is set in the inspection apparatus. Defect inspection method characterized by

  (2) Modeling a defect to be inspected, creating a defect estimation image by physical simulation, calculating defect detection sensitivity based on a synthesized image obtained by synthesizing the defect estimation image and the pattern estimation image, and detecting the defect The defect inspection method according to (1), wherein an inspection condition is determined based on sensitivity.

  (3) Defect inspection according to (2), wherein defect detection sensitivity is calculated based on both the pattern estimation image and the composite image, and inspection conditions are determined based on the defect detection sensitivity. Method.

  (4) The pattern estimation image, the defect estimation image, and the composite image are calculated as a plurality of images corresponding to each of a plurality of detection conditions that can be set by an inspection apparatus, and are most considered in consideration of the plurality of images. The defect inspection method according to (2) or (3), wherein the inspection condition is determined based on an inspection condition having a high detection sensitivity.

  (5) Any one of (2) to (4), wherein the pattern estimation image, the defect estimation image, and the composite image are calculated in consideration of variations in detection signals. The defect inspection method described in 1.

  (6) The defect inspection method according to (5), wherein the variation is a variation in the inspection object itself, such as edge roughness, corner rounding, pattern height, and film thickness of the inspection object pattern.

  (7) The defect inspection method according to (5), wherein the variation is caused by variation in aberration of a detection optical system of the inspection apparatus.

  (8) The defect inspection method according to (5), wherein the variation is caused by variations in image sampling in the inspection apparatus.

  (9) Based on a plurality of images calculated corresponding to each of the plurality of inspection areas, the defect detection sensitivity in each inspection area is calculated, and this is compared with the separately calculated yield prediction calculation result. The defect inspection method according to any one of (1) to (5), wherein an inspection condition is defined.

  (10) The defect inspection method according to (1) to (5) or 9, wherein the inspection method is a dark-field optical detection method.

  (11) The defect inspection method according to any one of (1) to (5) or (9), wherein the inspection method is a bright-field optical detection method.

  (12) The defect inspection method according to any one of (1) to (5) or (9), wherein the inspection method is an electron beam detection method.

  (13) In a defect inspection apparatus to be inspected with a pattern formed on the surface, means for holding pattern design data formed on the inspection object, means for extracting one or more basic patterns from the pattern design data, A pattern obtained from the inspection region to be inspected by integrating the means for calculating the detection output from each of the extracted basic patterns by physical simulation and the calculation result of the basic pattern corresponding to the pattern design data. And a means for creating an estimated image and a means for determining an inspection condition based on the pattern estimated image, wherein the determined inspection condition is configured to be set as an apparatus parameter. Defect inspection equipment.

  (14) Modeling the defect to be inspected, creating a defect estimation image by physical simulation, means for creating a synthesized image obtained by synthesizing both the defect estimation image and the pattern estimation image, and the synthesis Means for calculating defect detection sensitivity based on an image; and means for determining an inspection condition based on the calculated defect detection sensitivity, wherein the determined inspection condition can be set as an apparatus parameter. (13) The defect inspection apparatus according to (13).

  (15) Comprising means for creating a synthesized image obtained by synthesizing the pattern estimation image to be inspected and the defect estimation image, calculating defect detection sensitivity based on both the pattern estimation image and the synthesized image, Means for determining an inspection condition based on the calculated defect detection sensitivity, and is configured such that the determined inspection condition can be set as an apparatus parameter. Defect inspection equipment.

  (16) Based on a plurality of images calculated corresponding to each of the plurality of inspection regions, a means for calculating the defect detection sensitivity in each inspection region is compared with a separately calculated yield prediction calculation result. The defect inspection apparatus according to any one of (13) to (15), further comprising means for determining an inspection condition.

  (17) The defect inspection apparatus according to any one of (13) to (16), wherein the inspection means is a dark field optical detection method.

  (18) The defect inspection apparatus according to any one of (13) to (16), wherein the inspection means is a bright field optical detection system.

  (19) The defect inspection apparatus according to any one of (13) to (16), wherein the inspection means is an electron beam detection system.

  According to the present invention, the above-described configuration can improve the operation rate of the apparatus by setting the conditions in the offline work that does not occupy the actual apparatus, and the condition can be theoretically calculated. It is possible to obtain an advantage that it is easy to give a guideline for conversion.

It is explanatory drawing of the basic composition of this invention. It is explanatory drawing of the semiconductor wafer and semiconductor chip which are inspection objects. It is explanatory drawing of the vertical structure of a semiconductor device. It is explanatory drawing of the process of each layer of a semiconductor device, material, and a representative defect. It is explanatory drawing of the inspection apparatus of a dark field optical system. It is explanatory drawing of the inspection apparatus of a bright field optical system. It is explanatory drawing of an electron beam type inspection apparatus. It is explanatory drawing of the calculation (simulation) method in this invention. It is explanatory drawing of the parameter of the test object which causes a detection output variation. It is explanatory drawing of the parameter of the detection optical system which causes a detection output variation. It is explanatory drawing of the image sampling parameter which causes a detection output variation. It is explanatory drawing of a product yield prediction method. It is explanatory drawing of the 1st usage method of this invention. It is explanatory drawing of the apparatus structural example of this invention. It is explanatory drawing of the 2nd utilization method of this invention. It is explanatory drawing of the example of an operation screen in this invention.

  Hereinafter, the contents of the present invention will be described in detail using examples.

  An embodiment of the present invention will be described with reference to FIG. 1, which is an explanatory diagram of a basic configuration of the present invention. An inspection condition calculation unit 100 is a part for generating a detection image based on a physical simulation and setting an inspection condition. 111 to 115 are data that are the basis of detection image generation by physical simulation.

  Reference numeral 111 denotes pattern design data (pattern layout data such as GDSII), which is planar shape information of a circuit pattern. 112 is process condition data, which is information on the vertical structure such as the film thickness of each layer (explained in FIG. 3) of the semiconductor device, and data of the material used.

  Reference numeral 113 denotes material constant data. For optical inspection, it is the refractive index (n) and absorption coefficient (k), and for electron inspection, it is the mass and work function. These data are associated with the materials included in the process condition data 112 and the defect data 114 described below. The defect data 114 is a model of the size, shape, and material of the defect to be detected.

  Since the pattern design data 111 is very large-scale information, it is difficult to use it as it is for detection output calculation by physical simulation. For this reason, decomposition into basic patterns 121 is performed to extract basic pattern information that is a plurality of elements that are large enough to perform physical simulation and that can represent the entire pattern design data. The apparatus parameter data 115 is information on detection condition parameters that can be controlled by the various inspection apparatuses described with reference to FIGS. 5 to 7 and includes both the condition parameters of the illumination optical system and the condition parameters of the detection optical system.

  In the basic pattern detection output simulation 122, information on the basic pattern (planar shape information) decomposed in 121, information on the vertical structure and material of the pattern obtained from the process condition data 112, material constant data 113, and apparatus parameter data 115 Based on the above, basic pattern detection output calculation data 122A is calculated.

  In the defect detection output simulation 123, defect detection output calculation data 123A is calculated based on the condition parameters of the illumination optical system among the defect data 114, the material constant data 113, and the apparatus parameter data 115.

  In the integration process (estimated image creation) 124, the pattern design data 111 is integrated with the detection output calculation data 122A from the corresponding basic pattern to create an estimated image of the inspection region to be inspected. At this time, the condition parameter of the detection optical system is selected from the apparatus parameter data 115 to reflect the detection condition. In addition, the defect detection output calculation data 123A is also reflected in the detection conditions selected above to create a defect image, and is combined with the estimated image generated in 124 to generate an image of the inspection area where the defect exists (composite image). ).

  In the inspection condition setting 126, the estimated image of the inspection area created in 124 above, the synthesized image obtained by synthesizing both the estimated image of the inspection area and the defect image created in 125 above, or both the estimated image and the synthesized image Set image processing conditions.

  When the estimated image is used alone, defect detection by threshold processing can be set as the image processing condition setting. That is, in this case, a threshold value for defect detection is set only by the estimated image.

  When the composite image is used alone, it is possible to set the defect detection by threshold processing and estimate the defect detection sensitivity based on it. That is, in this case, a defect detection threshold value can be set while evaluating both the estimated image and the defect image, and the defect detection sensitivity corresponding to the threshold value can be evaluated.

  In the image processing condition setting by both the estimated image and the composite image, it is possible to set the image processing condition for defect detection by image comparison and estimate the defect detection sensitivity based on the setting. That is, in this case, only the defect image is extracted by subtracting the estimated image from the composite image, and a defect detection threshold value is set based on this, and the sensitivity of defect detection is evaluated for each threshold value. I can do it.

  As a result, the image processing conditions for one detection condition selected from the apparatus parameter data 115 are set for one inspection object, so these are set in the inspection apparatus and the inspection object is inspected.

  The above describes the method of creating an estimated image / composite image by calculation and setting the image processing conditions for one detection condition of the inspection apparatus. However, a plurality of detection conditions (illumination optical system parameters and detection optical system) that the inspection apparatus can take are described. After calculating multiple estimated images and composite images for each parameter, set the image processing conditions for each, select the detection conditions with the highest sensitivity, set them in the inspection device, and set the inspection target. It is also possible to carry out the inspection.

  FIG. 8 is an explanatory diagram of a calculation (simulation) method according to the present invention, taking as an example a case where the dark field optical detection method is used. Reference numeral 111 denotes pattern design data (schematic diagram) existing in the inspection area to be inspected, which is planar shape information of the circuit pattern.

  112 is process condition data, which is information on the vertical structure of the pattern and data on the material used. Reference numeral 113 denotes material constant data, which is a refractive index (n) and an absorption coefficient (k) for optical inspection. The pattern design data 111 is decomposed by the basic pattern decomposition 121 into basic patterns 811 that are sizes that can execute a physical simulation and that can represent the entire pattern design data.

  FIG. 8 shows an example of decomposition into 12 basic patterns A1 to C4. A is a straight line part, B is a convex corner part, C is a concave corner part, and 1-4 are classified according to the direction of the pattern. When the bright field optical detection method or the electron beam detection type inspection device is the target, although it is different from this figure, it is decomposed into basic patterns.

  The detection output simulation 122 includes basic pattern 111, process condition data 112 that is data of the vertical structure and material, material constant data 113, and apparatus parameter data 115A of the illumination optical system, that is, illumination wavelength, illumination polarization, illumination incident angle. Based on this information, the detection output from the basic pattern is calculated.

  As a simulation method used here, a finite element method (FEM), a time domain finite difference method (FDTD), a DDA method (Discrete dipole approximation), or the like is appropriate. All of these are methods for performing electromagnetic field calculations based on the Maxwell equation, and are particularly capable of calculating the reflected and scattered light when an object smaller than the wavelength of the illumination light is illuminated. Suitable for inspection of semiconductor wafers.

  Although not shown in the figure, the defect detection simulation 123 in FIG. 1 also uses the defect data 114 (defect shape, material data) and the apparatus parameter data of the illumination system (FIG. 115A) as shown in FIG. Similar to the above, the reflected scattered light is calculated.

  Here, the radiation distribution of scattered light from each basic pattern is calculated, and the calculation result is the light intensity, phase, and polarization distribution with respect to the radiation angle of the scattered light. 821 shows a schematic diagram of the radiation intensity distribution of scattered light corresponding to A1 to C4 of the basic pattern 811. FIG.

  Next, the detection output calculation result is integrated 830 into the pattern design data. For each part of the pattern design data 111, the corresponding basic patterns (A1 to C4) are integrated. The integrated result is shown in 831. In addition, a point light source with the radiation distribution (light intensity, phase, polarization distribution) of the scattered light calculated for each basic pattern is equivalently placed at the position of each basic pattern shown in 831 to Modeling the state of reflected and scattered light from the inspection area when illuminated under certain lighting conditions.

  In the imaging calculation simulation 840, the image of the inspection area is calculated in the following two stages. First, interference images of plural / plural types of point light sources corresponding to the integrated basic pattern are calculated. Specifically, the radiation distribution from each point light source is added in consideration of interference (phase distribution of each point light source).

  As a result, a diffraction image of the reflected scattered light from the inspection region at the position of the aperture stop 523 in FIG. 5 (dark field optical inspection apparatus) is calculated. The diffraction image information calculated here is light intensity, phase, and polarization distribution. This diffraction image information is filtered using the detection optical system device parameter data 115B, that is, detection NA, detection polarization, spatial filter conditions, etc. to reflect the setting conditions of the detection optical system, and then image calculation The imaging calculation result 841 of the inspection area to be inspected is calculated by the simulation 840. The imaging calculation simulation can use imaging calculation by FFT (Fast Fourier Transform).

  Although not shown in the figure, the defect image required in the synthesis process 125 in FIG. 1 can also be calculated by the same imaging calculation as described above.

  The above is an example of the case where the dark-field optical detection method is targeted. However, even when the bright-field optical detection method is targeted, the image calculation result of the inspection region to be inspected is obtained by substantially the same calculation. I can do it.

  Further, the image obtained by the inspection apparatus is obtained by sampling the image obtained by the detection optical system with the pixel of the sensor (531 in FIG. 5 and 631 in FIG. 6). Therefore, it is necessary to convert the detection image 851 in consideration of the pixel resolution of the sensor.

  In contrast, when the electron beam detection method is used, the detection output simulation 122 uses a Monte Carlo simulation, and the spot position of the electron beam with respect to the inspection region to be inspected and the intensity of the secondary electron detection output corresponding thereto. It is possible to create image data from the distribution. The imaging calculation described in the above 830 and 840 may be omitted.

  9 to 11 will be described with reference to image calculation taking account of variations in detection output. When an actual inspection object is inspected by an actual inspection apparatus, the inspection result may vary due to variations in the inspection object itself and variations in the inspection apparatus. Even when the inspection condition setting is obtained by calculation, it is effective for stable inspection to set inspection conditions that are not easily affected by the variation in consideration of these variations.

  FIG. 9 is an explanatory diagram of parameters to be inspected that cause variations in detection output. The parameters of the inspection target that cause the detection signal variation include the variation in the size (w) of the line edge roughness (LER) 902 of the pattern 901 shown in FIG. The variation of the size (R) of the corner round 903 of the pattern 901 shown in FIG. 9 or the height fluctuation (921) of the pattern 901 and the film thickness fluctuation (931) of the transparent film 930 covering the pattern shown in FIG. is there. Due to these fluctuations, the detection output also varies.

  FIG. 10 is an explanatory diagram of parameters of the detection optical system that cause variations in detection output. In the figure, 500 is an inspection object (wafer), and each point A, B, C of the inspection object is imaged as A ′, B ′, C ′ by the detection optical system 520 at the position 1000 of the detection sensor. At this time, since the objective lens 521 and the imaging lens 525 of the detection optical system actually have lens aberration, distortion occurs in the image. That is, even if the inspection object is the same, images detected by different detection optical systems of different inspection apparatuses have different distortions due to lens aberrations, resulting in variations in detection output.

  In addition, since the light path from the points A, B, and C to be inspected to the imaging points A ′, B ′, and C ′ is different, the image distortion depends on the points A ′, B ′, and C ′. It will be different. That is, even when the same inspection object is inspected by the same inspection apparatus, the detection output varies depending on the position in the lens field of view.

  FIG. 11 is an explanatory diagram of image sampling parameters that cause detection output variations. 1110 is an image to be detected, 1111 is a sensor pixel, and 1112 is a sensor pixel size. FIG. 11 (1) shows a case where an image to be detected is detected by one pixel. On the other hand, FIG. 11 (2) shows a case where the detection target image is detected across four pixels.

  The detection output (peak value of the detection waveform) obtained in this case is 1/4 in the case of FIG. 11 (2) as compared to FIG. 11 (1). In this way, detection variation occurs due to data sampling. Variations in data sampling are caused, for example, by variations in the position accuracy and speed accuracy of the stage of the inspection apparatus.

  As shown above, detection variations include (1) those due to inspection parameters, (2) those due to detection optical system parameters, and (3) those due to image sampling parameters. 8, detection output simulation 122, (2) is an imaging calculation simulation 840 in FIG. 8, and (3) is an image data sampling simulation 850 in FIG. It is possible to take into account.

As explained in FIG. 2, there are a plurality of different pattern areas in the semiconductor chip.
Even during inspection, the detection sensitivity differs in each area. In addition, the detection conditions that maximize the sensitivity may be different for each of the plurality of pattern regions. Based on this, a criterion for determining the optimum detection condition for the inspection object is required. Here, an embodiment will be described in which the detection condition is determined based on a comparison with the yield prediction result.

  FIG. 12 is an explanatory diagram of a product yield prediction method based on a critical area method (CAA), which is one of yield prediction methods. The critical area method is a method of calculating the degree of electrical failure from the relationship between the pattern design data and the number of defects generated with respect to the defect size. As a procedure, pattern design data 121 is input, and a defect rate curve 1220 with respect to the defect size is created.

  Here, 1201 is a defect having the size of R1, 1202 is a defect having the size of R1 in another position, 1203 is a defect having a size of R2 larger than 1201 and 1202, 1204 is a space portion of the pattern, and 1205 is a pattern of the pattern Wiring part. Here, the defect 1201 is in the space portion of the pattern and there is no possibility of causing an electrical failure. However, the defect 1202 is in the wiring portion of the pattern and may cause an electrical failure due to disconnection. Further, the large size defect 1203 may cause a short circuit or disconnection at any position on the pattern.

  A defect rate curve 1220 is created based on these theories. In addition, the defect occurrence rate curve 1221 is created based on the defect density data 1211 based on the results in the production line or based on theoretical formulas. The average fatality rate θ is obtained by obtaining the shaded portion 1230 from the product sum of these two curves 1220 and 1221.

  In this embodiment, in the semiconductor chip in FIG. 2, the respective products (S1 × θ1, S2 × θ2) of the areas S1, S2, S3 and the average lethalities θ1, θ2, θ3 of the blocks having different patterns in FIG. , S3 × θ3), and this is used as the importance of defect detection. On the other hand, the defect detection sensitivity of each block is obtained for a plurality of detection conditions by the method described with reference to FIG. 1, and the detection conditions that can obtain the optimum detection sensitivity are selected by comparing these.

  The selection criteria are as follows: (1) If there is a block with a high degree of importance in detecting defects, a detection condition that provides the highest sensitivity for that block is selected. (2) If there is no significant difference in the importance of defect detection in each block, select a detection condition that maximizes the average detection sensitivity of each block. Such a method is conceivable.

  FIG. 13 is an explanatory diagram of the first usage of the present invention. This is a method of using the present invention in the case where the development of test conditions is performed in accordance with the development of the system LSI type.

  In the development of LSI products, the basic process (vertical structure, material, etc.) is the same and the pattern design data changes. Therefore, in this method, the basic pattern calculation data 122A already calculated and the new pattern design data are changed. By performing the integration processing (124) on 111, it is possible to create an estimated image of the inspection area to be inspected that has been newly developed.

  It is also possible to create a combined (125) image with the created estimated image of the inspection region using the already calculated defect calculation data 123A. Based on these estimated images and synthesized images, it is possible to easily obtain the inspection conditions of the inspection object newly developed by calculation.

  FIG. 14 is an explanatory diagram of an apparatus configuration example of the present invention. 1411 is a pattern design data server, and 1412 is a storage unit for pattern design data. Reference numeral 1421 denotes a process condition data server, and 1422 denotes a process condition data storage unit. Reference numeral 1431 denotes an inspection condition simulation apparatus, which is a part for obtaining the inspection conditions by executing the simulation described with reference to FIGS.

  Reference numeral 1432 denotes a data storage unit. In FIG. 1, the material constant data 113, the defect data 114, or the basic pattern detection output calculation data 122A, the defect detection output calculation data 123A, the estimated image of the inspection area to be inspected, and the defect image And the composite image is saved.

  The pattern design data server 1411, the process condition data server 1421, and the inspection condition simulation apparatus 1431 are connected to each other via a local area network (LAN) 1401, and can exchange various data.

  Further, 1440 is an inspection apparatus A, and 1450 is an inspection apparatus B. The format may be any of a dark field optical type, a bright field optical type, and an electron beam type. Reference numeral 1460 denotes a defect review apparatus that reviews defects detected by the inspection apparatus. Reference numeral 1400 denotes a wafer to be inspected, which is inspected by the inspection apparatus A (1440) or B (1450) and then reviewed by the defect review apparatus 1460. The inspection condition simulation device 1431, the inspection device A (1440), the inspection device B (1450), and the defect review device 1460 are connected to each other via a local area network (LAN) 1202, etc., and various data can be exchanged. Is possible.

  For example, the inspection condition is set from the inspection condition simulation device 1431 to the inspection device A (1440) and the inspection device B (1450), or from the inspection device A (1440) and the inspection device B (1450) to the inspection condition simulation device 1431 It is possible to transfer the inspection image to the image, or to transfer new defect information (shape, material) from the defect review device 1460 to the inspection condition simulation device 1431 and store it in the data storage unit 1432 as defect data. Yes. By configuring the apparatus in this way, it is possible to set the inspection conditions efficiently by smoothly exchanging information.

  FIG. 15 is an explanatory diagram of the second usage method of the present invention. In defect inspection, for example, there is a demand to detect defects detected in bright-field optical inspection also in dark-field optical inspection, and this embodiment explains how to use the present invention to solve this. It is.

  Reference numeral 1510 denotes data obtained as a result of inspecting the wafer by the bright field optical inspection apparatus A, 1511 is the wafer, 1512 is the chip, 1513 is the coordinate data of the detected defect, and 1514 is the defect detection image. Reference numeral 1520 denotes defect review result data, which is data such as the shape and material of the defect. Reference numeral 1530 denotes a first simulation. The defect from which the data of 1510 is obtained is modeled based on the defect review result data 1520, and the first defect detection output calculation based on the simulation corresponding to the bright field optical system is used to calculate the first defect. The defect detection calculation image is obtained.

  This is compared with the actual detection image 1514 to verify the modeling accuracy. If the accuracy is not obtained, the modeling is repeated until the accuracy is obtained, and recalculation is performed. After the modeling accuracy is obtained, the second simulation 1540 is executed. Here, the second defect detection output calculation based on the second simulation corresponding to the dark field optical system is performed to obtain the second defect detection calculation image.

  This is combined with an estimated image stored in the already created data storage unit 1543 and corresponding to the inspection object and detection conditions (if not, an estimated image is newly created), and whether or not defects can be detected from the synthesized image Decision 1544 is made. If the detection is impossible, the detection condition is changed, and the calculation is repeated again to search for a detection condition capable of detecting the defect and set in the inspection apparatus (1550).

  FIG. 16 is an explanatory diagram of an example of an operation screen in the present invention. FIG. 16A is an example of an operation screen in disassembling pattern design data into basic patterns (121 in FIG. 8), 1601 is a product selection list, and 1602 process selection list. These are a list of available pattern design data and process condition (vertical structure / material) data. 1603 displays pattern design data corresponding to the product and process selected from the list, and 1604 displays process condition data. is there.

  Reference numeral 1605 is a definition list for definition when the pattern design data is decomposed into basic patterns, and 1606 is a display of the decomposed basic patterns. Thus, the efficiency of work can be raised by comprising so that various data can be listed.

  FIG. 16B is an example of the operation screen in the detection output simulation (122 in FIG. 8), and 1603 is a display of the decomposed basic pattern. 1607 is a selection list of inspection apparatus methods, which selects the above-described dark field optical method, bright field optical method, and electron beam method, and selects an inspection device when there are multiple models among the same methods. .

  An illumination parameter selection list 1608 is used to select illumination parameters that can be taken by the selected inspection method / apparatus. After selecting the conditions, the basic pattern detection output simulation is performed. Reference numeral 1609 is a display of the calculation result of the detection output of the basic pattern. Thus, the efficiency of work can be raised by comprising so that various data can be listed.

  FIG. 16 (3) is an example of the operation screen in the imaging calculation simulation (840 in FIGS. 1 and 8), and 1610 is a display of the result of integrating the basic pattern into the pattern design data. Reference numeral 1611 denotes a detection parameter selection list for selecting detection parameters that can be taken by the selected inspection method / apparatus. After selecting the conditions, an imaging calculation simulation is performed to produce an estimated image.

  Reference numeral 1613 denotes a display of the calculated estimated image. If an actual sample image is already obtained with the same illumination and detection parameters as the above calculation conditions, the actual image display 1614 is displayed alongside the estimated image display 1613 for the purpose of calculation verification. May be. Thus, the efficiency of work can be raised by comprising so that various data can be listed.

  FIG. 16 (4) is an example of an operation screen when selecting a detection condition. 1615 is an estimated image of the same object calculated by changing the detection conditions (illumination parameters and detection parameters). Illumination parameter conditions corresponding to images, and 1617 are detection parameter conditions corresponding to each image. Reference numeral 1618 denotes a defect detection sensitivity obtained by calculation. Thus, by making it possible to simultaneously check the estimated image and the defect detection sensitivity for a plurality of detection conditions, it becomes easy to select an optimal detection condition.

  In the above description, the example of the inspection of the semiconductor wafer substrate has been described. However, the present invention is not limited to the above-described embodiment, and can be applied to an inspection method and an inspection apparatus for an inspection object with a pattern, for example, a flat panel display. (Liquid crystal display, plasma display, organic EL display, etc.) and substrate inspection of patterned storage products (DTM: Discrete Track Media, BPM: Bit Pattern Media).

111 ... Pattern design data, 112 ... Process condition data, 200 ... Semiconductor wafer, 201 ... Semiconductor chip, 30 ... Element isolation layer, 311 ... Gate electrode part, 312 ... Contact part, 32-35 ... Wiring layer, 361 ... Scratch, 362 ..., 363 ... disconnection, 364 ... foreign matter, 500 ... wafer, 510 ... illumination optical system, 513 ... wavelength plate, 520 ... detection optical system, 521 ... objective lens, 525 imaging lens, 524 ... spatial filter, 531 ... sensor , 532 ... Image processing unit, 540 ... Stage, 550 ... Overall control unit, 610 ... Illumination optical system, 612 ... Condenser lens, 620 ... Detection optical system, 624 ... Polarized plate, 631 ... Sensor, 632 ... Image processing unit, 640 ... Stage, 650 ... Overall control unit, 710 ... Electronic lens barrel, 711 ... Electron gun, 712 ... Condenser lens, 713 ... Deflector, 714 ... Objective lens, 811 ... Basic pattern, 821 ... Scattered light from basic pattern Radiant intensity distribution, 831 ... pattern design data 841 ... Image calculation result, 851 ... Detected image considering sensor pixel resolution, 902 ... Line edge roughness, 903 ... Corner round, 921 ... Pattern height variation, 931 ... Film thickness Fluctuation, 1110 ... Image to be detected, 1111 ... Sensor pixel, 1212 ... Defect occurrence rate curve, 1220 ... Defect rate curve, 1510 ... Inspection data, 1511 ... Wafer, 1512 ... Chip, 1513 ... Defect coordinate data.

Claims (19)

In a defect inspection method for an inspection object having a pattern formed on the surface,
One or more basic patterns are extracted from the pattern design data formed on the inspection target, and the detection output of the basic patterns is calculated by physical simulation;
Create a pattern estimation image corresponding to the pattern design data by integrating the detection output of the basic pattern,
A defect inspection method, wherein an inspection condition is defined based on the pattern estimation image, and the inspection condition is set in an inspection apparatus.   The defect to be inspected is modeled, a defect estimation image is created by physical simulation, a defect detection sensitivity is calculated based on a composite image obtained by synthesizing the defect estimation image and the pattern estimation image, and the defect detection sensitivity is based on the defect detection sensitivity. 2. The defect inspection method according to claim 1, wherein an inspection condition is defined.   3. The defect inspection method according to claim 2, wherein a defect detection sensitivity is calculated based on both the pattern estimation image and the composite image, and an inspection condition is determined based on the defect detection sensitivity.   The pattern estimation image, the defect estimation image, and the composite image are calculated as a plurality of images corresponding to each of a plurality of detection conditions that can be set by an inspection apparatus, and have the highest detection sensitivity in consideration of the plurality of images. The defect inspection method according to claim 2, wherein the inspection condition is determined based on a high inspection condition.   The pattern estimation image, the defect estimation image, and the composite image are calculated in consideration of variations in detection signals, according to any one of claims 2 to 4. Defect inspection method.   6. The defect inspection method according to claim 5, wherein the variation is a variation in the inspection target itself, such as edge roughness, corner rounding, pattern height, and film thickness of the pattern to be inspected.   6. The defect inspection method according to claim 5, wherein the variation is caused by variation in aberration of a detection optical system of the inspection apparatus.   The defect inspection method according to claim 5, wherein the variation is caused by variations in image sampling in an inspection apparatus.   Based on a plurality of images calculated corresponding to each of the plurality of inspection regions, the defect detection sensitivity in each inspection region is calculated, and this is compared with the separately calculated yield prediction calculation result, thereby detecting the inspection conditions. The defect inspection method according to claim 1, wherein the defect inspection method is defined.   The defect inspection method according to claim 1, wherein the inspection method is a dark-field optical detection method.   The defect inspection method according to claim 1, wherein the inspection method is a bright-field optical detection method.   The defect inspection method according to claim 1, wherein the inspection method is an electron beam detection method. In the defect inspection equipment to be inspected with a pattern formed on the surface,
Means for holding pattern design data formed on the inspection object;
Means for extracting one or more basic patterns from the pattern design data;
Means for calculating a detection output from each of the extracted basic patterns by physical simulation;
Means for creating a pattern estimation image obtained from the inspection region to be inspected by integrating the calculation results of the basic pattern corresponding to the pattern design data;
A defect inspection apparatus comprising: means for determining an inspection condition based on the pattern estimation image, and configured so that the determined inspection condition can be set as an apparatus parameter. Means for modeling the defect to be inspected and creating a defect estimation image by physical simulation;
Means for creating a synthesized image obtained by synthesizing both the defect estimated image and the pattern estimated image;
Means for calculating defect detection sensitivity based on the composite image;
Means for determining an inspection condition based on the calculated defect detection sensitivity,
14. The defect inspection apparatus according to claim 13, wherein the predetermined inspection condition can be set as an apparatus parameter. Means for creating a composite image obtained by combining the pattern estimation image of the inspection object and the defect estimation image;
Defect detection sensitivity is calculated based on both the pattern estimation image and the composite image,
Means for determining an inspection condition based on the calculated defect detection sensitivity;
14. The defect inspection apparatus according to claim 13, wherein the predetermined inspection condition can be set as an apparatus parameter. Means for calculating defect detection sensitivity in each inspection area based on a plurality of images calculated corresponding to each of the plurality of inspection areas;
The defect inspection apparatus according to any one of claims 13 to 15, further comprising means for determining an inspection condition by comparing a separately calculated yield prediction calculation result.   The defect inspection apparatus according to claim 13, wherein the inspection means is a dark field optical detection method.   The defect inspection apparatus according to any one of claims 13 to 16, wherein the inspection means is a bright-field optical detection system.   The defect inspection apparatus according to any one of claims 13 to 16, wherein the inspection means is an electron beam detection system.

Images