JP5043741B2 - Semiconductor pattern inspection method and inspection apparatus - Google Patents

Description

  The present invention relates to a method for inspecting a pattern of a semiconductor wafer using a scanning electron microscope (hereinafter referred to as SEM), and more particularly to a recipe optimization process.

  In the manufacturing process of a semiconductor wafer, first, a coating material called a resist is applied on the semiconductor wafer. Next, an exposure mask (reticle) of a wiring pattern is overlaid on the resist. Next, visible light, ultraviolet light, or an electron beam is irradiated from above, and the resist is exposed. Thereby, a wiring pattern is formed on the semiconductor wafer.

  The shape of the wiring pattern formed on the semiconductor wafer varies depending on the intensity and aperture of visible light, ultraviolet light, or electron beam. Therefore, in order to form a highly accurate wiring pattern, it is necessary to inspect the wiring pattern.

  Conventionally, a critical dimension scanning electron microscope (CD-SEM) has been widely used for inspecting a wiring pattern formed on a semiconductor wafer. In the inspection of the semiconductor pattern, an image at a measurement point (MP) set in advance on the wiring pattern is acquired. From this image, various dimension values such as the wiring width of the pattern are measured, and the performance of the pattern is evaluated.

  Imaging and measurement of the wiring pattern are executed according to the recipe. The recipe describes the procedure, conditions, and the like of the wiring pattern imaging and measurement.

  The recipe is created by the operator registering data manually or by automatically registering design data.

  In the case of manual registration by an operator, knowledge and experience of the operator are required. It is difficult for beginners and less experienced operators to create recipes that operate correctly and stably. It is also difficult to determine whether the recipe setting created in this way is appropriate from the viewpoint of reducing damage to the sample and time efficiency. For this reason, when a recipe created by manual registration is executed, problems such as failure of imaging, significant damage to the sample, and the time required to execute the recipe occur. Such recipes require optimization.

  In the case of handwriting or automatic registration of a recipe based on design data, it is necessary to check the presence or absence of defects by executing the created recipe in an actual scanning microscope apparatus. If there is a problem, the operator must manually optimize the recipe.

  The recipe optimization process is a process in which settings are manually changed for each measurement point (MP). Even if an operator having knowledge and skills performs this work, a great deal of time and labor are consumed. If there are many measurement points (MP) that need to be evaluated, the operator's work time increases. Further, the apparatus is occupied for a long time by the process of optimizing the recipe. Therefore, it becomes a big problem in semiconductor manufacturing factories.

  An object of the present invention is to provide a semiconductor pattern inspection technique having a function of automatically performing recipe quality diagnosis and optimization work.

  According to the present invention, a low-magnification SEM image is acquired by a scanning electron microscope for a region including a measurement point set on a semiconductor pattern. Next, it is determined whether an addressing pattern, an autofocus pattern, and an autostigma pattern registered in advance as a recipe satisfy a predetermined recommended condition. When it is determined that the predetermined recommended condition is not satisfied, an optimum addressing pattern, autofocus pattern, and autostigma pattern are selected from the SEM image.

  When selecting an optimum pattern, a predetermined region centered on the measurement point is cut out in the SEM image, and the cut out image is further cut out to generate a plurality of block images. Create a histogram and brightness profile. An optimum pattern is selected from the luminance histogram and the luminance profile.

  According to the present invention, a semiconductor pattern inspection technique having a function of automatically performing recipe quality diagnosis and optimization work is provided.

  An outline of a scanning electron microscope apparatus according to the present invention will be described with reference to FIG. The scanning electron microscope of this example includes a cathode 8, a first anode 9, a second anode 10, a converging lens 12, a diaphragm plate 15, a converging lens 13, a scanning coil 16, and an orthogonal electromagnetic field (EXB) generator for secondary signal separation. 18 and an objective lens 14. A one-stage deflection coil 37 is disposed near or at the same position as the scanning coil 16. The deflection coil 37 operates as an objective lens adjuster that corrects the axial deviation of the objective lens 14. In addition, an astigmatism correction coil 38 and an adjuster 39 are disposed in the vicinity of the converging lens 13 or at the same position. The astigmatism correction coil 38 is a multi-pole coil for correcting astigmatism in the X and Y directions. The adjuster 39 functions as an astigmatism correction coil adjuster that corrects the axial deviation of the astigmatism correction coil 38.

  The scanning electron microscope apparatus of this example further includes a secondary signal detector 20, a signal amplifier 21, an image memory 28, an image display device 29, an image processing device 30, a computer 34, an input device 35, and a storage device 36. Have.

  The scanning electron microscope apparatus of this example further includes a high voltage control power source 23, a first convergent lens control power source 24, a second convergent lens control power source 25, an objective lens control power source 26, a scan coil control power source 27, and an objective lens adjuster control. A power supply 31, an astigmatism correction coil control power supply 32, and an astigmatism correction coil adjuster control power supply 33 are included. These power supplies are controlled by the computer 34.

  A voltage is applied between the cathode 8 and the first anode 9 by the high voltage control power source 23. Thereby, the primary electron beam 11 of a predetermined emission current is drawn from the cathode 8. An acceleration voltage is applied between the cathode 8 and the second anode 10 by the high voltage control power source 23. As a result, the primary electron beam 11 is accelerated and proceeds to the subsequent lens system. The primary electron beam 11 is converged by the converging lens 12, unnecessary electrons are removed by the aperture plate 15, converged by the converging lens 13, and minutely on the sample 17 held on the sample stage 22 by the objective lens 14. Converged as a spot.

  The objective lens 14 can take various forms such as an in-lens system, an out-lens system, and a snorkel system (semi-in-lens system). Further, the scanning electron microscope of this example may be a retarding method in which a voltage is applied to the sample to decelerate the primary electron beam. Further, the converging lens may be composed of an electrostatic lens composed of a plurality of electrodes.

  The primary electron beam 11 is scanned two-dimensionally on the sample 17 by the scanning coil 16. When the primary electron beam is irradiated, secondary electrons and the like are generated from the sample 17. The secondary signal 19 generated from the sample 17 travels to the upper part of the objective lens 14 and is then separated from the primary electrons by the secondary signal separation orthogonal electromagnetic field (EXB) generator 18 to the secondary signal detector 20. Detected. The signal detected by the secondary signal detector 20 is amplified by the signal amplifier 21 and then transferred to the image memory 28. The image transferred to the image memory 28 is displayed on the image display device 29 as a sample image.

  A one-stage deflection coil 37 (objective lens adjuster) is arranged in the vicinity of or at the same position as the scanning coil 16 and operates as an adjuster for correcting the axial deviation of the objective lens 14. In addition, an astigmatism correction coil 38 having multiple poles for correcting astigmatism in the X and Y directions is disposed between the objective lens 14 and the diaphragm plate 15. An adjuster 39 (astigmatism correction coil adjuster) that corrects the axial deviation of the astigmatism correction coil is arranged in the vicinity of the astigmatism correction coil 38 controlled by the astigmatism correction coil control power supply 32 or at the same position. . The objective lens adjuster 37 is controlled by an objective lens adjuster control power supply 31, and the astigmatism correction coil adjuster 39 is controlled by an astigmatism correction coil adjuster control power supply 33. An image processing device 30, a storage device 36, and an input device 35 are also connected to the computer 34.

  The image memory 28, the image display device 29, the image processing device 30, the computer 34, the input device 35, and the storage device 36 execute recipe diagnosis and optimization processing described below. The recipe diagnosis and optimization process is configured as a program executable by the computer 34. Recipes and various patterns, template images, SEM images, contact conditions, optical conditions, and the like are stored in the storage device 36.

  FIG. 2 shows an example of an image 300 of a wiring pattern used for semiconductor pattern inspection. In the wiring pattern inspection, in addition to the measurement point (MP) 301 to be evaluated, an addressing point (hereinafter referred to as AP) or an addressing pattern 302, an autofocus point (hereinafter referred to as AF) or an autofocus pattern 303, an autostigma point (hereinafter referred to as AST) or an autostigma pattern 304 is used.

  Measurement points 301 to be evaluated in the wiring pattern are set in advance. It is difficult to find the measurement point 301 on a low-magnification image. Therefore, a characteristic pattern or shape that can be easily found is set in the vicinity of the measurement point 301. This is the addressing pattern 302. The deviation (coordinates) between the addressing pattern 302 and the measurement point 301 is known. Addressing is a process of searching for an addressing pattern 302 and reaching the measurement point 301 therefrom. First, the addressing pattern 302 is visually found on a wide image with a relatively low magnification. Next, the field of view is moved by a predetermined deviation on the relatively high magnification image. Thereby, an image in which the measurement point 301 is the center of the visual field is obtained.

  The autofocus pattern 303 is a pattern used for performing autofocus adjustment. An auto stigma pattern 304 is a pattern used to perform auto stigma adjustment. Usually, autofocus and autostigma adjustment are performed at points other than the measurement point 301. This is because the sample surface is damaged when the measurement point is repeatedly irradiated with an electron beam. In order to avoid this, the autofocus pattern 303 and the autostigma pattern 304 are set at points different from the measurement point 301. Deviations (coordinates) from the autofocus pattern 303 and the autostigma pattern 304 to the measurement point 301 are registered in advance.

  The measurement point (MP) 301, the addressing pattern 302, the autofocus pattern 303, and the autostigma pattern 304 are referred to as imaging points. The coordinates of the imaging point, the size and shape of the pattern, the imaging sequence, the imaging conditions, and the template image of the addressing pattern 302 are managed as an imaging recipe.

  The deviation between the addressing pattern and the measurement point is determined by comparing the SEM image of the addressing pattern with known coordinates registered in advance as a template with the SEM image (actual imaging template) captured in the actual imaging sequence. It is obtained by calculating | requiring the deviation of.

  With reference to FIG. 3A and FIG. 3B, the process which performs length measurement with a scanning electron microscope apparatus is demonstrated. The procedure of measurement processing, setting conditions, etc. are set in advance as a recipe. Therefore, the length of the pattern is measured by executing the recipe. With reference to FIG. 3B, a normal recipe execution process will be described. In step S101, alignment is executed. In step S102, the visual field is moved to the coordinates of the measurement point. Thereby, the visual field including the measurement point is obtained. In step S103, addressing is performed. First, an addressing pattern is searched. Next, the field of view is moved from the addressing pattern by a predetermined offset amount. Thereby, a field of view in which the measurement point (MP) is arranged at the center is obtained. The offset amount from the addressing pattern to the measurement point (MP) is registered in advance. In step S105, the measurement point (MP) is moved to the autofocus pattern to perform autofocus. Moreover, it moves from the measurement point (MP) to the auto stigma pattern and performs auto stigma. Deviations from the measurement point (MP) to the autofocus pattern and the autostigma pattern are registered in advance.

  With reference to FIG. 3A, the execution process of the recipe by this invention is demonstrated. According to the present invention, when executing the recipe, the optimization information is acquired in step S104. Here, a large area image taken with a wide field of view is acquired. The wide field of view is a dimension based on the movable limit range of beam scanning. In Hitachi measuring SEM, one side is a 30μm square.

  Autofocus includes low-magnification autofocus and high-magnification autofocus. When both are executed, a large area image is acquired for each. All of these images have good resolution, high contrast, and the presence of a pattern can be easily confirmed.

  Usually, in order to obtain a wide-field image, the image is taken at a low magnification. However, when the magnification is low, the display area per pixel becomes large and the resolution is lowered. In this case, the number of pixels at the time of image acquisition may be increased, or a high-field image may be generated by joining high-magnification images. Thereby, a decrease in resolution can be avoided.

  In step S105, it moves to an autofocus pattern and performs autofocus. Moreover, it moves to the pattern for auto stigma and performs auto stigma. After performing auto focus adjustment and auto stigma adjustment, an image including a measurement point is acquired.

  In this way, in this example, three types of images are obtained: an image immediately after execution of autofocus adjustment, an image immediately after execution of autostigma, and a large area image obtained by photographing the vicinity of the measurement point (MP) with a wide field of view.

  The recipe optimization process will be described with reference to FIG. The target of optimization is the shape and position of the addressing pattern, autofocus pattern, and autostigma pattern, and the execution conditions of addressing, autofocus, and autostigma. Using the three images acquired in step S104 of FIG. 3B, the measurement results at that time, and the operation information of the apparatus (required time, matching score at the time of image recognition, etc.), a recipe optimization process is performed. .

  In step S401, recipe diagnosis is performed. That is, it is determined whether the coordinates, shape, S / N, and magnification of each pattern are appropriate. First, the recipe conditions already set are compared with the recommended conditions. This comparison is performed for each setting item. According to this example, an evaluation value (score) for quantitatively expressing the appropriateness of setting is given for each setting item. The score is compared with a threshold value to determine whether optimization is necessary. Next, the shape of the pattern is confirmed by an image taken immediately after step S104 and step S105 in FIG. 3B. Further, a plurality of blocks are cut out from the large area image, and a luminance histogram (FIG. 5) and profile (FIG. 8) are obtained for each block, thereby determining whether or not the recommended conditions are met.

  Examples of recipe diagnosis are shown in FIGS. As a result of the recipe diagnosis, if there is a problem with the recipe, the process proceeds to step S404 via step S402. If there is no problem with the recipe, the process proceeds to step S405 via step S403. In step S405, the recipe optimization process is terminated. In step S404, optimization is started. Specifically, the following steps S406 to S408 are executed.

  In step S406, an optimal addressing pattern is selected and a template image is created. In step S104 of FIG. 3A, a large area image photographed with a wide field of view was acquired. This large area image is cut out, an optimum addressing pattern is selected, and a template image is created. In step S407, optical conditions are set. Optical conditions include magnification, frame integration number, and the like. Set the optical conditions for the addressing pattern, autofocus pattern, and autostigma pattern. In step S408, the coordinates of the pattern are changed. That is, the optimum addressing pattern, autofocus pattern, and autostigma pattern are selected and their positions are registered. In the case of a continuous pattern, even if the position of the pattern is changed, the shape of the pattern does not change. However, in the case of a discontinuous pattern, changing the pattern position changes the pattern shape. These steps S406 to S408 will be described in detail later with reference to FIGS. When the optimization is thus completed, a new recipe is completed in step S409.

  Since the recipe optimization process uses information acquired in advance, it is not always necessary to work on the apparatus main body, and can be executed by an external apparatus connected to the apparatus.

  In steps S406 and S408, the optimum addressing pattern, autofocus pattern, and autostigma pattern are selected. Hereinafter, the process of selecting an optimum pattern will be described with reference to FIGS.

  This will be described with reference to FIG. Here, the optimum addressing pattern, autofocus pattern, and autostigma pattern candidate are selected using the luminance histogram. The large area image acquired in step S104 of FIG. 3A is divided into several blocks. For example, a region having one side of 15 μm centered on the measurement point (MP) is divided into a plurality of blocks. The dimension of each block corresponds to the visual field range (6.75 μm) assuming that the magnification suitable for addressing is 20 k times, for example. The blocks are arranged so as to overlap each other. By generating a large number of blocks in this way, the amount of information can be increased. A luminance histogram is created for each block.

  Curves 48 to 51 in FIG. 5 show examples of luminance histograms created for each block. In the SEM image, there is a correlation between the presence / absence and shape of the wiring pattern and the luminance of the image. The luminance change of the image represents the presence / absence of a pattern. It can be said that there are many patterns in the screen in an image with a large luminance change. Therefore, the pattern is expressed as the peak width of the luminance histogram. When there is no pattern, there is no change in luminance, and the screen has uniform luminance. In this case, the peak width is narrowed.

  Further, the luminance of the entire screen appears as a peak position on the histogram. That is, the peak position shifts to the low luminance side to the left of the center in an image with a dark entire screen, and the peak position shifts to the high luminance side to the right of the center in an image with a bright entire screen. Ideally, the entire screen has a moderate brightness and a large amount of pattern information.

  The curve 48 in FIG. 5 has a wide luminance range, that is, the luminance is not uniform. Therefore, such an image is appropriate as a pattern. The peak is skewed to the right as in curve 49, the peak is skewed to the left as in curve 50, or the average position of peak luminance is almost at the center as in curve 51, but the peak width is narrow and the field of view Those that have only uniform brightness as a whole are inappropriate.

  In this way, addressing pattern, autofocus pattern, and autostigma pattern candidates are detected from the shape of the luminance histogram.

  Next, an addressing pattern, an autofocus pattern, and an autostigma pattern are selected from these patterns as described below. For this optimal pattern detection, a profile representing the luminance distribution in the screen is used. This will be described with reference to FIG.

  With reference to FIG. 8, a method for selecting an optimum addressing pattern, autofocus pattern, and autostigma pattern will be described. Create a profile that represents the luminance distribution in the screen. The horizontal axis of the profile is the position in the screen (pixel arrangement), and the vertical axis is the luminance. Create an X-direction integration profile and a Y-direction integration profile.

First, a method for selecting an addressing pattern will be described. In addressing, a template image and an actual image are matched. Therefore, the addressing pattern is required to have a unique profile shape. Specificity means that the peak shape, position and number are different compared to other parts. As shown in the figure, the necessary condition for the addressing pattern is that there is a peak of luminance in at least one direction, and that peaks of the same shape are not repeated.
It is desirable that the autofocus pattern and the autostigma pattern have patterns of the same shape repeatedly around them. Therefore, the luminance distribution profile is preferably similar to that obtained with the surrounding blocks.

  As shown in the figure, a necessary condition for the autofocus pattern is that there is a peak in at least one direction. The same shape of peaks may be repeated.

  The autostigma pattern is preferably a pattern having edges in all directions. That is, a pattern having a peak in each of the profiles in the X direction and the Y direction is selected. As shown in the figure, the necessary condition for the autostigma pattern is that there are luminance peaks in two directions, the X direction and the Y direction, and that the peaks of the same shape are not repeated.

  Thus, in this example, a pattern candidate is selected from the luminance histogram, and an optimum addressing pattern, autofocus pattern, and autostigma pattern are selected from the luminance profile. Also, by automatically registering the imaging conditions as a new recipe, the complexity of conventional recipe optimization and the dependency on operator skill can be reduced.

  With reference to FIG. 6, the diagnosis and optimization of a recipe are demonstrated about the pattern for addressing. This corresponds to step S401 and step S406 in FIG. First, in step S601 to step S604, it is determined whether or not the coordinates, pattern shape, S / N, and magnification are appropriate for the addressing pattern. If at least one of these is inappropriate, the process proceeds to step S606. In step S606, an addressing pattern optimization process is performed. Here, a template image for matching is registered.

  First, one pattern is selected by two processes (histogram process and profile process). This is an addressing pattern. The two processes have been described with reference to FIGS. Next, the distance from the addressing pattern to the measurement point (MP) is measured on the large area image acquired in step S104 of FIG. 3B, and the relative distance from the measurement point (MP) to the addressing pattern is measured. Set as coordinates.

  Next, this addressing pattern portion is cut out from the large area image. The image cut out in this way is registered as a template image for matching. When the image is cut out, image processing such as edge enhancement may be added to facilitate use as a template image. Next, an appropriate image acquisition condition is set from the conditions at the time of acquiring a large area image.

  Image acquisition conditions are a magnification and the number of integrated frames. By setting this to an appropriate value, the S / N ratio is high, and a template image suitable for matching can be acquired.

  For example, a large area image with a side of 30 μm is created by joining images of 50 frames (view range 2.7 μm) and 8 frames. Next, an image block of 20k times (viewing field range 6.75 μm) is cut out from it and used as a template image. In this case, since the irradiation range of the beam is widened to 50/20, the beam irradiation amount per unit area is reduced. For this reason, the amount of signal obtained is also reduced, and the S / N ratio of the actual image is deteriorated.

  Therefore, if the cumulative number of images is set to 20, which is 50/20 times, the beam irradiation amount per unit area becomes equal. Therefore, it is possible to ensure an S / N ratio equivalent to that of the large area image.

  The relative coordinates of the addressing pattern obtained as described above, the template image for matching, and the imaging conditions are set in the recipe as new conditions.

  With reference to FIG. 7, recipe diagnosis and optimization will be described for the autofocus pattern and the autostigma pattern. In step S701 and step S702, it is determined whether or not at least one of the coordinates and the shape of the pattern is inappropriate for the autofocus pattern and the autostigma pattern. If not appropriate, the process proceeds to step S707.

  Step S707 will be described. First, a plurality of autofocus patterns and autostigma patterns are selected by histogram processing (FIG. 5). At this time, the size of the block to be divided is set to the size of the visual field range FOV at the execution magnification of autofocus and autostigma.

  The autofocus pattern is required to be a portion where the same pattern is repeatedly present in the surroundings so that it can cope with a case where the field of view is shifted. Autofocus is performed before and after addressing. Autofocus before addressing is performed after mechanically moving to the autofocus point. Therefore, there is a possibility that autofocus is performed at a point deviated from the position of the autofocus pattern 303. In other words, the autofocus before addressing and the autofocus after addressing need to be able to cope with the deviation of coordinates. Accordingly, a portion where a similar pattern exists around is selected from a plurality of patterns. A profile (FIG. 8) is used for this selection.

  The auto stigma corrects the beam shape from eight directions. Therefore, the autostigma pattern is required to be a pattern having edges in each of the X direction and the Y direction. Therefore, a portion where a pattern having edges in both the X direction and the Y direction exists is selected from the plurality of patterns. A profile (FIG. 8) is used for this selection. For each of the selected patterns, profiles in the X direction and the Y direction are created, and a pattern having peaks in both is selected.

  After the autofocus pattern and the autostigma pattern are selected in this way, the measurement is performed as relative coordinates from the measurement point (MP). On the large area image acquired in step S104 of FIG. 3B, the distance from the autofocus pattern and the autostigma pattern to the measurement point (MP) is measured, and the distance from the measurement point (MP) is measured. Set as relative coordinates. Template images are not created for autofocus patterns and autostigma patterns. Therefore, it is not necessary to cut out from the large area image.

  In steps S703 and S704, it is determined whether or not at least one of S / N and magnification is inappropriate for the autofocus pattern and the autostigma pattern. If not appropriate, the process proceeds to step S706.

  In step S706, the setting is changed so that the setting satisfies the recommended condition. At the same time, an image immediately after execution of autofocus (hereinafter, referred to as an autofocus pattern image) acquired at the time of executing the recipe is analyzed. An image immediately after execution of autostigma adjustment (hereinafter referred to as autostigma pattern image) is analyzed.

  As a result of changing the setting, if the S / N and the magnification become appropriate, the process proceeds to step S708, and the optimization process is terminated. If the S / N and the magnification are still inappropriate even after the setting is changed, the process proceeds to step S707 and the pattern is changed. For example, if the S / N shortage is observed even if the conditions are met, or if the pattern cannot be improved by adjusting the magnification (the magnification is characteristic, but the magnification can be changed, for example If it does, the pattern that just becomes a straight line) is set as a pattern NG.

  The above-described recipe optimization process can be performed automatically, or some options can be presented at each stage, and the operator can select and set them.

  In addition, a plurality of conditions may be presented based on the recommended score so that the options presented at that time can be selected according to the priority of important conditions (throughput, damage, accuracy, etc.) required by the operator for the recipe.

  According to the present invention, the burden on the operator's manual work is reduced and the optimization of the recipe becomes easier. In addition, since the appropriateness of the conditions becomes quantitative, it is possible to set a recipe that always operates stably without relying on the operator's knowledge and eliminating errors caused by manual work.

  The example of the present invention has been described above. However, the present invention is not limited to the above-described example, and various modifications can be easily made by those skilled in the art within the scope of the invention described in the claims. Will be understood.

It is a figure explaining the outline of the scanning electron microscope apparatus by this invention. It is a figure which shows the example of the image of the wiring pattern used for the test | inspection of the semiconductor pattern by this invention. It is a figure explaining the process which performs length measurement according to the recipe by this invention in the scanning electron microscope apparatus by this invention. It is a figure explaining the process which performs length measurement according to a normal recipe in the scanning electron microscope apparatus by this invention. It is a figure explaining the example of the optimization process of the recipe in the test | inspection method of the semiconductor pattern by this invention. It is a figure explaining the example of the histogram of the brightness | luminance used with the test | inspection method of the semiconductor pattern by this invention. It is a figure explaining the diagnosis and optimization of a recipe about the pattern for addressing in the inspection method of the semiconductor pattern by the present invention. It is a figure explaining the diagnosis and optimization of a recipe about the pattern for autofocus, and the pattern for autostigma in the inspection method of the semiconductor pattern by the present invention. It is a figure explaining the example of the profile of the brightness | luminance used with the inspection method of the semiconductor pattern by this invention.

Explanation of symbols

8 ... cathode, 9 ... first anode, 10 ... second anode, 11 ... primary electron beam, 12 ... first focusing lens, 13 ... second focusing lens, 14 ... objective lens, 15 ... aperture plate, 16 ... scanning coil , 17 ... Sample, 18 ... Secondary signal separation orthogonal electromagnetic field (EXB) generator, 19 ... Secondary signal, 20 ... Secondary signal detector, 21 ... Signal amplifier, 22 ... Sample stage, 23 ... High pressure control Power source 24... First convergent lens control power source 25. Second convergent lens control power source 26. Objective lens control power source 27 27 Scanning coil control power source 28 28 Image memory 29 29 Image display device 30 Image processing device 31 ... Objective lens adjuster control power supply, 32 ... Astigmatism correction coil control power supply, 33 ... Astigmatism correction coil adjuster control power supply, 34 ... Computer, 35 ... Input device, 36 ... Storage device, 37 ... Objective Lens adjuster, 38 Astigmatism correction coil, 39 ... astigmatism correction coil adjuster

Claims (12)

In a semiconductor pattern inspection method using a scanning electron microscope provided with a storage device, an image processing device, and a computer,
Acquiring a low-magnification SEM image by a scanning electron microscope for a region including a measurement point set on a semiconductor pattern according to an imaging condition registered in the storage device;
An addressing pattern determination step for determining whether an addressing pattern registered in the storage device satisfies a predetermined recommended condition;
In the addressing pattern determination step, when it is determined that a predetermined recommended condition is not satisfied, an addressing pattern selection step for selecting an optimal addressing pattern from the SEM image;
A template image creating step of cutting out an area including the addressing pattern selected in the addressing pattern selection step from the SEM image and creating a template image of the addressing pattern;
A template image update step for replacing the template image registered in the storage device with the template image created in the template image creation step;
Have
In the addressing pattern determination step, in the SEM image, a predetermined region centered on the measurement point is cut out, the cut out image is further cut out to generate a plurality of block images, and for each of the block images, Create a brightness histogram and a brightness profile, and determine whether the addressing pattern satisfies a predetermined recommended condition from the brightness histogram and the brightness profile,
In the addressing pattern selection step, an optimal addressing pattern is selected based on the luminance histogram and the luminance profile. In the inspection method of the semiconductor pattern according to claim 1,
An addressing pattern deviation calculation step for calculating a deviation from the addressing pattern selected in the addressing pattern selection step to the measurement point;
An addressing pattern deviation update step for replacing the deviation from the addressing pattern registered in the storage device to the measurement point with the addressing pattern deviation calculated in the addressing pattern deviation calculating step;
A method for inspecting a semiconductor pattern, comprising: In the inspection method of the semiconductor pattern according to claim 1,
An addressing pattern update step for replacing the addressing pattern selected in the addressing pattern selection step with an addressing pattern registered in the storage device;
A method for inspecting a semiconductor pattern, comprising: In the inspection method of the semiconductor pattern according to claim 1,
In the addressing pattern determination step, the addressing pattern selection step is executed when at least one of coordinates, position, S / N, and magnification of the addressing pattern does not satisfy a predetermined recommended condition. Semiconductor pattern inspection method. In the inspection method of the semiconductor pattern according to claim 1,
An autofocus pattern determination step for determining whether or not the autofocus pattern registered in the storage device satisfies a predetermined recommended condition;
An autofocus pattern selection step for selecting an optimal autofocus pattern from the SEM image when it is determined in the autofocus pattern determination step that a predetermined recommended condition is not satisfied;
An autofocus pattern deviation calculating step for calculating a deviation from the autofocus pattern selected in the autofocus pattern selecting step to the measurement point;
An autofocus pattern deviation update step for replacing the deviation from the autofocus pattern registered in the storage device to the measurement point with the autofocus pattern deviation calculated in the autofocus pattern deviation calculation step;
Have
In the autofocus pattern determination step, a predetermined region centered on the measurement point is cut out from the SEM image, and the cut out image is further cut out to generate a plurality of block images. , Creating a luminance histogram and a luminance profile, and determining from the luminance histogram and the luminance profile whether or not the autofocus pattern satisfies a predetermined recommended condition,
In the autofocus pattern selection step, an optimum autofocus pattern is selected based on the brightness histogram and the brightness profile, and a semiconductor pattern inspection method. In the inspection method of the semiconductor pattern according to claim 5,
The semiconductor is characterized in that the autofocus pattern selection step is executed when at least one of the coordinates and position of the autofocus pattern does not satisfy a predetermined recommended condition in the autofocus pattern determination step. Pattern inspection method. In the inspection method of the semiconductor pattern according to claim 5,
If at least one of S / N and magnification of the autofocus pattern does not satisfy a predetermined recommended condition in the autofocus pattern determination step, the optical condition of the autofocus pattern registered in the storage device The method for inspecting a semiconductor pattern is characterized in that if the predetermined recommended condition is not satisfied, the autofocus pattern selection step is executed. In the inspection method of the semiconductor pattern according to claim 1,
An autostigma pattern determination step for determining whether or not an autostigma pattern registered in the storage device satisfies a predetermined recommended condition;
An autostigma pattern selection step for selecting an optimal autostigma pattern from the SEM image when it is determined that the predetermined recommended condition is not satisfied in the autostigma pattern determination step;
An auto stigma pattern deviation calculating step for calculating a deviation from the auto stigma pattern selected in the auto stigma pattern selecting step to the measurement point;
Autostigma pattern deviation update step for replacing the deviation from the autostigma pattern registered in the storage device to the measurement point with the autostigma pattern deviation calculated in the autostigma pattern deviation calculation step;
Have
The autostigma pattern determination step cuts out a predetermined area centered on the measurement point in the SEM image, further cuts out the cut-out image to generate a plurality of block images, and each of the block images A brightness histogram and a brightness profile are created, and from the brightness histogram and the brightness profile, it is determined whether or not the autostigma pattern satisfies a predetermined recommended condition,
In the autostigma pattern selection step, an optimal autostigma pattern is selected based on the luminance histogram and the luminance profile. The semiconductor pattern inspection method according to claim 8,
In the autostigma pattern determination step, when at least one of the coordinates and position of the autostigma pattern does not satisfy a predetermined recommended condition, the autostigma pattern selection step is executed. Pattern inspection method. The semiconductor pattern inspection method according to claim 8,
If at least one of S / N and magnification of the autostigma pattern does not satisfy a predetermined recommended condition in the autostigma pattern determination step, the optical conditions of the autostigma pattern registered in the storage device The semiconductor pattern inspection method is characterized in that if the predetermined recommended condition is not satisfied, the autostigma pattern selection step is executed.   A computer-readable program for causing a computer to execute the semiconductor pattern inspection method according to claim 1. In a semiconductor pattern inspection apparatus having a scanning electron microscope equipped with a storage device, an image processing device, and a computer,
According to the imaging conditions registered in the storage device, a low-magnification SEM image is obtained for an area including a measurement point set on a semiconductor pattern, and a predetermined area centered on the measurement point is obtained in the SEM image. Cut out, further cut out the cut out image to generate a plurality of block images, create a luminance histogram and a luminance profile for each of the block images, and based on the luminance histogram and the luminance profile, It is determined whether or not the addressing pattern, autofocus pattern, and autostigma pattern registered in the storage device satisfy a predetermined recommended condition, and when it is determined that the predetermined recommended condition is not satisfied, the SEM From the image, the most suitable addressing pattern, autofocus pattern, or Inspection apparatus configured semiconductor patterns to select the over preparative stigma pattern.

Images