JP5439106B2 - Pattern shape evaluation apparatus and method using scanning charged particle microscope - Google Patents

Description

The present invention relates to a method and apparatus for measuring or evaluating a circuit pattern on a sample such as a semiconductor wafer using a scanning charged particle microscope in the design or manufacturing process of a semiconductor device.

When forming a circuit pattern on a semiconductor wafer, a coating material called a resist is applied on the semiconductor wafer, and an exposure mask (reticle) for the circuit pattern is superimposed on the resist, and then visible light, ultraviolet light, or electron beam is applied from there. A circuit pattern is formed on a semiconductor wafer by irradiating the resist, exposing and developing the resist, and etching the semiconductor wafer using the resist circuit pattern as a mask. Etc. are adopted.

  In recent years, in order to meet the needs for higher speed and higher integration of semiconductor devices, super-resolution exposure technology represented by Optical Proximity Correction (OPC) has been introduced, and pattern miniaturization and complexity have progressed. Yes. A three-dimensional transistor called Fin-FET (Fin-Field Effect Transistor) has also been devised.

  In the design and manufacture of semiconductor devices, it is necessary to measure the circuit pattern shape and feed back the evaluation results to the design and manufacturing process. For the measurement of the pattern shape, a critical dimension scanning electron microscope (CD-SEM), which is one of scanning charged particle microscopes, is widely used. As a conventional shape evaluation method using an SEM image, (1) a method of measuring dimensions such as a so-called CD value such as a line pattern width and a contact hole diameter, (2) for example, Japanese Patent No. 4158384 (Patent Document 2) There are a method for calculating an image feature amount having a high correlation with the disclosed pattern shape, and (3) a method for detecting a two-dimensional contour line of a pattern disclosed in, for example, Japanese Patent No. 4154374 (Patent Document 3). .

  On the other hand, for pattern observation by SEM, (1) a method of obtaining an SEM image (top-down image) by irradiating an electron beam from vertically above the pattern, and (2) electrons from an oblique direction relative to the pattern. There is a method of obtaining an SEM image (tilt image) by irradiating a beam. Furthermore, as a method for obtaining the latter tilt image, (1) the electron beam disclosed in, for example, Japanese Patent Application Laid-Open No. 2000-348658 (Patent Document 4) is deflected to irradiate the observation object obliquely. Imaging method (referred to as beam tilt method), (2) tilting the stage itself for moving the sample (referred to as stage tilt method), and (3) mechanically tilting the SEM electron optical system itself There is a system (referred to as a lens barrel tilt system).

JP 2007-250528 A Japanese Patent No. 4158384 Japanese Patent No. 4154374 JP 2000-348658 A

As patterns become finer, more complex, and three-dimensionally structured, design and manufacturing process control are becoming increasingly difficult. In addition to pattern height, line width, and sidewall inclination angle, subtle changes such as corner roundness and three-dimensional pattern cross-sectional shape have a great influence on the electrical characteristics of semiconductors. It is difficult to observe directly from the top-down direction. For example, in order to obtain information on the height, side wall inclination angle, and cross-sectional shape of the pattern, a tilt image obtained by observing the pattern from an oblique direction using an SEM was used. Measurement is considered effective.

  However, since the tilt image has various contour lines of a three-dimensional pattern, shape recognition by image processing becomes complicated. In addition, since the tilt image includes an edge due to a shape change outside the measurement target due to the roughness of the pattern surface, it is not easy to measure the shape by image processing together with the complicated shape recognition. In recent years, the number of measurement points to be evaluated has increased, and it is not realistic from the viewpoint of time and labor to manually analyze an image or specify an image processing method for each measurement point. Furthermore, in the first place, it is difficult to determine from which direction it is appropriate to observe the pattern in order to evaluate the shape of the pattern.

The present invention provides a method for performing stable and automatic shape evaluation from a tilt image of a semiconductor pattern including a complicated structure as described above. In order to solve this problem, the present invention provides a pattern shape evaluation apparatus and method using a scanning charged particle microscope having the following features. In the following description, a scanning electron microscope (SEM) will be described as an example of a scanning charged particle microscope. However, the present invention is not limited to this, and a scanning ion microscope (Scanning Ion Microscope: The present invention can also be applied to a scanning charged particle microscope such as a SIM) or a scanning transmission electron microscope (STEM).
That is, in the present invention,
(1) In a circuit pattern shape evaluation apparatus and method therefor, an imaging region designation step for designating an area including a circuit pattern to be evaluated as an imaging region, and an irradiation direction of charged particles irradiated to the imaging region (imaging An imaging direction designating step for designating a direction), an imaging step for imaging the imaging area from the imaging direction to obtain a captured image, and a design data input step for inputting design data of a circuit pattern included in the imaging area; , An expected contour calculation step of calculating an expected position (expected contour line) of the contour line of the circuit pattern observed on the captured image from the design data and the imaging direction, and the captured image and the predicted contour line A matching step for obtaining a correspondence relationship and an image processing range or an image processing method for processing the captured image based on the predicted contour line are set. An image processing range / method setting step to be performed, and a shape evaluation step for evaluating the shape of the circuit pattern by processing the captured image in accordance with the image processing range or the image processing method.

  Thus, it is possible to estimate where the contours of the pattern to be evaluated or its peripheral pattern may exist in the tilt image by using the predicted contour calculated based on the design data. It is possible to set an appropriate image processing range and image processing method for evaluating the pattern. Thus, for example, in detecting the contour line of the pattern to be evaluated, failures such as erroneous detection of a contour line different from the contour line to be detected can be reduced, and stable shape evaluation can be realized.

  (2) Further, in the present invention, the predicted contour calculated in the predicted contour calculation step and the image processing range or image processing method set in the image processing range / method setting step are stored as a measurement recipe. The shape evaluation step is performed based on the measurement recipe.

  Thus, by creating a file for designating an image processing range and an image processing method as a measurement recipe, pattern shape evaluation can be automatically performed based on the measurement recipe. Further, by using the design data, a wafer or SEM device is not required for creating a measurement recipe (offline), and a measurement recipe can be created immediately after tape out. As a result, it is possible to improve overall throughput including preparation for evaluation including recipe generation to actual SEM imaging / shape evaluation. Furthermore, by managing the measurement recipe as a file, the measurement recipe can be used in common at the time of shape evaluation of the same pattern or shape evaluation using a plurality of SEMs.

  (3) In the present invention, in the imaging step, the imaging from the imaging direction is performed by a method of deflecting the irradiation direction of charged particles in the imaging direction, a method of tilting a sample stage on which an evaluation target is placed, or charging The particle microscope electron optical system itself is mechanically tilted, and the predicted contour calculation step according to claim 1 calculates the predicted contour according to the imaging method.

  In order to observe a pattern from an oblique direction, it is necessary to incline the irradiation direction of charged particles relative to the pattern, and there can be a plurality of methods as described above. In addition, the position of the pattern outline on the tilt image generated depending on the method may be different. Therefore, by calculating the expected contour line from the design data in consideration of such a difference in method, it is possible to estimate a contour line that is closer to the contour line on the actually captured SEM.

  (4) In the present invention, in the predicted contour calculation step, the design data in which two-dimensional layout information of the circuit pattern shape included in the imaging region is written and the design value of the height of the circuit pattern are obtained. Respectively input, calculating a pseudo three-dimensional shape of the circuit pattern from the design data and the design value of the height, and calculating a contour line of the pseudo three-dimensional shape observed from the imaging direction. Was used to calculate the expected contour line.

  As the design data format, “GDSII”, “OASIS”, etc. are generally known, but only two-dimensional layout information (without depth information) is recorded in the design data file. Many. For this reason, by providing a mechanism for separately inputting the height information (for example, film thickness information for each layer) of each pattern, the 3D shape of the target pattern can be estimated internally together with the two-dimensional layout information. Can do. Of course, the three-dimensional shape may be obtained by inputting design data in which the three-dimensional shape information of the pattern is written from the beginning. However, the cross-sectional shape of the actually generated pattern is not just a rectangle, and the line width may change continuously in the height direction (cannot be represented by a single two-dimensional layout information and height information) .

  In such a case, when the evaluation target is a resist pattern, exposure and development simulation may be performed from the mask design data to estimate the three-dimensional shape information. In addition, in order to estimate an accurate contour position on the tilt image of the three-dimensional shape, an SEM simulation may be used in combination with the three-dimensional shape. Similarly, when the evaluation target is an etching pattern, three-dimensional shape information can be estimated by performing an etching simulation after estimating the shape of the resist pattern.

  (5) Also, in the present invention, in the image processing range / method setting step, a predicted misalignment range between the predicted contour and the contour actually observed on the captured image is set, and the predicted contour The image processing range or the image processing method is set based on the expected misalignment range.

  A shape divergence may occur between the expected contour line and the contour line actually observed on the captured image. Causes include pattern thickening / thinning, rounded corners, roughness, pattern translation, pattern displacement on the SEM image due to secondary electron generation mechanism, etc. . In order to avoid failure of pattern shape evaluation due to the shape divergence, an expected range of misalignment is set in advance, and an image processing range or image processing method is set so that processing does not fail even if misalignment occurs in the same range. By doing so, shape evaluation robust to shape divergence is realized.

  The following (a) to (c) can be cited as the estimation method of the position shift expected range. (A) A method of setting based on prior knowledge about the manufacturing process. For example, the corners of the pattern are likely to be rounded, knowledge that a positional shift may occur between two exposures during double patterning, and methods such as referencing past cases are included. (B) A method of estimating the contour line actually observed on the captured image based on exposure, development, etching, SEM simulation or the like as described in the item (4). At this time, since it may be difficult to accurately provide simulation parameters that meet actual conditions, it is possible to estimate the expected misalignment range from a plurality of estimation results estimated by varying the parameters. (C) A method in which the pseudo three-dimensional shape described in the item (1) or the predicted contour is deformed by simple shape deformation or image processing simulating the simulation, and estimation is performed based on the deformation range.

(6) In the present invention, in the shape evaluation step, according to the image processing range or the image processing method, (a) the dimension of the circuit pattern is measured, (b) the outline of the circuit pattern is detected, (c ) Image feature values correlated with the three-dimensional shape of the circuit pattern are calculated.
(7) Furthermore, in the circuit pattern shape evaluation apparatus and method, an imaging area designating step for designating an area including the circuit pattern to be evaluated as an imaging area, and an irradiation direction of the charged particles that irradiate the imaging area An imaging direction designating step for designating (imaging direction), a design data input step for inputting design data of a circuit pattern included in the imaging area, and an expected position of the outline of the circuit pattern observed on the captured image Based on the predicted contour line, the predicted contour calculation step for calculating (expected contour line) from the design data and the imaging direction, the predicted contour display step for displaying the predicted contour line on the screen, and the displayed predicted contour line An imaging direction determining step for determining an imaging direction, and an imaging region captured from the imaging direction determined in the imaging direction determining step. An imaging step for obtaining an image was included.

  That is, in order to perform shape evaluation using a tilt image, it is necessary to set an imaging direction in which an appropriate shape evaluation of a pattern to be evaluated can be performed. Depending on the imaging direction, the pattern (or part of the pattern) to be evaluated in the captured image may be behind the surrounding pattern. Further, for example, in contour detection, it is not necessary to observe only the contour to be detected. In order to detect a contour line satisfactorily by image processing, it is desirable that the periphery of the contour line to be detected is imaged to some extent in consideration of the spread (base) of the edge signal in the SEM image. In order to set the imaging direction based on such a judgment, a GUI that prompts the user to determine the imaging direction by displaying the expected contour of the pattern observed from the arbitrarily set imaging direction on a display etc. is effective. is there. In addition, by designating the pattern to be evaluated, it is possible to determine whether or not the pattern to be evaluated can be observed in the computer based on the predicted contour, and to automatically determine an appropriate imaging direction.

According to the present invention, it is possible to perform shape evaluation stably and automatically from a tilt image of a semiconductor pattern including a complicated structure. This makes it possible to obtain information on the height of the pattern, the side wall inclination angle, and the cross-sectional shape of the three-dimensional pattern, which are difficult to observe directly from the top-down direction. It is possible to provide feedback to the design and manufacturing process. In addition, by automating such shape evaluation, it is possible to greatly reduce the work time and labor of the evaluator even for a large number of measurement points.

It is a figure which shows the structure of the SEM apparatus for implement | achieving this invention. It is a figure which shows the method of imaging the signal amount of the electron discharge | released from on a semiconductor wafer. It is a figure which shows the flow of the whole process in this invention. It is a figure which shows the flow of the whole process in this invention. It is a top view of the pattern which shows the two-dimensional layout information of the pattern which can be put on an upper layer. It is a top view of the pattern which shows the two-dimensional layout information of the pattern which can be placed in a lower layer. It is a perspective view to the three-dimensional shape estimated from the two-dimensional layout information of the pattern of an upper layer and a lower layer. It is a figure which shows the variation of the shape deformation / shift of a pattern. It is a figure which shows the top-down image of the line pattern 702. It is a graph which shows the SEM signal profile between ab in the top-down image 701. FIG. It is a graph showing the integration profile of a SEM signal. It is a graph which shows the example which measures line width from the integration profile of a SEM signal. The figure which shows the cross-sectional shape of the line pattern which shows the state by which the electron beam was irradiated from diagonally. It is a tilt image of a line pattern observed by irradiating an electron beam from an oblique direction. It is a perspective view of the pattern which shows the method of extracting the outline of line segment ab which is a part of normal linear pattern using a tilt image, and shows the state by which the length measurement cursor was set to the appropriate range. The method of extracting the outline of line segment ab which is a part of the linear pattern formed thinner than usual using a tilt image, and the perspective of the pattern which shows the state where the length measurement cursor was set to the appropriate range FIG. The pattern which shows the method of extracting the outline of line segment ab which is a part of the linear pattern formed thinner than usual using a tilt image, and shows the state where the length measurement cursor was set in the large inappropriate range FIG. The pattern which shows the method of extracting the outline of line segment ab which is a part of the linear pattern formed more thinly than usual using a tilt image, and shows the state where the length measurement cursor was set to the small and inappropriate range FIG. It is a figure which shows the extraction method of the outline of a curve pattern using a tilt image. It is a figure which shows the curve extracted without being influenced by the other outline by setting the length measurement cursor 1003 curved along the curve 1002 in order to detect the curve ab. It is a perspective view of the pattern which shows the result of having extracted the outline of the curve pattern using the tilt image. It is a figure which shows the cross-sectional shape of the line pattern 1101 which has a small notch in a footing part. It is a figure which shows the cross-sectional shape of the line pattern 1103 which has a big notch in a footing part. It is a figure which shows the tilt image at the time of irradiating an electron beam from diagonal and observing the line pattern 1101 or 1103. It is a figure which shows the outline of the SEM signal profile between a-b equivalent to a footing part vicinity in a tilt image. It is a perspective view of an ideal rectangular parallelepiped line pattern. It is a perspective view of a line pattern in which fine irregularities called roughness are present on the surface. It is a figure which shows the tilt image of the line pattern in which the fine unevenness | corrugation called roughness is present on the surface. It is a graph which shows the result of having calculated the variation of the image brightness value in the fixed area | region in a tilt image as an image feature-value. It is a perspective view of two line patterns. When the right footing part (line segment ab) of the pattern 1301 is a site to be evaluated, (a) a pattern cross-sectional view showing a state in which an electron beam is irradiated from almost right above the pattern; and (b) It is a pseudo tilt image at that time. When the right footing part (line segment ab) of the pattern 1301 is a part to be evaluated, (a) the incident angle of the electron beam is large and the part to be evaluated is a shadow of the previous pattern. A cross-sectional view of the pattern and (b) a pseudo tilt image at that time. It is a figure which shows the example of a pseudo tilt image or the display example of an observation direction, and the determination method of an imaging direction. When the right footing portion (line segment ab) of the pattern 1301 is a part to be evaluated, (a) a pattern indicating a state in which the electron beam is incident on a part to be evaluated with an appropriate incident angle of the electron beam And (b) a pseudo tilt image at that time. (a) is a figure which shows the cross-sectional shape of a pseudo | simulation 3D line pattern, (b) is a figure which shows the pseudo tilt image obtained when the cross-sectional shape 1401 of the estimated line pattern is imaged from the imaging direction 1402. FIG. (A) is a figure which shows the cross-sectional shape of the pattern 1404 in which side wall inclination | tilt angle (theta) differs greatly with respect to a pseudo | simulation 3D line pattern, (b) is a figure which shows the tilt image of the pattern 1404. It is a figure which shows the extraction method of the outline of a linear pattern. It is a figure which shows the extraction method of the outline of a curve pattern. It is a figure which shows the structure of the apparatus system for implement | achieving this invention, and shows the state which has arrange | positioned SEM control apparatus, an imaging / measurement recipe creation calculating apparatus, and a server separately. It is a figure which shows the structure of the apparatus system for implement | achieving this invention, and shows the state which united and arrange | positioned the SEM control apparatus, the imaging / measurement recipe creation calculating apparatus, and the server. It is a perspective view of the pattern of cross-sectional shape 1701. FIG. It is a tilt image obtained when a pattern having a cross-sectional shape 1701 is imaged by a stage tilt method or a lens barrel tilt method at a tilt angle θ. It is a tilt image obtained when a pattern having a cross-sectional shape 1701 is imaged by the beam tilt method at a tilt angle θ. It is sectional drawing of the pattern which shows the case where a cross-sectional shape is a mere rectangle. The line width is a cross section of a pattern having a cross-sectional shape that continuously changes in the height direction. It is a figure which shows the processing flow which estimates the three-dimensional shape or SEM image of a pattern. It is sectional drawing of the line pattern which shows the state which irradiates the electron beam from two directions, 2002 and 2003. It is a figure which shows the tilt image of the pattern 2001 when an electron beam is irradiated from the direction of 2002. FIG. It is a figure which shows the tilt image of the pattern 2001 when an electron beam is irradiated from the direction of 2003. FIG.

  The present invention provides a method for performing stable and automatic shape evaluation from a tilt image of a semiconductor pattern including a complicated structure using a scanning charged particle microscope. Hereinafter, a case where the embodiment according to the present invention is applied to a scanning electron microscope (SEM) will be described. However, the present invention is not limited to this, and a scanning ion microscope (Scanning Ion Microscope) is described. : SIM) or a scanning charged particle microscope such as a scanning transmission electron microscope (STEM).

1. SEM
1.1 SEM components
FIG. 1 shows a block diagram of a schematic configuration of an SEM that acquires a secondary electron image (Secondary Electron: SE image) or a reflected electron image (Backscattered Electron: BSE image) of a sample. Further, the SE image and the BSE image are collectively referred to as an SEM image. The image acquired here is a top-down image obtained by irradiating the measurement object with an electron beam from the vertical direction, or a part of a tilt image obtained by irradiating the electron beam from an arbitrary tilted direction. Or include everything.

  The electron optical system 102 includes an electron gun 103 therein and generates an electron beam 104. After the electron beam emitted from the electron gun 103 is narrowed down by the condenser lens 105, the electron beam is focused and irradiated at an arbitrary position on the semiconductor wafer 101 which is a sample placed on the stage 121. Further, the irradiation position and aperture of the electron beam are controlled by the deflector 106 and the objective lens 108.

  Secondary electrons and reflected electrons are emitted from the semiconductor wafer 101 irradiated with the electron beam, and the secondary electrons separated from the orbit of the irradiated electron beam by the ExB deflector 107 are detected by the secondary electron detector 109. . On the other hand, the reflected electrons are detected by the reflected electron detectors 110 and 111. The backscattered electron detectors 110 and 111 are installed in different directions. The secondary electrons and backscattered electrons detected by the secondary electron detector 109 and the backscattered electron detectors 110 and 111 are converted into digital signals by the A / D converters 112, 113, and 114 and input to the processing / control unit 115. Thus, the image is stored in the image memory 117 and the CPU 116 performs image processing according to the purpose.

  FIG. 2 shows a method of imaging the signal amount of electrons emitted from the semiconductor wafer when the semiconductor wafer is irradiated with an electron beam. For example, as shown on the left side of FIG. 2, the electron beam is scanned and irradiated in the x and y directions as 201 to 203 or 204 to 206. It is possible to change the scanning direction by changing the deflection direction of the electron beam. The locations on the semiconductor wafer irradiated with the electron beams 201 to 203 scanned in the x direction are denoted by G1 to G3, respectively. Similarly, locations on the semiconductor wafer irradiated with electron beams 204 to 206 scanned in the y direction are denoted by G4 to G6, respectively.

  The signal amounts of electrons emitted in G1 to G6 are the brightness values of the pixels H1 to H6 in the image 209 shown on the right in FIG. 2 (subscripts 1 to 6 in G and H correspond to each other). Reference numeral 208 denotes a coordinate system indicating the x and y directions on the image (referred to as an Ix-Iy coordinate system). Thus, the image frame 209 can be obtained by scanning the inside of the visual field with the electron beam. Further, in practice, a high S / N image can be obtained by scanning the inside of the visual field several times with an electron beam in the same manner and averaging the obtained image frames. The number of addition frames can be set arbitrarily. Although FIG. 1 shows an embodiment provided with two reflected electron image detectors, it is possible to eliminate the reflected electron image detector, to reduce the number, or to increase the number.

  As a method for obtaining a tilt image obtained by observing a measurement object from an arbitrary tilt angle direction using the apparatus shown in FIG. 1, (1) deflecting an electron beam irradiated from an electron optical system and tilting the irradiation angle of the electron beam (2) A method of tilting the stage 121 itself for moving a sample such as a semiconductor wafer (referred to as a stage tilt method). In FIG. 2, the stage is tilted at a tilt angle 122 with respect to the xyz coordinate system 100), and (3) a system in which the electron optical system itself is mechanically tilted (referred to as a lens barrel tilt system). .

  A processing / control unit 115 in FIG. 1 is a computer system that includes a CPU 116 and an image memory 117. In order to capture an area including a circuit pattern to be evaluated based on an imaging recipe as an imaging area, a stage controller 119 or a deflection Processing and control such as sending a control signal to the control unit 120 or performing various kinds of image processing based on a measurement recipe for a captured image in an arbitrary imaging region on the semiconductor wafer 101 is performed. The processing / control unit 115 is connected to a processing terminal 118 (including input / output means such as a display, a keyboard, a mouse, etc.), and displays a GUI or the like on the GUI (accepts input from the user). Graphic User Interface).

  Reference numeral 121 denotes an XY stage, which moves the semiconductor wafer 101 and enables imaging of an arbitrary position of the semiconductor wafer. Changing the imaging position with the XY stage 121 is called stage shift, for example, changing the observation position by deflecting an electron beam with the deflector 106 is called image shift. In general, the stage shift has a wide movable range but the positioning accuracy of the imaging position is low. On the other hand, the image shift has a property that the movable range is narrow but the imaging position positioning accuracy is high.

1.2 Imaging / Measurement Recipe
Details of the imaging recipe and the measurement recipe will be described.
First, an imaging recipe is a file that specifies an imaging sequence and imaging conditions for imaging an imaging region to be evaluated without positional displacement and with high definition. For example, there is a risk that the imaging position may be shifted due to the positioning accuracy of the above-described stage shift and image shift. As a countermeasure, as shown in FIG. 3 of Japanese Patent Application Laid-Open No. 2007-250528 (Patent Document 1), a positioning template (addressing point; hereinafter referred to as AP) provided with a position and a pattern is used. Before imaging the imaging area to be evaluated and imaged, the position deviation amount is detected once by imaging the position of the template, and a field of view is displayed in the imaging area to be evaluated so as to correct the positional deviation amount. By moving, it is possible to obtain a captured image with little visual field shift.

  Also, an autofocus point (hereinafter referred to as AF) for imaging to adjust the focus of the electron beam applied to the evaluation object, and an autostigma point (hereinafter referred to as AST) for correcting astigmatism of the electron beam shape. ), Or an auto brightness / contrast point (hereinafter referred to as ABCC) for adjusting the brightness / contrast of the captured image, and imaging the AF, AST, ABCC before imaging the imaging area to be evaluated. Each adjustment can be performed to obtain a high-definition captured image. The imaging region to be evaluated, the location of AP, AF, AST, ABCC, the presence / absence of imaging, and the imaging order are designated as the imaging sequence. Also, the imaging area, AP, AF, AST, and ABCC to be evaluated are designated as imaging conditions such as probe current, acceleration voltage, electron beam scanning direction, scanning range (imaging range), and number of additional frames. To do. An observation direction (tilt angle) can also be specified as an imaging condition.

  Next, the measurement recipe is a file that designates a procedure for measuring the shape of the imaging target and evaluating it from the captured SEM image. A measurement point (hereinafter referred to as MP) to be measured and evaluated in the SEM image is determined, and a desired dimension of the pattern in the MP is measured and evaluated. There may be a plurality of MPs in the SEM image, and the entire field of view of the SEM image may be the MP.

  As a shape evaluation method, (1) a method of measuring a dimension such as a line pattern width or a contact hole diameter called a so-called CD value, and (2) a pattern shape disclosed in, for example, Japanese Patent No. 4158384 (Patent Document 2) Examples include a method for calculating a highly correlated image feature amount, and (3) a method for detecting a two-dimensional contour line of a pattern disclosed in, for example, Japanese Patent No. 4154374 (Patent Document 3).

  For example, the CD value of (1) includes the line pattern line width measurement, the gap measurement between line patterns, the receding amount of the line end, the contact hole diameter measurement, OPC (Optical Proximity Correction). Measurement of the shape and the like can be mentioned, and hereinafter, such variations of measurement in MP are referred to as length measurement species. In addition to the category such as “line width measurement”, information on the measurement part such as which part of the wiring area is measured and where the distance is measured, and in which direction if “retraction amount” is measured, for example. Information on the direction of measurement of whether to measure the amount of retraction of can be included in the length measurement type.

  In order to calculate the dimensions (1) to (3), the image feature amount, and the contour line, image processing for accurately detecting the edge of the pattern shape is required. In order to accurately detect the edge, an area having a certain size including the edge is set, and the SEM signal is integrated within the area, for example, in the line direction, thereby being less susceptible to image noise and line edge roughness. There is a method of calculating a profile and detecting an edge position using the profile. An area of a certain size including the edge is designated by a box called a length measurement cursor. The MP position, length measurement type, position and shape of the length measurement cursor, and length measurement method (length measurement algorithm and length measurement parameter) are managed as a measurement recipe, and the SEM evaluates the SEM image based on the measurement recipe.

  In the present invention, the terms “imaging recipe” and “measurement recipe” are used as described above. However, the separation of the imaging recipe and the measurement recipe is an example, and the setting items specified in each recipe can be managed in any combination. Therefore, when the imaging recipe and the measurement recipe are not particularly distinguished, they are collectively referred to as a recipe or an imaging / measurement recipe.

  The imaging recipe creation device 123 generates an imaging recipe, and the processing / control unit 115 captures an SEM image of an area including a circuit pattern to be evaluated based on the imaging recipe. The measurement recipe creating device 124 generates a measurement recipe by a method described later, and the image processing device 125 performs shape measurement and evaluation of the imaging target from the SEM image based on the measurement recipe. Reference numerals 123, 124, and 125 are connected to a processing terminal 126 (including input / output means such as a display, a keyboard, and a mouse), and display a processing result or the like to a user, or accept a user input. User Interface). Reference numeral 127 denotes a storage for storing a database such as semiconductor circuit pattern design layout information (hereinafter referred to as design data). The database stores captured SEM images, measurement / evaluation results (pattern dimensions, image features, contours, etc.). Line), and information such as imaging / measurement recipes can be stored and shared. The processes performed in 115, 123, 124, and 125 can be divided into a plurality of devices or combined and processed in any combination.

2. Processing flow of the present invention
The present invention provides a method for performing stable and automatic shape evaluation from a tilt image of a semiconductor pattern including a complicated structure as described above. Hereinafter, the processing flow of the present invention will be described. The outline of the processing flow will be described with reference to FIGS. 3 and 4, and the details will be supplementarily described in other drawings as necessary.

2.1 Design Data Input: Corresponding to Step 1) in FIG. 4 First, an area including a circuit pattern to be evaluated is designated as an imaging area (Step 301). As the coordinates of the circuit pattern to be evaluated, for example, the coordinates of the hot spot (danger point) detected based on the result of an exposure simulation or the like executed by an EDA (Electronic Design Automation) tool is input. Alternatively, there is a case where the user inputs his / her own judgment (referring to the information of the EDA tool as necessary). In addition, an imaging condition in the imaging area is input (step 302). The imaging conditions include probe current, acceleration voltage, electron beam scanning direction and scanning range (imaging range), the number of added frames, and the irradiation direction of charged particles that irradiate the imaging region (as described above). Imaging direction).

  By changing the observation direction (tilt angle), a tilt image obtained by observing the pattern from an oblique direction can be obtained. A method for determining the tilt angle will be described later. Further, design data of a circuit pattern included in the imaging area is input (step 303). The design data is, for example, two-dimensional layouts 401 and 402 having a circuit pattern shape in FIG. In the figure, a pattern is drawn in an xyz coordinate system (corresponding to 100 in FIG. 1).

2.2 Creation of pseudo 3D pattern; corresponding to step 2) of FIG. 4 Based on the design data input at step 303, a pseudo three-dimensional shape (pseudo 3D pattern) of the circuit pattern shape included in the imaging region is calculated in the computer. Calculate (step 305).

  As the design data format, “GDSII”, “OASIS”, etc. are generally known, but only two-dimensional layout information (without depth information) is recorded in the design data file. Many. Therefore, by providing a mechanism capable of separately inputting height information (for example, film thickness information for each layer) of each pattern, the three-dimensional shape of the target pattern (for example, FIG. 4) is combined with the two-dimensional layout information. 403, 404) can be estimated. The height information can be one of input information in step 304. An example of estimating a three-dimensional shape of a pattern composed of multiple layers will be described with reference to FIGS. 501 in FIG. 5A is two-dimensional layout information of a pattern that can be placed on the upper layer, and 502 in FIG. 5B is two-dimensional layout information of a pattern in the lower layer. By giving the height h1 of the upper layer and the height h2 of the lower layer to this, a three-dimensional shape such as 503 in FIG. 5C can be estimated. This can be similarly estimated for three or more layers and laminated patterns (such as hard masks).

  Of course, the three-dimensional shape may be obtained by inputting design data in which the three-dimensional shape information of the pattern is written from the beginning. However, the cross-sectional shape of the actually generated pattern is not a simple rectangle as illustrated in 1801 of FIG. 18A, and the line width may continuously change in the height direction as illustrated in 1804 of FIG. 18B. (A single line width (eg, 1802 in FIG. 18A) and height information (eg, 1803 in FIG. 18A) cannot be expressed). In such a case, a modification may be made such as rounding the corners of the pattern in consideration of the inclination angle (taper angle) of the side wall, the top rounding or the bottom footing of the pattern. Further, in order to estimate the three-dimensional shape of the pattern more accurately, as shown in FIG. 19, when the evaluation object is a resist pattern, mask design data is input (step 1901), and the resist film is determined from the mask design data. The three-dimensional shape information may be estimated by performing the exposure and development simulation (steps 1902 and 1903). Similarly, when the evaluation target is an etching pattern, three-dimensional shape information can be estimated by performing an etching simulation after the above-described resist pattern shape estimation (step 1904).

2.3 Expected contour estimation on tilt image; corresponding to step 3) in FIG. 4, based on the pseudo 3D pattern estimated in step 305 and the imaging direction of the SEM which is one of the input information in step 302, the pseudo By calculating a contour line obtained when the 3D pattern is observed from the imaging direction, an expected position (predicted contour line) of the pattern contour line actually observed on the tilt image from the imaging direction is estimated ( Step 306). In other words, it is possible to estimate where in the tilt image each contour line of the pattern to be evaluated or its peripheral pattern can exist based on the predicted contour line calculated based on the design data. It is possible to set an appropriate image processing range and image processing method (position and shape of the length measurement cursor, length measurement method, etc. described in the description of the measurement recipe) in evaluating the pattern.

  Thereby, for example, in detecting the contour line of the pattern to be evaluated, failures such as erroneous detection of a contour line different from the contour line to be detected can be reduced, and stable shape evaluation can be realized. In FIG. 4, 405 is an example of an expected contour line (405 can be called an estimated tilt image). This figure shows where the expected contour line is located in the Ix-Iy coordinate system (corresponding to 208 in FIG. 2), which is a coordinate system indicating the x and y directions on the tilt image.

  In order to observe the pattern from an oblique direction, it is necessary to tilt the irradiation direction of the charged particles relative to the pattern. As described above, the beam tilt method, stage tilt method, lens barrel tilt method, etc. There can be multiple ways. However, the position of the pattern contour on the tilt image generated due to such a difference in tilt method may be different. The difference in the position of the pattern outline on the tilt image due to the difference in the tilt method will be described with reference to FIGS. A tilt image obtained when the pattern of the cross-sectional shape 1701 as shown in FIG. 17A is imaged by the stage tilt method or the lens barrel tilt method at a tilt angle θ (illustrated) is a beam at 1709 in FIG. A tilt image obtained when the image is taken by the tilt method is indicated by 1710 in FIG. 17C. In the tilt images 1709 and 1710, the edge positions on the tilt images corresponding to the corners a, b and c of the pattern in the cross-sectional shape 1701 are indicated by a, b and c.

  In imaging the cross-sectional shape 1701 by the stage tilt method, the xyz coordinate system is inclined with respect to the pattern as 1707. The trajectories of the electron beams irradiated to the corners a, b, and c are schematically shown as 1702, 1703, and 1704, and the electron beams are scanned from 1702 to 1704 via 1702 and 1703. Since the scanning speed of the electron beam with respect to the x direction of the xyz coordinate system 1707 (equivalent to a direction 1705 orthogonal to the electron beam irradiation direction) is constant, the x direction between 1702-1703 and 1703-1704. Are respectively L1 and L2 (illustrated), the distances between edges ab and bc on the tilt image 1709 are constant times A * L1 and A * L2 of L1 and L2, respectively (A Is a constant).

  In the imaging of the cross-sectional shape 1701 by the lens barrel tilt method, the xyz coordinate system is not inclined with respect to the pattern as in 1708, but in the direction 1705 orthogonal to the electron beam irradiation direction as in the stage tilt method. Since the scanning speed of the electron beam is constant, a tilt image 1709 is obtained.

  On the other hand, in imaging of the cross-sectional shape 1701 by the beam tilt method, the xyz coordinate system is 1708. Since the scanning speed of the electron beam with respect to the x direction (equivalent to the direction 1706 in the figure) of the xyz coordinate system 1708 is constant, the intervals in the x direction between 1702-1703 and 1703-1704 are respectively M1. , M2 (illustrated), the distances between edges ab and bc on the tilt image 1710 are constant times A * M1 and A * M2 of M1 and M2, respectively (A is a constant).

  Thus, even with the same tilt angle, the obtained tilt images differ as shown by 1709 and 1710 depending on the tilt method. Therefore, the present invention is characterized in that a contour line closer to the contour line on the actually captured SEM is estimated by calculating the predicted contour line in consideration of such a difference in tilt method.

  It should be noted that the beam tilt method in which the coordinate system is not inclined such as 1708 with respect to the pattern to be evaluated is easy to illustrate, and in this specification, the drawings other than FIGS. ing.

2.4 Estimating the expected range of misalignment between the predicted contour line and the actual contour line; corresponding to step 4 in FIG. 4; the predicted contour estimated in step 306 and the contour line actually observed on the captured image The estimated misalignment range is estimated (step 307). A shape divergence may occur between the expected contour line and the contour line actually observed on the captured image. As the cause, as illustrated in FIG. 6, (b) Thickness 602 / (c) Thinness 603, (d) Rounded corner 604, (e) Roughness 605, ( f) Pattern displacement on the SEM image due to pattern translation 607 and secondary electron generation mechanism (the actual pattern edge position does not always match the pattern edge position on the image) Etc. In order to avoid failure of pattern shape evaluation due to the shape divergence, an expected range of misalignment is set in advance, and an image processing range or image processing method is set so that processing does not fail even if misalignment occurs in the same range. By doing so, shape evaluation robust to shape divergence is realized. The following (a) to (c) may be mentioned as the estimation method of the position shift expected range.

  (A) A method of setting based on prior knowledge about the manufacturing process. For example, the corners of the pattern are likely to be rounded, knowledge that a positional shift may occur between two exposures during double patterning, and methods such as referencing past cases are included. Also, the extent of the range is set based on the worst amount of misalignment that can occur, or it is set based on the amount of misalignment that is expected to be statistically less than the actual amount of misalignment that occurs to some extent. can do.

  (B) As described above with reference to FIG. 19, the pattern shape actually generated is accurately estimated based on exposure, development or etching simulation, and is observed on the captured image based on the pattern shape. A method for estimating a contour line. At this time, since it may be difficult to accurately provide simulation parameters that meet actual conditions, it is possible to estimate the expected misalignment range from a plurality of estimation results estimated by varying the parameters. Further, in order to accurately estimate the positional deviation of the pattern on the SEM image caused by the secondary electron generation mechanism, an SEM simulation may be combined. The SEM simulation is to estimate an obtained SEM image by probabilistically obtaining how the electrons move by irradiation with an electron beam by Monte Carlo simulation or the like. For example, when the evaluation target is a resist pattern, SEM simulation is performed based on the pattern shape estimated after the development simulation (step 1905), and the obtained SEM image is estimated. Similarly, when the evaluation target is an etching pattern, an SEM simulation is performed based on the pattern shape estimated after the etching simulation (step 1906), and the obtained SEM image is estimated. If it is difficult to accurately give the parameters of the SEM simulation, it is possible to estimate the expected misalignment range from a plurality of estimation results obtained by estimating the parameters.

(C) A method in which a pseudo 3D pattern or an expected contour is deformed by simple shape deformation or image processing simulating the exposure simulation, development simulation, etching simulation or SEM simulation, and estimation is performed based on the deformation range.
The amount of shape deformation that can occur and the reference value of the predicted misalignment range can be one of the input information in step 304. In step 307, the predicted misalignment range may be estimated based on the input information.

  4 is an enlarged view of a portion 408 (three line segments ab, ac, ad) (shown by bold lines) of the predicted contour line in the estimated tilt image 405 of FIG. It shows with. In this example, it is shown that the contour line is expected to move within the range of the dotted line. The expected misalignment range may vary from place to place rather than a fixed amount. In this example, it is estimated that the expected range of deformation or shift such as rounding of the corner portion (for example, 413) is larger than the expected range of deformation or shift of the contour line in the straight line portion (for example, 412). As another example, an enlarged view of a portion 409 (line segment ef, illustrated by a thick line) and 410 (line segment g-h, illustrated by a thick line) of an expected contour line in the estimated tilt image 405, 409B and 410B. The estimated misalignment ranges are indicated by dotted lines 414 and 415, respectively.

2.5 Setting of image processing range on predicted contour line; corresponding to step 5) in FIG. 4. Based on the predicted contour line and the predicted misalignment range, the image processing range (also called length measurement cursor) or image processing method is used. Set (step 308). As an example of a length measurement cursor in the estimated tilt image 416 of FIG. 4, an image processing range 417 for detecting the footing edge of the pattern 406, and an image processing for evaluating the degree of roundness of the corner portion where the patterns 406 and 407 intersect. An image processing range 419 for detecting a footing edge of a range 418 and a pattern 407 is shown. In the estimated tilt image 416, as an example of setting the image processing range, the image processing range is set so that the image processing range covers up to the predicted misregistration range 411 indicated by a dotted line. With this setting, if the deformation or shift of the evaluation part is within the expected position shift range, the evaluation part does not fall outside the image processing range.

  An image processing range and an image processing method will be described with reference to FIGS. 7A to 7D taking line width measurement of a line pattern in a top-down image as an example. In order to measure the line width 707 of the line pattern 702 on the top-down image 701 in FIG. 7A, it is necessary to accurately and stably measure the positions of the left and right edges of the line. Therefore, an area of a certain size (length measurement cursor area, for example, 703A and 703B) including the edge is set on each of the left and right edges, and the edge position is detected by processing the SEM signal in the area. The length measurement cursor depends on its position, the distance from the line edge to the outer cursor end (704A or 704B), the distance from the line edge to the inner cursor end (705A or 705B), and the range in the y direction (706A or 706B). Can be defined. A graph 708 in FIG. 7B shows an SEM signal profile 709 between a and b in the top-down image 701. In order to increase the S / N of the SEM signal profile 709, the SEM signal in the graph 710 of FIG. 7C is obtained by, for example, averaging the SEM signals in the y direction for the range (706A or 706B) in the y direction of the measuring cursor. Can be obtained.

  By integrating in the line direction, a profile that is less susceptible to image noise and line edge roughness can be obtained. The left and right white band peak positions 712A and 712B are detected in the profile 711, and the interval between them is measured as the line width. A graph 710 in FIG. 7C is an example of an image processing method for line width between peaks of the profile 711, but there may be variations in which place and which place in the profile are measured as the line width. For example, the line edge position on the left side is represented by the brightness value H1 (shown) of the white band peak of the SEM signal profile in the x-direction range (707A, 708A) of the length measurement cursor in the profile 711 in the graph 713 of FIG. A coordinate 714A, the lowest brightness value H2 (illustrated) and its x-coordinate 715A are obtained, and a position 716A between 714A-715A where the brightness value is Hx = H2 + (H1-H2) * P (P is a settable parameter) To do. Similarly, an image processing method is also conceivable in which the right-side line edge position 716B is obtained and the interval between them is measured as a line width. The image processing method described with reference to FIGS. 7A to 7D is an example, and various algorithms can be considered for pattern shape extraction and measurement. In addition, an appropriate image processing range setting method may differ depending on the difference in image processing method.

Hereinafter, a method for setting the length measurement cursor in the tilt image will be described with reference to FIGS.
8A and 8B are measurement examples of the side wall width of the line pattern in the tilt image. In FIG. 8A, reference numeral 801 indicates the cross-sectional shape of the line pattern, and reference numeral 802 indicates the direction of the electron beam irradiated obliquely onto the object. A tilt image when the line pattern 801 is observed from the direction is indicated by 803 in FIG. 8B. A region 804 is an upper surface portion of the line pattern, and a region 805 is a region corresponding to the right side wall portion of the line pattern. In order to measure the width 807 of the region 805 corresponding to the right side wall portion in the tilt image 803, by setting the length measuring cursors 806A and 806B, the boundary edge between the upper surface portion and the right side wall portion, and the right side wall portion, respectively. And the boundary edge of the background.

An appropriate range of the length measurement cursor in a tilt image including many edges will be described with reference to FIGS. Consider a case in which the outline of a line segment ab that is a part of the pattern 901 on the tilt image in FIG. 9A is extracted by image processing and the shape is evaluated. The pattern 903 on the tilt image in FIG. 9B, the pattern 904 on the tilt image in FIG. 9C, and the pattern 906 on the tilt image in FIG. 9D have the same design data as the pattern 901 in FIG. Etc. The line width is narrowed. Therefore, the line width h2 (on the image of the normal pattern 901 in FIG. 9A (illustrated) on the image 903 in FIG. 9B, 904 in FIG. 9C, and 906 in FIG. In the figure, h2 <h1. The length measurement cursor 902 in FIG. 9B is illustrated as an appropriate range, and the length measurement cursor 905 in FIG. 9C and the length measurement cursor 907 in FIG. 9D are illustrated as an inappropriate range. First, the length measurement cursor 902 in FIG. 9B sufficiently includes the line segment ab and the periphery thereof in the normal pattern 901 in FIG. 9A and the deformation pattern 903 in FIG. 9B, and does not include other line segments.
On the other hand, a large length measurement cursor is set like the length measurement cursor 905 in FIG. 9C so that the line segment ab does not fall outside the range of the length measurement cursor even when the pattern is deformed or shifted due to the parameter variation or the like. Then, other line segments (for example, the line segment cd and the line segment ef) are included, and it becomes difficult to detect the line segment ab. On the other hand, if a small length measuring cursor is set like the length measuring cursor 907 in FIG. 9D, it may be difficult to detect the line segment ab by image processing. This is because, as described with reference to FIGS. 7A to 7D, the SEM signal profile at the edge position is not a waveform like an impulse signal in which the brightness value is increased only at the edge position, but is a waveform having a gentle spread. Therefore, the image processing range for detecting the edge needs to have a certain width around the edge position.

  Further, with the smaller length measurement cursor 907 in FIG. 9D, there is a high risk that the line segment ab is outside the range of the length measurement cursor due to deformation or shift of the pattern. However, depending on the pattern shape and viewing direction, it may be difficult to geometrically set a wide measuring cursor that does not include other line segments. In such a case, not only the length measurement cursor but also the image processing method can be changed. For example, when the width of the length measurement cursor centered on the edge position to be detected is narrow, the measurement is not performed using the background lightness value H2 away from the edge position exemplified by the graph 713 in FIG. 7D. The method of measuring only the edge peak position (for example, 712A) exemplified in the graph 710 of FIG. In addition, when a line segment outside the target is included in the length measurement cursor, an image recognition process that excludes the line segment outside the target is added in the image processing.

  A length measuring cursor for detecting curved edges will be described with reference to FIGS. In order to detect a curve 1002 (curve ab) that is a part of the curved pattern 1001 in the tilt image 1000 of FIG. 10A, another contour is set by setting a measurement cursor 1003 that is curved along the curve 1002. A curve can be extracted without being affected by a line. The extraction result is shown at 1005 in FIG. 10B. Further, when integrating the SEM signal profile in such contour extraction of a curved shape, the integration is not performed in the y direction as in the integration profile 711 in FIGS. 7C and 7D, but is integrated along the curve. An image processing method such as Further, the extraction of the contour line is not limited to a part of the pattern typified by 1005 in FIG. 10B, and a contour line or the entire contour line of the field of view can be extracted as in 1007 in FIG. 10C.

  SEM image 701 in FIG. 7A, SEM image 803 in FIG. 8B, SEM image 901 in FIG. 9A, SEM image 903 in FIG. 9B, SEM image 904 in FIG. 9C, SEM image 906 in FIG. 9D, and SEM image 1000 in FIG. A contour similar to the contour of the displayed pattern is given by the predicted contour. As described above, the present invention is characterized in that the image processing range is set in consideration of the mixing of other line segments into the image processing range, the deformation or shift of the pattern, or a certain width necessary for the image processing. For this purpose, an image processing range or an image processing method is set based on information on an expected contour line (for example, 405) on a tilt image and information on an estimated position shift range (for example, 411).

2.6 Measurement recipe generation
The predicted contour calculated in the predicted contour estimation step 306 and the image processing range or image processing method set in the image processing range / method setting step 307 are stored as a measurement recipe (step 309). In this way, by creating a file for designating an image processing range and an image processing method as a measurement recipe, it is possible to automatically perform pattern shape evaluation on a tilt image based on the measurement recipe. In addition, as described above, the design data is used to determine the image processing range or image processing method, so that no wafer or SEM device is required to create a measurement recipe (offline), and a measurement recipe is created immediately after tape-out. it can. As a result, it is possible to improve overall throughput including preparation for evaluation including recipe generation to actual SEM imaging / shape evaluation. Furthermore, by managing the measurement recipe as a file, the measurement recipe can be used in common at the time of shape evaluation of the same pattern or shape evaluation using a plurality of SEMs.

2.7 Tilt image capture of evaluation points; corresponding to step 6) in FIG.
The imaging region input in step 301 is imaged from the imaging direction input in step 302 to obtain a captured image (step 310). FIG. 4 shows a captured tilt image 420.

2.8 Matching of tilt image and expected contour line & image processing range setting on tilt image; corresponding to step 7) of FIG. 4 The captured image captured at step 310 and the predicted contour line estimated at step 306 are matched, A correspondence relationship between positions is obtained (step 311). When the correspondence between the position of the captured image and the predicted contour line is known, since the image processing range and the image processing method are set for the predicted contour in step 308, the position of the captured image and the image processing range is set. The image processing range and the image processing method can be set on the captured image (step 312). The length measurement cursors 417, 418, and 419 arranged in the actually captured tilt image 421 in FIG. 4 are shown.

  For example, if the length measurement cursor is not well-positioned, for example, the expected misalignment range is too wide and the length measurement cursor cannot include the expected misalignment range, or a large misalignment greater than the expected misalignment range occurs. In some cases. In such a case, the length measurement cursor can be moved based on the actually captured image. The relative positions of the patterns 424 and 425 in the captured image 422 are greatly deviated from the relative positions of the patterns 424 and 425 in the captured image 421 in FIG. Therefore, the actual footing edge is located at the end of the image processing range 419 for detecting the footing edge of the pattern 425. Therefore, an example in which the length measurement cursor is moved based on the actually captured image is illustrated in the captured image 423. The length measuring cursor 426 in the picked-up image 423 corresponding to the length measuring cursor 419 in the picked-up image 422 moves upward, and as a result, is captured at the center of the footing edge length measuring cursor of the pattern 425.

  FIGS. 15A and 15B show two examples of contour detection. The left figure shows an example of detecting a linear pattern, and the right figure shows an example of detecting a curved pattern. By matching the predicted contour 1501 in FIG. 15A and 1506 in FIG. 15B with the captured image, the length measurement cursor 1503 in FIG. 15A and the length measurement cursor 1508 in FIG. 15B arranged on the predicted contour are displayed on the actual image in FIG. 15B and 1507 in FIG. 15B, and by searching for the contour line in the length measurement cursor (the search direction is schematically indicated by 1504 in FIG. 15A and 1509 in FIG. 15B), FIG. The position 1505 of the outline of 15A and 1510 of FIG. 15B are detected (in FIG. 15A and B, four points on the outline are inspected in each of the left and right sides).

2.9 Evaluation of circuit pattern to be evaluated
In step 312, the shape of the circuit pattern is evaluated by processing the captured image according to the image processing range or the image processing method set on the captured image. Shape evaluation includes variations such as pattern length measurement (step 313), image feature value calculation (step 314), pattern outline extraction (step 315), and pattern 3D shape measurement (step 316).

11A to 11D and FIGS. 12A to 12D, an embodiment of the image feature amount calculation in step 314 of FIG. 3 will be described.
FIGS. 11A to 11D are examples of evaluating the dent (notch shape) of the pattern footing portion using the image feature amount from the tilt image. 11A in FIG. 11A and 1103 in FIG. 11B show the cross-sectional shape of the line pattern. A small notch 1102 is provided in the footing portion of the pattern 1101 in FIG. 11A, and a large notch 1104 is provided in the footing portion of the pattern 1103 in FIG. Exists. Reference numeral 1100 in FIG. 11A indicates the direction of the electron beam irradiated from the oblique direction. A tilt image when the line pattern 1101 or 1103 is observed from the above direction is shown as 1105 in FIG. 11C. An area 1106 in FIG. 11C is an area corresponding to the upper surface portion of the line pattern, and an area 1107 is an area corresponding to the right wall portion of the line pattern.
An outline of the SEM signal profile between a and b corresponding to the vicinity of the footing portion in the tilt image 1105 is shown in 1109 of FIG. 11D. In the profile 1109, the lightness value in the footing portion 1110 is low, but this degree may vary depending on the size of the notch (the larger the notch, the darker, etc.). Therefore, in the evaluation of the notch shape, the brightness value in the footing unit 1110 can be calculated as an image feature amount, and the size of the notch shape can be evaluated according to the value instead of directly measuring the notch shape. As an image processing range for calculating a profile for calculating the image feature quantity correlated with the three-dimensional shape of such a pattern, a length measuring cursor 1108 that appropriately captures the footing portion is arranged as shown in FIG. 11C. .
12A to 12D are examples of evaluating the degree of unevenness on the pattern surface using image feature amounts from tilt images. Taking a line pattern as an example, the pattern of FIG. 12A is not an ideal rectangular parallelepiped as shown by 1201, and there are cases where fine irregularities called roughness are present on the surface as schematically shown by 1204 of FIG. 12B. A tilt image when observing the line pattern 1201 in FIG. 12A or 1204 in FIG. 12B is shown in 1205 in FIG. 12C. A region 1206 is an upper surface portion of the line pattern, and a region 1207 is a region corresponding to the right side wall portion of the line pattern.

  The top surface of the pattern (1202 in the pattern 1201 in FIG. 12A, the region corresponding to 1208 in the tilt image 1205 in FIG. 12C) and the side wall (1203 in the pattern 1201 in FIG. 12A, and 1209 in the tilt image 1205 in FIG. 12C). It is difficult to accurately measure the height of fine irregularities on the pattern surface and evaluate the roughness level from the distribution of the height information. Therefore, paying attention to the property that the brightness value of the image changes according to the unevenness of the pattern surface, within a certain region in the image (1208 for the evaluation of the pattern upper surface, 1209 for the evaluation of the side wall) A variation (for example, standard deviation) in image brightness value is calculated as an image feature amount. As shown in the graph 1210 of FIG. 12D, it can be evaluated that the roughness is larger as the image feature amount (lightness value variation) is larger. A length measurement cursor 1208 or 1209 as shown in FIG. 12C is arranged as an image processing range for designating an area for calculating the image feature amount.

  An example of the 3D shape measurement in step 316 in FIG. 3 will be described with reference to FIGS. In FIG. 20A, 2001 indicates the cross-sectional shape of the line pattern, and 2002 and 2003 indicate two directions of the electron beam irradiated obliquely onto the object. Images of the pattern 2001 are taken from the directions 2002 and 2003, respectively, and two obtained tilt images are shown in 2004 in FIG. 20B and 2009 in FIG. 20C. A region 2005 in FIG. 20B and a region 2010 in FIG. 20C correspond to the upper surface portion of the line pattern, and a region 2006 in FIG. 20B and a region 2011 in FIG. 20C correspond to the right side wall portion of the line pattern.

  The tilt image 2009 corresponding to the observation direction 2003 having a larger tilt angle has a larger area corresponding to the right side wall (the area 2011 is larger than the area 2006). By setting the length measurement cursors 2007 and 2008 in the tilt image 2004, the boundary edge between the upper surface portion and the right side wall portion and the boundary edge between the right side wall portion and the ground can be detected, respectively. Similarly, by setting the length measurement cursors 2012 and 2013 in the tilt image 2009, it is possible to detect the boundary edge between the upper surface portion and the right side wall portion and the boundary edge between the right side wall portion and the base.

If the boundary edge corresponding to the same location in the cross-sectional shape 2001 in FIG. 20A can be detected from the two images in different observation directions 2004 in FIG. 20B and 2009 in FIG. 20C, the corresponding point problem in stereo vision is solved. The height information can be obtained by the principle of surveying (in this case, if the height of the boundary edge between the upper surface portion and the right side wall portion and the boundary edge between the right side wall portion and the base is measured, the accurate height of the pattern 2001 ( Film thickness) can be measured). When the predicted contour line in the present invention is estimated with respect to the tilt image 2004 of FIG. 20B and the 2009 of FIG. 20C, contour lines similar to the contour lines illustrated as the tilt images 2004 and 2009 are obtained. The length measuring cursors 2007 and 2008 shown in FIG. 20B and the length measuring cursors 2012 and 2013 shown in FIG. 20C can be set.
Also, a case will be described in which the height information at an arbitrary location of the pattern 2001 in FIG. 20A is estimated. For example, when the height at the point 2014 on the tilt image 2004 in FIG. 20B is estimated, it is necessary to search for the point on the tilt image 2009 in FIG. 20C corresponding to the point 2014. For example, a position having an image pattern similar to the point 2014 is searched while shifting the template 2015 in the tilt image 2009 of FIG. 20C. At this time, if a corresponding point is found between the tilt images 2004 and 2009, height information at the same point can be measured.

  In the search for the position of the template 2015 in FIG. 20C corresponding to the point 2014 in FIG. 20B, the predicted contour line in the present invention can be used. That is, since the point 2014 existing in the pattern upper surface portion 2005 in the tilt image 2004 is also present in the pattern upper surface portion 2010 in the tilt image 2009, the search for the position of the template 2015 is limited to the range of the pattern upper surface portion 2010. Can do. Thus, even if an image pattern similar to the point 2014 exists at a position outside the range of the pattern upper surface portion 2010, the template 2015 is not erroneously associated. In addition, the search time can be shortened. If the predicted contour line is used, the pattern structure on the tilt image (for example, the pattern upper surface portion and sidewall portion ranges 2005 and 2006 in FIG. 20B and 2010 and 2011 in FIG. 20C) can be known in advance. Setting is possible.

2.10 Expected contour display & imaging direction setting
The predicted contour estimated in step 306 is displayed on the screen (step 317), and the imaging direction can be determined based on the displayed predicted contour (re-input as the imaging direction in step 302). The determined imaging direction can be output to an imaging recipe, and a captured image is obtained in step 310 based on the imaging recipe.
That is, in order to perform shape evaluation using a tilt image, it is necessary to set an imaging direction in which an appropriate shape evaluation of a pattern to be evaluated can be performed. Depending on the imaging direction, the pattern (or part of the pattern) to be evaluated in the captured image may be behind the surrounding pattern.

  Further, for example, in contour detection, it is not necessary to observe only the contour to be detected. In order to detect a contour line satisfactorily by image processing, it is desirable that the periphery of the contour line to be detected be imaged somewhat in consideration of the spread (base) of the edge signal in the SEM image as described above. In order to set the imaging direction based on such a judgment, a GUI that prompts the user to determine the imaging direction by displaying the expected contour of the pattern observed from the arbitrarily set imaging direction on a display etc. is effective. is there. In addition, by designating the pattern to be evaluated, it is possible to determine whether or not the pattern to be evaluated can be observed in the computer based on the predicted contour, and to automatically determine an appropriate imaging direction.

A display example of the predicted contour line will be described with reference to FIGS. Consider an example in which two line patterns 1301 and 1302 are arranged on the base 1303 in FIG. 13A (the base is hatched for easy identification). The part to be evaluated is the right footing part (line segment ab) of the pattern 1301. FIG. 13B shows an expected contour line (pseudo tilt image) in a tilt image obtained when the electron beam is irradiated from the direction 1304 in FIG. 13B (a) to the direction 1306 in FIG. 13D (a) as the imaging direction of the SEM. (B) 1307 to FIG. 13D (b) 1309 are shown in order (the base is hatched for easy identification). In the case of the imaging direction 1304 as shown in FIG. 13B (a), the line segment ab to be evaluated in the pseudo tilt image 1307 (in this case, the top-down image) as shown in FIG. Observation overlaps with the line segment cd on the upper surface.
On the other hand, in the case of the imaging direction 1305 as shown in FIG. 13C (a), the line segment ab to be evaluated in the pseudo tilt image 1308 as shown in FIG. 13C (b) is the shadow of the adjacent line pattern 1302. It is difficult to observe. In the imaging direction 1306 as shown in FIG. 13D (a), the line segment ab to be evaluated in the pseudo tilt image 1309 can be sufficiently observed. By displaying the pseudo tilt image in this way, it is possible to determine and determine the quality of the imaging direction without actually capturing a pattern. In particular, it is often more effective to observe the footing part, which is difficult to observe from the top-down direction as in this example, from an oblique direction with a tilt angle as large as possible, but in high-aspect line and space, etc. , It often becomes the shadow of the surrounding pattern.

  According to this method, it is possible to determine the maximum tilt angle among the observable tilt angles. A pseudo tilt image (1307 in FIG. 13B (b) to 1309 in FIG. 13D (b), etc.) may be displayed on the display screen for determining and determining the quality of the imaging direction. The three-dimensional appearance of the pattern (displayed using the pseudo 3D pattern created in step 305 in FIG. 3. The base 1303 displayed in FIG. 13A and the displays of the patterns 1301 and 1302 thereon) may be displayed. The relationship between the cross-sectional shape of the pattern and the imaging direction (a combination of the imaging direction 1306 and the cross-sectional shapes 1301 and 1302 displayed in FIG. 13D (a)) may be displayed.

  Moreover, the modification of FIG. 13A-D is demonstrated using FIG. 14A and B. FIG. FIG. 14A (a) shows a pseudo tilt image 1403 obtained when the cross-sectional shape 1401 of the line pattern estimated from the pseudo 3D pattern is imaged from the imaging direction 1402. For the purpose of observing the corner a (illustrated) of the footing portion of the pattern 1401, an edge corresponding to the corner a can be observed in the pseudo tilt image 1403 (line segment a-a '). However, a shape divergence may occur between the pseudo 3D pattern and the actual pattern. For example, it is assumed that the actual pattern has a significantly different sidewall inclination angle θ (illustrated) as indicated by 1404 in FIG. 14B (reverse taper) with respect to the pseudo 3D pattern 1401. In that case, the corner b (illustrated) above the pattern 1404 is shaded, and the corner a of the footing portion cannot be observed in 1406 of FIG. 14B (b) which is a tilt image of the actual pattern.

  There is a risk of erroneously determining the quality of the imaging direction due to the shape deviation. Therefore, in step 317, the predicted misalignment value estimated in step 307 (including information on the amount of shape deformation that can occur) can be considered. That is, the amount of shape deformation that can occur can be added to the pseudo 3D pattern or pseudo tilt image, and the result can be displayed. In the example of FIGS. 14A and 14B, a pseudo 3D pattern or a pseudo tilt image to which the shape deformation amount that can be generated with respect to the sidewall inclination angle θ is added is displayed, and the value of the shape deformation amount that can be generated is estimated. If this is difficult, it is possible to display a pseudo 3D pattern or a pseudo tilt image created by changing the shape deformation amount within a certain range.

3. System configuration
An embodiment of the system configuration in the present invention will be described with reference to FIG.
In FIG. 16A, 1601 is a mask pattern design apparatus, 1602 is a mask drawing apparatus, 1603 is an exposure / development apparatus for a mask pattern on a wafer, 1604 is a wafer etching apparatus, 1605 and 1607 are SEM apparatuses, and 1606 and 1608 are the above-mentioned respectively. SEM control device for controlling the SEM device, 1609 is an EDA (Electronic Design Automation) tool server, 1610 is a database server, 1611 is a storage for storing the database, 1612 is an imaging / measurement recipe creation device, 1613 is an imaging / measurement recipe server, Reference numeral 1614 denotes an image processing apparatus image processing server for measuring and evaluating a pattern shape, and these can transmit and receive information via a network 1615.

  A storage 1611 is attached to the database server 1610, (a) design data (mask design data (without / with OPC), wafer transfer pattern design data), (b) imaging / measurement recipe generation rule, (c) generation (D) Captured image, (e) Measurement / evaluation result (pattern measurement value, image feature amount, pattern outline, pattern 3D shape, etc.), (f) pseudo 3D pattern and misregistration An expected range, (g) a part or all of various simulation data (created in FIG. 19) can be stored and referenced by linking with a product type, manufacturing process, date and time, data acquisition device, and the like.

  In the figure, two SEM devices 1605 and 1607 are connected to the network as an example. However, in the present invention, an imaging / measurement recipe is stored in the database server 1611 or the imaging / measurement in any of a plurality of SEM devices. It can be shared by the recipe server 1613, and the plurality of SEM devices can be operated by one imaging / measurement recipe creation. In addition, by sharing the database among multiple SEM devices, the success or failure of past imaging or measurement and the accumulation of failure causes can be accelerated. By referring to this, it is possible to help generate a good imaging / measurement recipe. it can.

  FIG. 16B shows an example in which 1606, 1608, 1609, 1610, and 1612 to 1614 in FIG. 16A are integrated into one device 1616. As in this example, it is possible to divide or integrate arbitrary functions into arbitrary plural devices.

  According to the present embodiment, it is possible to estimate where in the tilt image each contour line of the pattern to be evaluated or its peripheral pattern can exist based on the predicted contour line calculated based on the design data. It is possible to set an image processing range and an image processing method appropriate for evaluating the pattern to be evaluated. Thereby, for example, in detecting the contour line of the pattern to be evaluated, failures such as erroneous detection of a contour line different from the contour line to be detected can be reduced, and stable shape evaluation can be realized.

DESCRIPTION OF SYMBOLS 100 ... xyz coordinate system (Coordinate system of an electron optical system) 101 ... Semiconductor wafer 102 ... Electron optical system 103 ... Electron gun 104 ... Electron beam (primary electron) 105. .. Condenser lens 106 ... Deflector 107 ... ExB deflector 108 ... Objective lens 109 ... Secondary electron detector 110, 111 ... Backscattered electron detector 112-114 ... A / D converter 115 ... processing / control unit 116 ... CPU 117 ... image memory 118, 126 ... processing terminal 119 ... stage controller 120 ... deflection control unit 121 ... stage 123 ..Imaging recipe creation device 124 ... Measurement recipe creation device
125 ... Image processing apparatus (shape measurement / evaluation) 127 ... Database (storage) 703A, 703B ... Measuring cursor 1601 ... Mask pattern design apparatus 1602 ... Mask drawing apparatus 1603 ... Exposure Developing device 1604. Etching device 1605, 1007... SEM device 1606, 1608... SEM control device 1609... EDA tool server 1610. ... Imaging / measurement recipe server 1614 ... Image processing device server (shape measurement / evaluation) 1615 ... Network.

Claims (9)

An apparatus for imaging a circuit pattern of a semiconductor device using a scanning charged particle microscope and evaluating the shape of the circuit pattern from the captured image,
An imaging area specifying means for specifying an area including a circuit pattern to be evaluated as an imaging area;
Imaging direction designating means for designating an irradiation direction (imaging direction) of charged particles irradiated by the scanning charged particle microscope with respect to the imaging area designated by the imaging area designating means;
Imaging means for capturing an imaging area designated by the imaging area designating means from the imaging direction to obtain a captured image;
Design data input means for inputting design data of a circuit pattern included in the imaging area designated by the imaging area designation means;
The predicted position (predicted contour line) of the contour line of the circuit pattern observed on the captured image obtained by imaging with the imaging unit is calculated using the design data input to the design data input unit and the information on the imaging direction. Expected contour calculation means to perform,
Collating means for obtaining a correspondence relationship between a captured image obtained by imaging with the imaging means and an expected contour calculated by the expected contour calculating means;
An image processing range / method setting means for setting an image processing range or an image processing method for processing a captured image obtained by imaging with the imaging means based on the expected contour calculated by the expected contour calculating means;
Shape evaluation means for evaluating the shape of the circuit pattern by processing a captured image obtained by imaging with the imaging means in accordance with the image processing range or image processing method set by the image processing range / method setting means. An apparatus for evaluating the shape of a circuit pattern. An apparatus for imaging a circuit pattern of a semiconductor device using a scanning charged particle microscope and evaluating the shape of the circuit pattern from the captured image,
An imaging area specifying means for specifying an area including a circuit pattern to be evaluated as an imaging area;
An imaging direction designating unit for designating an irradiation direction ( first imaging direction) of the charged particles irradiated by the scanning charged particle microscope with respect to the imaging region designated by the imaging region designating unit;
Design data input means for inputting design data of a circuit pattern included in the imaging area designated by the imaging area designation means;
Design data obtained by inputting an expected position (expected contour line) of a contour line of a circuit pattern observed on a captured image obtained by imaging the imaging region to the design data input unit and information on the first imaging direction An expected contour calculation means to calculate using,
Display means for displaying on the screen the predicted contour calculated by the predicted contour calculation means;
An imaging direction determining means for determining a second imaging direction based on the expected contour line displayed on the display means;
An apparatus for evaluating the shape of a circuit pattern, comprising: an imaging unit that captures an image of the imaging area from the second imaging direction determined by the imaging direction determination unit and obtains a captured image. A method of imaging a circuit pattern of a semiconductor device using a scanning charged particle microscope and evaluating the shape of the circuit pattern from the captured image,
An imaging area designating step for designating an area including a circuit pattern to be evaluated as an imaging area;
Irradiation direction of charged particles irradiated with the scanning charged particle microscope to the designated imaging region (
An imaging direction specifying step for specifying an imaging direction),
An imaging step of capturing the specified imaging region from the imaging direction with the scanning charged particle microscope to obtain a captured image;
A design data input step for inputting design data of a circuit pattern included in the designated imaging region;
A predicted contour calculation step of calculating a predicted position (expected contour line) of the contour line of the circuit pattern observed on the captured image obtained by imaging using the input design data and the information on the designated imaging direction. When,
A collation step for obtaining a correspondence relationship between the captured image obtained by imaging and the calculated expected contour line;
An image processing range / method setting step for setting an image processing range or an image processing method for processing a captured image obtained by imaging based on the calculated expected contour line;
And a shape evaluation step for evaluating the shape of the circuit pattern by processing the captured image in accordance with the set image processing range or image processing method. The predicted contour calculated in the predicted contour calculation step and the image processing range or image processing method set in the image processing range / method setting step are stored as a measurement recipe, and the shape evaluation is performed based on the measurement recipe. 4. The circuit pattern shape evaluation method according to claim 3, wherein steps are performed. In the imaging step, imaging from the imaging direction is performed by a method of deflecting the irradiation direction of charged particles in the imaging direction, a method of tilting a sample stage on which an evaluation target is placed, or a charged particle microscope electron optical system itself. 4. The circuit pattern shape evaluation method according to claim 3, wherein the predicted contour is calculated according to the imaging method in the predicted contour calculation step according to claim 1. In the predicted contour calculation step, design data in which two-dimensional layout information of the circuit pattern shape included in the imaging region is written and a design value of the height of the circuit pattern are input, respectively, and the design data and the Calculating a pseudo three-dimensional shape of the circuit pattern from a design value of height, and calculating an expected contour line by calculating a contour line of the pseudo three-dimensional shape observed from the imaging direction. 4. The circuit pattern shape evaluation method according to claim 3, wherein: In the image processing range / method setting step, a predicted misalignment range between the predicted contour and a contour actually observed on the captured image is set, and based on the predicted contour and the predicted misalignment range. 4. The circuit pattern shape evaluation method according to claim 3, wherein the image processing range or the image processing method is set. In the shape evaluation step, the size of the circuit pattern is measured according to the image processing range or the image processing method, the outline of the circuit pattern is detected, or the image feature quantity correlated with the three-dimensional shape of the circuit pattern is calculated. The circuit pattern shape evaluation method according to claim 3. A method of imaging a circuit pattern of a semiconductor device using a scanning charged particle microscope and evaluating the shape of the circuit pattern from the captured image,
An imaging area designating step for designating an area including a circuit pattern to be evaluated as an imaging area;
An imaging direction designating step of designating an irradiation direction ( first imaging direction) of charged particles irradiated with the scanning charged particle microscope with respect to the designated imaging region;
A design data input step for inputting design data of a circuit pattern included in the designated imaging region;
Using the input design data and the information on the designated first imaging direction, the predicted position (expected contour line) of the contour line of the circuit pattern observed on the captured image obtained by imaging the imaging region An expected contour calculation step to calculate,
An expected contour display step for displaying the calculated predicted contour on the screen;
An imaging direction determining step for determining a second imaging direction based on the expected contour line displayed on the screen;
A circuit pattern shape evaluation method comprising: an imaging step of imaging the designated imaging area from the second imaging direction determined in the imaging direction determination step to obtain a captured image.

Images