Scanning electron microscope and differential map generation method

Abstract

In generating the tilt map, the azimuth angle error caused by leakage magnetic fields from the objective lens is reduced. [Solution] The calculator 41 calculates tilt information based on the set of detected values ​​obtained from each minute surface in the sample. A tilt map is constructed from multiple tilt information corresponding to multiple minute surfaces. Each tilt information includes information that identifies the azimuth angle of each minute surface. The corrector 42 corrects each tilt information so that the azimuth angle of each minute surface is corrected according to the current sample observation conditions. The azimuth angle and tilt angle may be corrected sequentially or simultaneously.

Description

Technical Field

[0004]

[0001] The present invention relates to a scanning electron microscope and a tilt map generation method, and particularly relates to a technique for correcting a tilt map reflecting the shape of a sample surface.

Background Art

[0002] A scanning electron microscope is cited as one of charged particle beam devices. As a three-dimensional shape reconstruction method using a scanning electron microscope, a method using reflected electrons and a method using secondary electrons are known. Hereinafter, the three-dimensional shape reconstruction method using reflected electrons will be described.

[0003] A typical split-type reflected electron detector of a scanning electron microscope has a detection surface composed of four detection regions. In the process of two-dimensionally scanning an electron beam on a sample, reflected electrons are emitted from minute surfaces existing at each position on the sample. The reflected electrons are detected by the split-type reflected electron detector. The distribution of reflected electrons on the detection surface of the split-type reflected electron detector changes depending on the tilt state of the minute surface that has received the electron beam (and emitted the reflected electrons). Specifically, the tilt state of the minute surface means the tilt direction and tilt amount of the minute surface. Based on the four detection signals output from the four detection regions, tilt information representing the tilt state of the minute surface is calculated.

[0004] Specifically, the tilt direction of the minute surface is specified as the azimuth angle of the minute surface. Specifically, the tilt amount of the minute surface is specified as the inclination angle of the minute surface. Typically, the azimuth angle is the rotation angle around an axis parallel to the optical axis, and the inclination angle is the intersection angle with respect to an axis parallel to the optical axis. The azimuth angle and the inclination angle may be defined based on an axis other than the optical axis.

[0005] A gradient map is constructed from multiple inclination information corresponding to multiple positions (i.e., multiple minute surfaces) on the sample surface. The gradient map can also be called a gradient map. By applying known integration processes to the gradient map, the three-dimensional shape of the sample is reproduced (see, for example, Patent Documents 1 and 2). Photometric stereo and shading methods are known as such three-dimensional shape reconstruction methods. [Prior art documents] [Patent Documents]

[0006] [Patent Document 1] Japanese Patent Publication No. 2006-172919 [Patent Document 2] Japanese Patent Publication No. 2022-185757 [Overview of the Initiative] [Problems that the invention aims to solve]

[0007] Conventional three-dimensional shape reconstruction methods assume that backscattered electrons emitted from a sample travel in a straight line to the detection surface of a segmented backscattered electron detector. However, our research has shown that this assumption does not hold true depending on the sample observation conditions.

[0008] In other words, when the sample is close to the objective lens, when a high acceleration voltage is set and a high-intensity magnetic field is generated at the objective lens, or when a specific objective lens (e.g., a semi-in lens) is used, the leakage magnetic field from the objective lens has a non-negligible effect on the backscattered electrons emitted from the sample. More specifically, the leakage magnetic field deflects the backscattered electrons, causing errors in the calculated azimuth angle. This problem can also occur when generating a gradient map using secondary electrons, etc.

[0009] The object of the present invention is to improve the accuracy of the tilt map. Alternatively, the object of the present invention is to reduce or eliminate azimuth angle errors caused by leakage magnetic fields from the objective lens. Alternatively, the object of the present invention is to provide a new correction technique for tilt maps. [Means for solving the problem]

[0010] The scanning electron microscope according to the present invention is a detector that detects electrons emitted from each minute surface in a sample during the process of scanning an electron beam over the sample, and includes a detector having a plurality of detection regions, a calculator that calculates a tilt map consisting of a plurality of tilt information representing the tilt state of a plurality of minute surfaces in the sample based on a plurality of detection signals output from the plurality of detection regions, and a corrector that corrects the tilt map, wherein each of the tilt information includes information that specifies the azimuth angle of each of the minute surfaces, and the corrector corrects each of the tilt information so that the azimuth angle of each of the minute surfaces is corrected according to the current sample observation conditions.

[0011] The tilt map generation method according to the present invention includes the steps of detecting electrons emitted from each minute surface in a sample using a plurality of detection regions during the process of scanning a charged particle beam over the sample, and generating a tilt map consisting of a plurality of tilt information representing the tilt state of a plurality of minute surfaces in the sample based on a plurality of detection signals output from the plurality of detection regions, wherein each of the tilt information includes information that specifies the azimuth angle of each minute surface, and in the step of generating the tilt map, each of the tilt information is corrected so that the azimuth angle of each minute surface is corrected according to the current sample observation conditions. [Effects of the Invention]

[0012] According to the present invention, the accuracy of the tilt map can be improved. Alternatively, according to the present invention, azimuth angle errors caused by leakage magnetic fields from the objective lens can be reduced or eliminated. Alternatively, a new correction technique for tilt maps can be provided. [Brief explanation of the drawing]

[0013] [Figure 1] It is a block diagram showing a configuration example of a scanning electron microscope according to an embodiment. [Figure 2] It is a plan view showing an example of a detection surface in a detector. [Figure 3] It is a cross-sectional view showing a sample surface and a minute surface. [Figure 4] It is a diagram showing the emission of electrons from a minute surface. [Figure 5] It is a diagram showing a coordinate system. [Figure 6] It is a diagram showing a tilt vector corresponding to tilt information. [Figure 7] It is a diagram showing a plurality of tilt vectors corresponding to a plurality of minute surfaces. [Figure 8] It is a diagram showing an example of a tilt map. [Figure 9] It is a diagram showing a straight electron orbit. [Figure 10] It is a diagram showing an electron orbit curved under the influence of a leakage magnetic field. [Figure 11] It is a diagram showing an example of a correction function according to the first embodiment. [Figure 12] It is a diagram showing an example of a correction table according to the first embodiment. [Figure 13] It is a flowchart showing an operation according to an embodiment. [Figure 14] It is a diagram showing a three-dimensional shape generated based on a tilt map before correction. [Figure 15] It is a diagram showing a three-dimensional shape generated based on a tilt map after correction. [Figure 16] It is a diagram showing a correction table according to the second embodiment. [Figure 17] It is a diagram showing a correction table according to the third embodiment. [Figure 18] It is a flowchart showing a correction value calculation method by experiment. [Figure 19] It is a diagram showing a correction value calculation using a sphere. [Figure 20] It is a diagram showing a correction value calculation using a simulator. [Figure 21] This figure shows multiple spheres dispersed within the scanning area. [Figure 22] This figure shows the labeling process applied to backscattered electron images. [Figure 23] This diagram shows multiple horizontal lines set for multiple regions. [Figure 24] This figure shows a two-dimensional azimuth correction function. [Figure 25] This figure shows a two-dimensional tilt angle correction function. [Modes for carrying out the invention]

[0014] The embodiments will be described below with reference to the drawings.

[0015] (1) Outline of the Embodiment The scanning electron microscope according to this embodiment includes a detector, a calculator, and a corrector. The detector is a detector that detects electrons emitted from each minute surface in a sample during the process of scanning the sample with an electron beam, and has a plurality of detection regions. The calculator calculates a tilt map consisting of a plurality of tilt information representing the tilt state of a plurality of minute surfaces in a sample, based on a plurality of detection signals output from the plurality of detection regions. The corrector corrects the tilt map. Each tilt information includes information that identifies the azimuth angle of each minute surface. The corrector corrects each tilt information so that the azimuth angle of each minute surface is corrected according to the current sample observation conditions.

[0016] According to the above configuration, the azimuth angle error included in each piece of tilt information is reduced or eliminated by correcting the tilt map. Therefore, for example, it becomes possible to correctly reconstruct the three-dimensional shape of the sample based on the tilt map.

[0017] When the stray magnetic field from the objective lens acts on electrons emitted from each minute surface (emitted electrons), an azimuth angle error occurs in the calculation process of the tilt information. The degree of the stray magnetic field's effect, i.e., the degree of the azimuth angle error, changes depending on the sample observation conditions (current sample observation conditions) set for the scanning electron microscope during sample observation. Therefore, in the above configuration, the current sample observation conditions (which can also be called the current sample measurement conditions) are taken into consideration when correcting the tilt map. The current sample observation conditions may include, for example, parameters that affect the strength of the stray magnetic field, parameters that define the spatial relationship between the stray magnetic field and emitted electrons, etc.

[0018] Each microsurface is a local region that receives an electron beam and emits electrons. The electrons emitted from each microsurface are, for example, backscattered electrons. Multiple detection regions refer to at least three detection regions. The tilt state of a microsurface refers to the tilt direction and amount of the microsurface. The tilt direction of a microsurface is expressed as the azimuth angle of the microsurface, and the amount of tilt of a microsurface is expressed as the tilt angle of the microsurface. Normally, the tilt map is corrected after its generation, but the generation and correction of the tilt map may be performed simultaneously.

[0019] In the embodiment, the compensator corrects each tilt information based on azimuth correction information corresponding to the current sample observation conditions. The azimuth correction information is generated in advance or generated when the tilt map is corrected.

[0020] In this embodiment, the azimuth correction information includes an azimuth correction value that is commonly applied to multiple microsurfaces. This configuration allows for easy correction of the tilt map. Alternatively, the azimuth correction information includes multiple azimuth correction values ​​that are applied to multiple microsurfaces. This configuration further improves the accuracy of the corrected tilt map. When the azimuth error fluctuates significantly within the scanning area, it is preferable to provide multiple azimuth correction values ​​corresponding to multiple coordinates within the scanning area.

[0021] In this embodiment, the azimuth correction information is information obtained in advance through experiments or simulations under sample observation conditions corresponding to the current sample observation conditions, or information generated from such information. A scanning electron microscope is used in the above experiments. A simulator (usually a computer) is used in the above simulations.

[0022] The scanning electron microscope according to this embodiment includes a storage unit that stores multiple azimuth correction information obtained by repeating experiments or simulations while sequentially setting multiple sample observation conditions. The corrector identifies the azimuth correction information corresponding to the current sample observation condition based on the multiple azimuth correction information. In this embodiment, the current sample observation condition includes at least one of the acceleration voltage and working distance.

[0023] In the embodiment, each inclination information further includes information that identifies the inclination angle of each minute surface. The corrector corrects each inclination information so that both the azimuth angle and inclination angle of each minute surface are corrected. The incident angle of the electron beam changes depending on the position within the scanning area. Changes in the incident angle cause inclination angle errors. With the above configuration, both azimuth angle errors and inclination angle errors are reduced, so the accuracy of the inclination map can be further improved. Correction for inclination angle errors caused by other factors may also be performed.

[0024] In this embodiment, the compensator corrects each inclination information based on the inclination angle correction value corresponding to each minute surface so that the inclination angle of each minute surface is corrected.

[0025] The tilt map generation method according to the embodiment comprises a detection step and a generation step. In the detection step, electrons emitted from each minute surface in the sample are detected by multiple detection regions during the process of scanning the sample with a charged particle beam. In the generation step, a tilt map consisting of multiple tilt information representing the tilt state of multiple minute surfaces in the sample is generated based on multiple detection signals output from the multiple detection regions. Each tilt information includes information that identifies the azimuth angle of each minute surface. In the generation step, each tilt information is corrected so that the azimuth angle of each minute surface is corrected according to the current sample observation conditions.

[0026] The above method for generating a gradient map is performed in a charged particle beam apparatus. The gradient map may be corrected after its generation, or it may be generated and corrected simultaneously. The charged particle beam irradiated onto the sample is, for example, an electron beam, an ion beam, etc.

[0027] (2) Details of the embodiment Figure 1 shows a scanning electron microscope according to an embodiment. The illustrated scanning electron microscope has the function of constructing (reconstructing) the three-dimensional shape of the sample surface by detecting backscattered electrons emitted from the sample. Secondary electrons may be detected instead of backscattered electrons.

[0028] The scanning electron microscope has a measuring unit 10 and a calculation and control unit 12. The measuring unit 10 corresponds to the observation unit. The measuring unit 10 has a microscope tube 14 and a housing 16. The housing 16 supports the microscope tube 14. Inside the microscope tube 14 are an electron gun, a focusing lens, a scanning coil, an objective lens, etc. These are the components that scan the electron beam 13 while irradiating the sample 24 with the electron beam 13. The internal space of the housing 16 is the sample chamber 18. A movable stage 26 is provided in the sample chamber 18. The movable stage 26 holds the sample 24. The position and orientation of the sample 24 are changed by the movable stage 26.

[0029] The lower end of the objective lens 20 enters the sample chamber 18. A detector 22 is provided near the lower end surface of the objective lens 20. The detector 22 may be attached to the objective lens 20. The detector 22 is a backscattered electron detector, more specifically, a segmented backscattered electron detector. In the illustrated configuration example, the detection surface of the detector 22 is divided into four segments, that is, the detector 22 has four detection regions. In the configuration example shown in Figure 1, four detection signals s1, s2, s3, and s4 are output in parallel from the four detection regions.

[0030] The distance from the lower end surface (or reference level) of the objective lens 20 to the surface of the sample 24 (i.e., the electron beam focus) is called the working distance (WD). In Figure 1, the secondary electron detector, X-ray detector, etc., are omitted from the illustration.

[0031] The arithmetic control unit 12 will now be described. In this embodiment, the arithmetic control unit 12 includes a computer as an information processing device. Reference numeral 48 indicates a processor. The processor 48 is, for example, a CPU that executes programs. In Figure 1, multiple functions performed by the processor 48 are represented by multiple blocks. The arithmetic control unit 12 has a storage unit, which is not shown. Specifically, a memory unit is connected to the processor 48. The memory stores information for generating and correcting the slope map.

[0032] The control unit 28 controls the operation of each element constituting the scanning electron microscope. The scanning signal generator 30 generates the scanning signal that is supplied to the scanning coil in the microscope tube 14. The scanning signal is also supplied to the preprocessing circuit 34.

[0033] Four detection signals s1 to s4 are input in parallel to the four input circuits 32A to 32D. Each input circuit 32A to 32D has a preamplifier, an I / V converter, etc. The I / V converter converts the current signal into a voltage signal. The four detection signals output in parallel from the four input circuits 32A to 32D are sent to the pre-processing circuit 34.

[0034] The preprocessing circuit 34 has four A / D converters. Each A / D converter converts each detection signal (analog signal) into a detection value sequence (digital signal). Within the preprocessing circuit 34, each detection value set is associated with each coordinate identified by the scan signal. A detection value set associated with a given coordinate consists of four detection values ​​S1, S2, S3, and S4 obtained from that coordinate.

[0035] The data processing unit 36 ​​includes a slope map generation unit 38 and a three-dimensional shape reconstruction unit 40. The slope map generation unit 38 is a module that generates a slope map and includes a calculator 41 and a corrector 42.

[0036] The arithmetic unit 41 calculates slope information representing the inclination state of a minute surface present at each coordinate within the scanning region, based on the detection set associated with that coordinate. A slope map is constructed from multiple slope information corresponding to multiple coordinates. Specifically, each piece of slope information is a slope vector defined by the x-direction slope amount and the y-direction slope amount.

[0037] The corrector 42 corrects the tilt map according to the sample observation conditions (current sample observation conditions) set in the measurement unit 10 during sample observation. In the configuration example shown in Figure 1, the corrector 42 individually corrects multiple tilt information corresponding to multiple coordinates based on a combination of multiple setting values ​​that define the current sample observation conditions.

[0038] More specifically, the corrector 42 determines an azimuth correction value that is commonly applied to multiple coordinates, for example, based on a combination of the current acceleration voltage and the current WD. A correction function or correction table in memory is used for this purpose. Subsequently, the corrector 42 corrects each tilt information based on the determined azimuth correction value so that the azimuth error that occurred at each coordinate is reduced or eliminated. This will be described in detail later. When correcting the tilt map, either the current acceleration voltage or the current WD may be referenced, or other setting values ​​such as the magnification may be referenced along with the current acceleration voltage and the current WD.

[0039] The three-dimensional shape reconstruction unit 40 reconstructs the three-dimensional shape of the sample surface based on the corrected tilt map. Specifically, the three-dimensional shape reconstruction unit 40 applies integration processing to the corrected tilt map and generates the three-dimensional shape as a result. The method of reconstructing the three-dimensional shape based on the tilt map is a known technique.

[0040] The display unit 44 is configured, for example, as an LCD. The screen of the display unit 44 displays SEM images (backscattered electron images, secondary electron images, etc.), images representing the three-dimensional shape of the sample surface, etc. An input device 46 is connected to the control unit 28. The input device 46 includes a keyboard, a pointing device, etc.

[0041] Figure 2 shows the detection surface 55 of the detector 22 shown in Figure 1. As described above, the detection surface 55 consists of four detection regions D1, D2, D3, and D4. Each detection region D1, D2, D3, and D4 has a fan-shaped form. The x-direction is the first horizontal direction, and the y-direction is the second horizontal direction. Two detection regions D1 and D3 are aligned in the x-direction, and two detection regions D2 and D4 are aligned in the y-direction. The detector 22 has an aperture 50 through which the electron beam passes. The detection surface 55 may be composed of three detection regions or five or more detection regions.

[0042] The slope map will be explained below using Figures 3 to 8.

[0043] Figure 3 shows a vertical cross-section of sample 24. The surface of sample 24 has a three-dimensional shape. The average plane defined on the surface of sample 24 is the sample surface 54. In the illustrated example, the sample surface 54 is a plane perpendicular to the optical axis.

[0044] The microscopic surface 56 is the surface that receives the electron beam 52 and emits backscattered electrons. The microscopic surface 56 can also be called the electron beam incidence surface. In Figure 3, the microscopic surface 56 is exaggerated. This is also true in Figures 4 and 5, which will be explained below.

[0045] In Figure 4, (A) shows the first backscattered electron emission state, and (B) shows the second backscattered electron emission state. In both states, the sample surface and the detection surface 55 are parallel. The electron beam 58 (which can also be called the optical axis) passes through the center of the detector 22. The sample surface and the electron beam 58 are perpendicular to each other.

[0046] In the first backscattered electron emission state, the normal to the minute surface 56A is parallel to the electron beam 58. The group of backscattered electrons 60 emitted from the minute surface 56A reaches the four detection regions D1, D2, D3, and D4 that constitute the detection surface 55 of the detector 22 almost equally.

[0047] In the second backscattered electron emission state, the normal to the minute surface 56B is tilted with respect to the electron beam 58. The group of backscattered electrons 62 emitted from the minute surface 56B arrives at the detection surface 55 in the detector 22 with an uneven distribution. Specifically, in the illustrated example, the detected value obtained from detection region D1 is the largest, the detected value obtained from detection region D3 is the smallest, and the two detected values ​​obtained from the two detection regions D2 and D4 are the intermediate values, and these two detected values ​​are approximately equal. The four detected values, or set of detected values, represent information about the tilt state (tilt direction and amount) of the minute surface.

[0048] Figure 5 shows the coordinate system. For simplicity, in Figure 5, the sample surface is perpendicular to the optical axis 200, and the detection surface 55 of the detector 22 is parallel to the sample surface. The optical axis 200 passes through the center of the detector 22. The z axis is parallel to the optical axis 200. The x axis corresponds to the alignment of detection regions D1 and D3. The y axis corresponds to the alignment of detection regions D2 and D4. The x, y, and z axes are orthogonal to each other.

[0049] The symbol 56C indicates a minute surface parallel to the sample surface. The symbol 56D indicates a tilted minute surface. The normal vector R corresponds to the normal of the tilted minute surface 56D. As shown in the figure, the tilt direction of the minute surface 56D is represented by the azimuth angle φ. The azimuth angle φ is the rotation angle around the axis parallel to the optical axis 200. The amount of tilt of the minute surface 56 is represented by the tilt angle θ. The tilt angle θ is the angle of intersection of the normal vector R with respect to the axis parallel to the optical axis. Note that a coordinate system other than the coordinate system shown in Figure 5 may be used.

[0050] Several formulas have been proposed to calculate the inclination information of a microsurface based on a set of detected values ​​obtained from the microsurface. In this embodiment, the following formulas (1-1) and (1-2) are used.

number

[0051] Gx is the gradient amount (gradient component) in the x-direction, and Gy is the gradient amount (gradient component) in the y-direction. The gradient vector is defined by these two gradient amounts. The gradient vector corresponds to the gradient information. S1 shows the detected value obtained from detection region D1. S2 shows the detected value obtained from detection region D2. S3 shows the detected value obtained from detection region D3. S4 shows the detected value obtained from detection region D4. k is the conversion coefficient.

[0052] Figure 6 shows the gradient vector S defined by Gx and Gy as described above. The direction of the gradient vector S represents the azimuth angle φ of the infinitesimal surface. The magnitude of the gradient vector S represents the magnitude of the inclination angle θ of the infinitesimal surface.

[0053] During the two-dimensional scanning process of an electron beam, multiple sets of detection values ​​corresponding to multiple coordinates are obtained sequentially. For each coordinate, a gradient vector (i.e., gradient information of a small surface) is calculated based on the set of detection values ​​obtained from that coordinate. The matrix composed of multiple gradient vectors corresponding to multiple coordinates is the gradient map.

[0054] Figure 7 shows a gradient vector array 64. The gradient vector array 64 consists of multiple gradient vectors S corresponding to multiple coordinates. As described above, each gradient vector S is defined by a gradient amount Gx in the x direction and a gradient amount Gy in the y direction.

[0055] Figure 8 illustrates a slope map 66. The slope map 66 is composed of multiple slope information corresponding to multiple coordinates. In this embodiment, each slope information is composed of a slope amount Gx in the x direction and a slope amount Gy in the y direction, as described above.

[0056] Next, using Figures 9 to 11, we will explain the necessity of correcting the slope map and the method of correcting the slope map.

[0057] Figure 9 shows a conventional backscattered electron trajectory model. (A) is an oblique view, and (B) is a plan view. Backscattered electrons emitted from the minute surface 56E travel in a straight line (see reference numeral 68) and reach a specific position on the detection surface 55. Conventionally, the inclination information of the minute surface is calculated based on this assumption.

[0058] However, the above model does not hold true depending on the sample observation conditions. For example, the strength of the leakage magnetic field increases when the acceleration voltage is high. When the working width (WD) is small, the effect of the leakage magnetic field on backscattered electrons increases. The strength of the leakage magnetic field also increases when using certain objective lenses. In such cases, a deflection effect occurs on backscattered electrons moving through the leakage magnetic field.

[0059] Figure 10 shows a backscattered electron trajectory model that takes into account the effect of the leakage magnetic field from the objective lens. (A) is an oblique view, and (B) is a plan view. When a leakage magnetic field acts on backscattered electrons emitted from the minute surface 56F, a rotational force is generated on the backscattered electrons about an axis parallel to the optical axis, causing the backscattered electron trajectory to become arc-shaped (see reference numeral 70). The backscattered electrons will arrive at a different position than the destination that would be assumed if a linear trajectory (see reference numeral 68A) were assumed. As a result, the tilt information calculated for each coordinate within the scanning area becomes inaccurate, meaning that the azimuth angle error included in the tilt information increases. If the three-dimensional shape is reconstructed based on a tilt map containing such errors, distortion will occur in the three-dimensional shape.

[0060] Therefore, in this embodiment, the tilt map is corrected using azimuth correction information corresponding to the current sample observation conditions. In the first embodiment, the azimuth correction information is, for example, an azimuth correction value φc that is applied commonly to multiple coordinates, that is, to multiple microplanes. The azimuth correction value φc is determined in advance by experiment or simulation. This will be described later.

[0061] In the first embodiment, for example, as shown in Figure 11, a correction function 72 is provided for each acceleration voltage (AV). The horizontal axis represents WD, and the vertical axis represents the azimuth correction value φc. The correction function 72 is merely an example. For example, N correction functions 72 are provided corresponding to N acceleration voltages, where N is an integer greater than or equal to 2. A specific correction function 72 corresponding to the current acceleration voltage is selected from among the N correction functions 72, and the azimuth correction value φc is determined by providing the current WD to that specific correction function 72. The azimuth correction value φc may also be determined based on multiple correction functions 72 corresponding to the current acceleration voltage.

[0062] Based on the determined azimuth correction value φc, the inclination amount Gx in the x-direction and the inclination amount Gy in the y-direction are corrected according to equations (2-1) and (2-2) below.

number

[0063] Gx' is the corrected x-axis tilt, and Gy' is the corrected y-axis tilt.

[0064] The above correction reduces or eliminates azimuth errors caused by leakage magnetic fields. By correcting the tilt information for each coordinate within the scanning area, a corrected tilt map is obtained.

[0065] The correction function 72 shown in Figure 11 is obtained as a function that approximates multiple correction values ​​obtained by experiment or simulation, as will be described later. A correction function including two parameters, acceleration voltage and WD, may be used. The azimuth correction value may be determined based on either acceleration voltage or WD. The azimuth correction value may be determined based on three or more parameters, including acceleration voltage and WD.

[0066] Figure 12 shows a group of correction tables 82 according to the first embodiment. The group of correction tables 82 is used in place of the group of correction functions. The group of correction tables 82 corresponds to a database and consists of multiple correction tables 84 corresponding to multiple acceleration voltages. Each correction table 84 consists of multiple records 86 corresponding to multiple WD intervals. Each record 86 has two numerical values ​​88 and 90 that define the upper and lower limits of the WD interval, and also has an azimuth correction value φc92 corresponding to the WD interval. Based on such a group of correction tables 82, a specific azimuth correction value φc may be determined from the current sample observation conditions.

[0067] In addition, multiple WDs may be registered in the correction table 84 instead of multiple WD intervals. In that case, two WDs close to the current WD may be identified, and the actual azimuth correction value to be used may be determined by weighted interpolation of the two azimuth correction values ​​corresponding to those two WDs.

[0068] Figure 13 shows the operation or process according to the embodiment as a flowchart. S10 shows a process that is performed in advance, and S20 shows a process that is performed when the sample is measured.

[0069] S10 includes S12 and S14. In S12, the experiment or simulation is repeatedly performed while changing the sample observation conditions (combination of acceleration voltage and WD). As a result, in S14, a group of correction functions consisting of multiple correction functions corresponding to multiple sample observation conditions is generated, or a group of correction tables consisting of multiple correction tables corresponding to multiple sample observation conditions is generated. The group of correction functions or correction tables is registered in memory as a storage unit.

[0070] S20 includes steps S22 through S30. In S22, the sample is measured. In S24, a slope map is generated. S22 and S24 may be executed simultaneously, or S24 may be executed after S22 is completed. In S26, the azimuth correction value is determined based on the current acceleration voltage and current WD combination. A specific correction function or correction table stored in memory is used for this purpose.

[0071] In S28, the tilt map is corrected based on the determined azimuth correction value. In S30, the three-dimensional shape is reconstructed based on the corrected tilt map.

[0072] Figure 14 shows the three-dimensional shape 74 reconstructed based on the uncorrected slope map. Figure 15 shows the three-dimensional shape 78 reconstructed based on the corrected slope map. In Figures 14 and 15, the vertical axis is the height axis.

[0073] In the three-dimensional shape 74, the portion indicated by reference numeral 76 has relatively large distortion. Other portions of the three-dimensional shape 74 also have relatively large distortion. On the other hand, in the three-dimensional shape 78, the portion indicated by reference numeral 80 has relatively small distortion. Other portions of the three-dimensional shape 78 also have relatively small distortion. In this way, by correcting the slope map, the results of processing using the corrected slope map can be improved.

[0074] Figure 16 shows a correction table 94 according to the second embodiment. In the second embodiment, the configuration shown in Figure 1 is also used. In the second embodiment, different azimuth correction values ​​are obtained for each coordinate within the scanning area. For example, when using a semi-in-lens, the strength of the stray magnetic field tends to change within the scanning area. In such cases, the adoption of the second embodiment is desirable.

[0075] More specifically, N (=m × n) combinations are defined based on m acceleration voltages and n WDs. Both m and n are integers greater than or equal to 2. In the second embodiment, N correction tables 94 corresponding to the N combinations are created in advance, and the N correction tables 94 are registered in memory. Each correction table 94 has multiple azimuth correction values ​​φc corresponding to multiple coordinates.

[0076] In correcting the slope map, first, a specific correction table 94 corresponding to the current acceleration voltage and current WD is selected from among N correction tables 94. Next, according to that specific correction table 94, an azimuth correction value corresponding to that coordinate is determined for each coordinate. The slope information corresponding to that coordinate is corrected using the determined azimuth correction value. Instead of multiple correction tables 94, a single or multiple correction functions may be used.

[0077] Figure 17 shows the correction table group 96 according to the third embodiment. In the third embodiment, the configuration shown in Figure 1 is also used. Within the scanning area, the electron beam incidence angle changes according to the coordinates. Due to the change in the electron beam incidence angle, a tilt angle error separate from the azimuth angle error occurs. Therefore, in the third embodiment, both azimuth angle correction and tilt angle correction are performed when correcting the tilt map.

[0078] As described above, N (=m × n) combinations are defined based on m acceleration voltages and n WDs. The correction table group 96 consists of N correction tables 98 corresponding to the N combinations. Each correction table 98 consists of multiple records corresponding to multiple coordinates 100. Each record includes an azimuth angle correction value φc (see reference numeral 102) and a tilt angle correction value (Gxc, Gyc) (see reference numeral 104).

[0079] In the third embodiment, a specific correction table 98 is selected from the correction table group 96 based on the current acceleration voltage and current WD combination. Using this specific correction table 98, the azimuth angle and the tilt angle are corrected for each coordinate.

[0080] For example, when correcting the azimuth angle, equations (2-1) and (2-2) described above are used, while when correcting the tilt angle, equations (3-1) and (3-2) below are used.

number

[0081] Gx'' is the corrected tilt amount in the x-direction. Gy'' is the corrected tilt amount in the y-direction. The tilt angle may be corrected after the azimuth angle, or the azimuth angle may be corrected after the tilt angle. The azimuth angle and tilt angle may be corrected simultaneously.

[0082] Next, we will explain how to calculate the azimuth correction value and the tilt correction value.

[0083] Figures 18 and 19 show the method for calculating the azimuth correction value experimentally. This calculation method calculates the azimuth correction value used in the first embodiment described above. The same sample observation conditions as those to be set for future sample measurements are set, and the azimuth correction value is calculated under the set sample observation conditions. Alternatively, multiple azimuth correction values ​​are calculated while changing the sample observation conditions. This will be explained in detail below.

[0084] A sphere, serving as the sample, is placed within the scanning region of the electron beam. The sphere is actually placed on a sample stage. Preferably, the sphere is a perfect sphere. An ellipsoid or other type of sphere may be used instead. The diameter of the sphere may be, for example, about 30% of the shorter side of the scanning region. The electron beam is scanned two-dimensionally within the scanning region containing the sphere (see S40). Based on the set of detection values ​​obtained for each coordinate during this process, a backscattered electron image as a SEM image is generated, and a tilt map is also generated.

[0085] Figure 19 shows the backscattered electron image 114 produced by the experiment. The backscattered electron image 114 includes a spherical image 116. The center O of the spherical image 116 is identified, and a horizontal line passing through the center O is defined (see S42). The point (intersection) A where the horizontal line intersects the contour 116a of the spherical image 116 is identified (see S44). In the backscattered electron image 114, a predetermined point (representative point) B between the center O and point A is identified (see S46). The representative point B is, for example, the midpoint between the center O and intersection A.

[0086] From the slope map generated by the sample measurement, the slope amount Gx in the x-direction and the slope amount Gy in the y-direction are extracted as slope information corresponding to point B. If there is no influence of the leakage magnetic field, the slope amount Gy in the y-direction is zero. If there is an influence of the leakage magnetic field, a slope amount Gy in the y-direction occurs, and the greater the influence of the leakage magnetic field, the larger the slope amount Gy in the y-direction becomes. Based on the slope amount Gy in the y-direction, the magnitude of the deflection effect (i.e., azimuth angle error) due to the leakage magnetic field can be estimated (see S48).

[0087] For example, the azimuth correction value φc can be calculated based on equation (4) below.

number

[0088] The above azimuth correction value φc cancels out the azimuth error. In the first embodiment, the azimuth correction value φc obtained as described above is applied commonly to all coordinates (i.e., all infinitesimal surfaces).

[0089] In Figure 19, multiple azimuth correction values ​​may be calculated within a certain interval 118 between the center O and intersection A, and the average value calculated from these multiple azimuth correction values ​​may be used as the actual azimuth correction value. Alternatively, an azimuth correction value corresponding to point D may be calculated along with the azimuth correction value corresponding to point B, and the average value obtained from the two azimuth correction values ​​may be used as the actual azimuth correction value. Point D is, for example, the midpoint between the center O and intersection C. Intersection C is the point where the horizontal line 117 intersects with the contour 116a, and is on the opposite side from point A.

[0090] Alternatively, a vertical line may be defined instead of the horizontal line 117, and the azimuth correction value may be calculated using this line. Alternatively, an average value may be calculated based on a first azimuth correction value corresponding to a representative point on the horizontal line 117 and a second azimuth correction value corresponding to a representative point on the vertical line, and this average value may be used as the azimuth correction value actually used.

[0091] Figure 20 shows the method for calculating the azimuth correction value using simulation. This calculation method calculates the azimuth correction value used in the first embodiment described above. The same sample observation conditions as those to be set during future sample measurements are set, and the azimuth correction value is calculated under the set sample observation conditions. Alternatively, multiple azimuth correction values ​​are calculated while changing the sample observation conditions. This will be explained in detail below.

[0092] In Figure 20, the simulator 120 is configured, for example, by a computer. The simulator 120 includes a structural model 124, an orbit calculation formula 126, a correction value calculation formula 128, etc. The structural model 124 includes an objective lens model, a detector model, a sample model, etc. The user specifies the type of objective lens and a set of parameters such as acceleration voltage and WD.

[0093] Based on the input information, the simulator 120 calculates the trajectory of backscattered electrons emitted from the sample using the trajectory calculation formula 126. Subsequently, based on the calculated trajectory, the simulator 120 calculates an azimuth correction value using the correction value calculation formula 128. This azimuth correction value is then registered in the memory of the scanning electron microscope.

[0094] When carrying out the second embodiment described above, the azimuth correction value array may be calculated experimentally prior to that, as follows.

[0095] As shown in Figure 21, the group of spheres 132 is distributed within the scanning region 130. Ideally, the group of spheres 132 is arranged so that multiple spheres are evenly distributed throughout the entire scanning region 130. Preferably, the diameter of each sphere is 10% or less of the length of the shorter side of the scanning region 130. In the illustrated example, the group of spheres 132 consists of multiple spheres of various sizes. Multiple spheres of the same size may also be used.

[0096] After the arrangement of the sphere group 132, the electron beam is scanned in two dimensions within the scanning region 130. Based on the set of detected values ​​for each coordinate obtained as a result, a backscattered electron image and a tilt map are generated.

[0097] Figure 22 shows the generated backscattered electron image 134. The backscattered electron image 134 includes a group of spherical images 136. Labeling is applied to the backscattered electron image 134, which extracts each spherical image as an isolated region, and assigns labels (#1 to #9) to each spherical image.

[0098] As shown in Figure 23, in the backscattered electron image 134 after labeling, a horizontal line 138 passing through the center of each spherical image 137 is defined. Similarly, the intersection points where the horizontal line 138 intersects the contour of the spherical image 137 are identified. An azimuth correction value is calculated from the tilt information corresponding to a predetermined point between the center and the intersection point.

[0099] As described above, multiple azimuth correction values ​​corresponding to multiple representative coordinates within the scanning area are obtained. By mapping the multiple azimuth correction values ​​onto the xy-θc coordinate system and approximating the multiple azimuth correction values ​​with a surface (polynomial), a correction function that includes coordinates x and y as variables is obtained. This correction function is used in the second embodiment described above. The correction function or an equivalent correction table may be obtained by simulation.

[0100] Based on the azimuth angle correction values ​​for each coordinate obtained as described above, and the tilt angle correction values ​​for each coordinate obtained as described below, a correction function or correction table used in the third embodiment can be generated. In that case, multiple correction functions or multiple correction tables can be generated while changing the sample observation conditions.

[0101] When calculating tilt angle correction values ​​experimentally, first, a planar sample covering the entire scanning area is placed on the sample stage. A planar sample is a sample with a top surface that is a perfect plane (horizontal plane). Next, the electron beam is scanned in two dimensions within the scanning area. This acquires a set of detected values ​​for each coordinate. Based on the set of detected values ​​for each coordinate, tilt information, namely the tilt amount Gx in the x direction and the tilt amount Gy in the y direction, is determined.

[0102] Since these tilt amounts Gx and Gy correspond to the tilt angle error originating from the electron beam incidence angle, the tilt angle correction values ​​(Gxc, Gyc) can be obtained using equations (5-1) and (5-2) shown below.

number

[0103] The tilt angle correction values ​​(Gxc, Gyc) correspond to the tilt angle correction value θc, as follows:

number

[0104] When equations (3-1) and (3-2) above are used during tilt angle correction, the tilt angle correction values ​​(Gxc, Gyc) are used. When θ is directly corrected during tilt angle correction, the tilt angle correction value θc is used. A set of correction functions or correction tables for the third embodiment may be generated by simulation.

[0105] Figure 24 shows the correction function 106 that can be used in the second and third embodiments. The first horizontal axis is the x-axis, and the second horizontal axis is the y-axis. The vertical axis is the axis showing the azimuth correction value φc. The correction function 106 is defined as a surface that approximates multiple measured values ​​(azimuth correction values) 108 obtained by experiment or simulation. In practice, multiple correction functions 106 corresponding to multiple sample observation conditions are generated.

[0106] Figure 25 shows the correction function 110 that can be used in the third embodiment. The first horizontal axis is the x-axis, and the second horizontal axis is the y-axis. The vertical axis is the axis representing the tilt angle correction value θc. The correction function 110 is defined as a surface that approximates multiple measured values ​​(azimuth angle correction values) 112 obtained by experiment or simulation. In practice, multiple correction functions 110 corresponding to multiple sample observation conditions are generated.

[0107] The tilt map generation method, including the tilt map correction described above, may be implemented in charged particle beam apparatus other than electron microscopes. In the above embodiment, acceleration voltage and WD were used as sample observation conditions, but only one of them may be used. Alternatively, three or more parameters related to the error of tilt information may be used as sample observation conditions. In the above embodiment, in tilt map correction, each piece of tilt information is corrected so that at least the azimuth angle is corrected. If the tilt angle error changes depending on the composition of the sample, the composition may be estimated for each individual coordinate, and the tilt angle may be corrected based on the estimated composition. [Explanation of Symbols]

[0108] 10 Measurement unit (observation unit), 12 Calculation control unit, 20 Objective lens, 22 Detector (segmented backscattered electron detector), 24 Sample, 38 Tilt map generation unit, 40 Three-dimensional shape reconstruction unit, 41 Calculator, 42 Corrector.

Claims

1. A detector that detects electrons emitted from each minute surface of a sample during the process of scanning the sample with an electron beam, the detector having a plurality of detection regions, A calculator that calculates a tilt map consisting of multiple tilt information representing the tilt state of multiple minute surfaces in the sample, based on multiple detection signals output from the multiple detection regions, A corrector for correcting the aforementioned slope map, Includes, Each of the aforementioned inclination information includes information that identifies the azimuth angle of each of the aforementioned minute surfaces, The corrector corrects the inclination information so that the azimuth angle of each minute surface is corrected according to the current sample observation conditions. A scanning electron microscope characterized by the following features.

2. In the scanning electron microscope according to claim 1, The correction of the azimuth angle of each of the aforementioned minute surfaces reduces the azimuth angle error caused by the leakage magnetic field from the objective lens acting on electrons emitted from each of the aforementioned minute surfaces. A scanning electron microscope characterized by the following features.

3. In the scanning electron microscope according to claim 1, The corrector corrects each of the inclination information based on the azimuth angle correction information corresponding to the current sample observation conditions. A scanning electron microscope characterized by the following features.

4. In the scanning electron microscope according to claim 3, The aforementioned azimuth correction information includes an azimuth correction value that is applied in common to the plurality of minute surfaces. A scanning electron microscope characterized by the following features.

5. In the scanning electron microscope according to claim 3, The azimuth correction information includes a plurality of azimuth correction values ​​applied to the plurality of minute surfaces, A scanning electron microscope characterized by the following features.

6. In the scanning electron microscope according to claim 3, The aforementioned azimuth angle correction information is information obtained in advance through experiments or simulations under sample observation conditions corresponding to the current sample observation conditions, or information generated from such information. A scanning electron microscope characterized by the following features.

7. In the scanning electron microscope according to claim 3, It includes a memory unit that stores multiple azimuth angle correction information obtained by repeatedly conducting experiments or simulations while sequentially setting multiple sample observation conditions. The corrector identifies the azimuth correction information corresponding to the current sample observation conditions based on the plurality of azimuth correction information. A scanning electron microscope characterized by the following features.

8. In the scanning electron microscope according to claim 3, The current sample observation conditions include at least one of the following: acceleration voltage and working distance. A scanning electron microscope characterized by the following features.

9. In the scanning electron microscope according to claim 1, Each of the aforementioned inclination information further includes information that identifies the inclination angle of each of the aforementioned minute surfaces. The corrector corrects the tilt information such that both the azimuth angle and tilt angle of each minute surface are corrected. A scanning electron microscope characterized by the following features.

10. In the scanning electron microscope according to claim 9, The corrector corrects the inclination information so that the inclination angle of each minute surface is corrected based on the inclination angle correction value corresponding to each minute surface. A scanning electron microscope characterized by the following features.

11. The process involves scanning a charged particle beam across a sample and detecting electrons emitted from each minute surface of the sample using multiple detection regions. A step of generating a slope map consisting of multiple slope information representing the slope state of multiple minute surfaces in the sample, based on multiple detection signals output from the multiple detection regions, Includes, Each of the aforementioned inclination information includes information that identifies the azimuth angle of each of the aforementioned minute surfaces, In the process of generating the tilt map, the tilt information is corrected so that the azimuth angle of each minute surface is corrected according to the current sample observation conditions. A method for generating a slope map characterized by the following features.

Images