JP3304681B2 - Electron microscope and three-dimensional atomic array observation method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a point defect which is a cause of failure in an integrated device such as a memory device and a high-speed operation device,
The present invention relates to a three-dimensional atomic arrangement observation method suitable for three-dimensional observation of impurity atoms and their clusters at the level of one atom, and to an electron microscope having image processing means by this method.

[0002]

2. Description of the Related Art A method of three-dimensionally observing point defects, impurity atoms and their clusters in a crystal at the level of one atom by using an electron microscope is disclosed in Japanese Patent Application Laid-Open No. 4-3372.
There is a three-dimensional atomic arrangement observation apparatus and method described in Japanese Patent Publication No. 36-36. The publication discloses observation of the sample in a two-dimensional image (electron microscope image) having a contrast depending on the atomic species of the sample formed by an electron beam scattered within a certain angle range by the sample, The same structure in the sample is performed from various directions by tilting the sample, and the resulting positional displacement of the same atom between a plurality of two-dimensional images having different observation directions is image-processed based on the observation direction. Thus, a technique for specifying three-dimensional coordinates and constructing a three-dimensional atomic arrangement image has been disclosed. This publication also discloses that an atomic species is identified by analyzing the relationship between the scattering angle range of the electron beam used for imaging and the image contrast.

[0003]

SUMMARY OF THE INVENTION The above prior art invention includes a configuration of an electron microscope apparatus suitable for three-dimensional atomic arrangement observation, a method of observing a two-dimensional image, and a plurality of two-dimensional images observed from various directions. Describes that a three-dimensional image is reconstructed by specifying three-dimensional coordinates on the basis of the above, but no specific means for realizing the reconstruction of the three-dimensional image is described. Therefore, the inventor has applied an image reconstruction method (for example, a filtered back projection method) which has been mainly used in X-ray CT (computed tomography) technology or the like to a three-dimensional image reconstruction based on an electron microscope image. The application to the law was considered. In the X-ray-CT technique, a conventional image reconstruction method is based on a plurality of two-dimensional projection images obtained by irradiating a sample with X-rays from all directions, and completes a three-dimensional internal structure of the sample. It has been proved mathematically and experimentally that it can be reconstructed as follows.

However, in order to apply this method to an electron microscope image, several conditions must be satisfied.

First, the relationship between the sample thickness through which the electron beam has passed and the image contrast on a two-dimensional projected image must be linear. Specifically, it is necessary to establish a linear relationship such that if the thickness of the sample is doubled, the value of the projection data will be doubled, and if the thickness is the same, the value of the projection data will be the same even if the observation direction is changed. It is. However, the electron beam passing through the crystal is X
This linear relationship is generally not guaranteed because of the complex scattering process followed by lines. Therefore, in three-dimensional atomic array observation using an electron microscope, the presence of projection data (ie, a two-dimensional image) that deviates from this linear relationship causes artifacts on the reconstructed three-dimensional image.

Second, it is necessary to accurately set the inclination angle of the electron beam of the sample with respect to the optical axis to a predetermined value in the housing of the electron microscope. This is because if the tilt angle of the sample is slightly deviated from a desired angle (or an angle to be set), an artifact occurs in the reconstructed image. However, in the electron microscope, it is very difficult to set the tilt axis of the sample in accordance with the design specification, that is, only with the hardware of the sample holder, due to the processing accuracy and operation accuracy of the sample holder. It is virtually impossible to exactly match the angle with the mechanical movement of the sample holder. Therefore, in order to accurately set the tilt angle of the sample, the tilt axis must be analyzed from the projected image and corrected.

Third, the projection images obtained by observation from each direction need to be information based on the same object. That is, in the normal X-ray-CT technique, the entire irradiation region of the X-ray beam on the sample is the observation target region. Therefore, the projection images obtained by the irradiation of the X-ray beam from each direction are all based on the information of the same object. It will be. On the other hand, in the three-dimensional atomic arrangement observation using an electron microscope, for example, a crystal defect or an impurity localized in a sample such as a semiconductor wafer or a device becomes an observation target region. However, the irradiation of the electron beam is performed including a non-observation target region (hereinafter, a peripheral region) surrounding the observation target region in the sample. In the peripheral region, the electron beam irradiated in the same manner as in the observation target region is scattered, so that the projection image obtained by irradiating the observation target region localized in the sample with the electron beam from each direction includes an electron beam. Information on the peripheral region that differs depending on the irradiation direction of the beam (or the inclination angle of the sample with respect to the electron beam) is included. Therefore, in a three-dimensional image constructed based on a plurality of projection images, artifacts occur due to a variation in the information amount of a peripheral region for each projection image.

[0008] Since the software of the conventional image reconstruction method is premised on the use of ordinary X-ray-CT technology, consideration has been given to eliminating the artifacts generated in the three-dimensional atomic arrangement observation with the above-mentioned electron microscope. I didn't.

SUMMARY OF THE INVENTION It is an object of the present invention to provide a method for reconstructing a three-dimensional atomic array structure with high accuracy by eliminating or reducing an obstacle which is an obstacle in reconstructing the three-dimensional image. An object of the present invention is to provide an electron microscope provided with means for processing a two-dimensional image reconstruction.

[0010]

In order to achieve the above object, the present invention provides a method for observing a sample using an electron microscope.
Detection of an electron beam for forming an electron microscope image (two-dimensional projected image) and image processing of the two-dimensional projected image are performed in the following manner.

First, in order to maintain the linearity of the image contrast, of the electrons scattered by the sample when irradiating the sample with an electron beam, the image contrast and the sample thickness have a substantially linear relationship, and the image contrast is high. Forms a two-dimensional projection image with electrons scattered in an angular range depending on the atomic number. More specifically, two-dimensional projection is performed by detecting electrons (so-called high-angle scattered electrons) in a scattering angle range in which the intensity of electrons due to thermal diffuse scattering is more dominant than the intensity of electrons due to elastic scattering. Image the image. For this purpose, an electron beam detector is installed inside the housing of the electron microscope.
It may be provided at a position suitable for detecting high-angle scattered electrons from a sample (sample holder) based on the optical axis of the electron optical system. Also,
When using an electron beam detector (for example, Newvicon or Harpicon) that can arbitrarily set the detection area on the electron detection surface, the detection area is set to a detection angle range suitable for detecting high-angle scattered electrons.

Second, in order to accurately set the tilt angle of the sample with respect to the irradiation electron beam, the analysis of the tilt axis involves rotating the sample by the sample holder and tilting the sample holder itself (mechanical tilt axis). By tilting the sample at several angles around it, at least two foreign substances present in the sample (for example, atoms or atomic clusters of elements different from the element that is the main component of the sample) or a change in the position of a characteristic point such as a crystal defect can be detected. Attention is paid to the positions of these feature points in the two-dimensional projected image. Next, one of the marked feature points is specified, the position is set as the origin, the tilt angle of the sample is sequentially changed based on the mechanical tilt axis, the specific feature point expected with respect to the tilt angle and other The effective tilt axis of the sample (that is, the tilt axis that is the center of the actual sample tilt with respect to the irradiation electron beam) is analyzed based on the difference between the positional relationship with the feature point and the actually observed positional relationship. For example, a figure whose shape is known between a specified feature point and another feature point is imagined in a three-dimensional space, and deformation of the figure in a plurality of projection images (at different inclination angles) used for extracting the feature point is determined. Analyze the tilt axis from the degree.

To facilitate understanding, a specific example will be described. First, two orthogonal coordinate systems for expressing the positional relationship between the above-mentioned feature points, one based on the mechanical configuration of an electron microscope ( Hereinafter, an observation system) and another one based on the above-described effective tilt axis (hereinafter, a sample system) are set. One of the coordinate axes of the observation system is set to or parallel to the optical axis of the electron optical system of the electron microscope, and one of the coordinate axes of the sample system is set to or above the effective tilt axis. The origins of the two coordinate systems may be set to the specific feature points described above. Here, when the sample holder is rotated at an angle of ± α,
Correspondingly, feature points other than the specific feature point move at an angle of ± α around a coordinate axis parallel to the effective tilt axis of the sample system. At this time, the trajectory of the position change of the feature point has an angle α / 2 on a plane perpendicular to the effective tilt axis of the sample system.
Draws an isosceles triangle with two angles, but in an electron microscope image (that is, a plane perpendicular to the optical axis of the electron optics of the observation system), this is true unless the optical axis of the electron optics and the effective tilt axis are parallel. A locus of similar position change due to the feature point does not appear. Therefore, the shape of the trajectory imagined on the plane of the sample system,
The deviation between the mechanical tilt axis (having a fixed positional relationship with the optical axis of the electron optical system) and the effective tilt axis is obtained by associating the shape of the trajectory actually measured on the plane of the observation system.

Thirdly, with regard to the mixing of the peripheral area, the area included or not included in the projection image depending on the sample tilt angle is both ends of the projection data, and this area is a re-projection when the projection data is back-projected. The fact that it exists only in the peripheral part in the constituent image is used. This area is deleted from the projection data and the reconstructed image, and the missing angle information recovery processing is performed only on the area that becomes the reconstructed image having the same structure to restore the cross-sectional structure.

[0015]

First of all, electrons that are high-angle scattered by the sample to be observed as electrons for forming a projected image, that is, elastic scattered electrons, which are subject to interference by the electron optical system and the conditions of the sample thickness, in terms of the intensity of the scattered electrons. By using the electrons scattered at an angle at which the thermally diffuse scattered electrons which do not interfere with the condition (1) are dominant, the influence of these conditions on the image contrast can be eliminated. That is, by using scattered electrons in such an angular range where the image contrast and the sample thickness can be approximated to a linear relationship, the theory of general image reconstruction can be applied to a two-dimensional image formed by an electron beam passing through a crystal. Adaptable and reconfigured 3
It is possible to quantitatively analyze the position and contrast of the different atoms in the three-dimensional structure.

Second, by using a tilt axis analysis method in which a plurality of unique portions present in the sample are marked and the actual tilt of the sample is analyzed from a change in the position of these portions in the electron microscope image with respect to the tilt angle of the sample. The inclination angle of the sample holder set in the electron microscope can correspond to the actual inclination angle of the sample with respect to the irradiation electron beam. That is,
The effective tilt axis for the irradiation electron beam of a sample oriented in any direction in three dimensions, which could not be accurately grasped due to the mechanical accuracy limitations of the sample holder and its tilting device only with the hardware of the electron microscope, Analysis can be based on electron microscope images (images). Therefore, it is possible to almost certainly set the fine sample inclination angle which is essential for the three-dimensional image observation at the atomic level.

Third, if information on an unnecessary area, that is, a peripheral area surrounding a desired observation area is removed from the projection image, a structure which becomes an obstacle to the three-dimensional image reconstruction processing is included around the desired area to be analyzed. However, the structure of interest can be three-dimensionally reconstructed.

The above three methods can be incorporated in the control or processing sequence in a hardware control device or an image processing device of the electron microscope. For example, an electron beam source that emits an electron beam, a first electron optical system that irradiates the sample with the electron beam, a sample holder that holds the sample, and a sample tilting device that tilts the sample holder with respect to the irradiated electron beam. A second electron optical system that forms a projection image with an electron beam emitted from the sample, an electron beam detection device that detects the projection image, and a data processing device that processes data of the projection image of the electron beam detection device. In the electron microscope described above, the setting of the first scattered electron beam detection condition is performed by irradiating a sample with a hollow electron beam (hollow cone beam) by a first optical system or by using an electron having a two-dimensional electron beam detection surface. This can be performed by controlling the setting of the electron beam detection area on the detection surface in the line detection device. In the above-mentioned second tilt axis analysis method, a projection image for each sample tilt angle set by the sample tilt device is processed by the data processing device, and the effective tilt axis obtained from the mechanical tilt axis of the sample tilt device and the projected image is obtained. A control signal is sent from the data processing device to the sample tilting device (feedback is applied) based on the data calculated for the deviation from the typical tilting axis. When the sample tilting device has a tilt axis adjusting device for adjusting the center axis position of the tilt of the sample holder, the data processing device may control the tilt axis adjusting device. Further, the above-described third image processing method may be installed in the data processing device as image processing software.

[0019]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Intuitively, image reconstruction is performed by recording projection data 2 in various directions of a two-dimensional cross-sectional image 1 of an object as shown in FIG. This is processing for reconstructing the constituent image 3. If the two-dimensional reconstructed image 3 and the two-dimensional cross-sectional image 1 match, it means that the original image has been completely restored. The present invention clarifies observation conditions and processing necessary for adapting the image reconstruction method to an actually observed electron microscope image, and provides a method that satisfies the conditions.

FIG. 2 shows the basic structure of the electron microscope used in the embodiment of the present invention. It comprises a field emission type electron gun 4, a condenser lens 5, a beam deflection scanning coil 6, an objective lens 7, a sample fine movement / tilt mechanism 8, an electron beam detector 9, a control and image processing processor 10. FIG. 3 is a flowchart showing a process of constructing a three-dimensional atomic array image. Setting observation conditions for the two-dimensional projection image 13, obtaining a plurality of two-dimensional projection images 13 observed from various directions by inclining the thinned sample 12 around a certain tilt axis 14, A step of specifying the tilt axis 14 from the image 13 by image analysis, a step of setting the tilt axis 14 according to design specifications, and constructing each two-dimensional cross-sectional image 1 orthogonal to the projection image from the plurality of two-dimensional projection images 13 And analyzing the three-dimensional atomic arrangement image from the three-dimensional image constructed by stacking the two-dimensional cross-sectional images 1. The details of each step will be described below.

First, a process for obtaining the two-dimensional projected image 13 will be described. The sample 12 thinned by the electron beam 11 having a thickness of 1 to 2 atoms or less is scanned, and among the electrons scattered by the atoms in the sample, the scattered electron beam 15 having a certain scattering angle range α to β.
Is detected, and the intensity is displayed on a CRT in synchronization with the scanning of the electron beam, and the two-dimensional projected image 13 is observed. The detection angle range is set so as to satisfy the following several conditions. First, in order to set the detection angle range, the scattering process of the electron beam in the sample will be considered. An electron beam passing through the sample 12 is scattered by atoms in the sample 12. The scattered electron dose and the scattering angle generally have a relationship as shown in FIG. It has a distribution having a peak at a certain scattering angle and a broader tail toward the higher scattering angle side, and this distribution shifts toward the higher scattering angle side as the atomic number Z of the atom scatters electrons increases. Therefore, the electron dose in the scattering angle range α to β shown in FIG. Here, as shown in FIG.
It is assumed that the electron beam 11 is incident on the sample 12 having (silver) particles at each position, and the amount of the electron beam 15 in the scattering angle range α to β is measured. When the electron beam 11 enters the position (1), the amount of the electron beam 15 in the scattering angle range α to β is small because the electron beam only passes through the Si crystal. When the electron beam 11 enters the position (2), the amount of the electron beam 15 in the scattering angle range α to β increases because the electron beam passes through Ag atoms having a large atomic number in addition to the Si crystal. When the electron beam 11 enters the position (3), the amount of the electron beam 15 in the scattering angle range α to β increases by the amount of the electron beam passing through two Ag atoms. When the electron beam 11 is incident on the position (4), Si
Since one atom is missing, the amount of the electron beam 15 in the scattering angle range α to β is smaller than that in the position (1). As described above, the amount of the electron beam 15 in the scattering angle range α to β corresponds to the number of atoms that scattered the electron beam and the atomic number Z thereof.

Next, the correspondence between the amount of the electron beam 15 and the image contrast of the two-dimensional image formed by the electron beam will be considered.
The electrons scattered by the sample are divided into coherent elastic scattered electrons and non-coherent heat diffuse scattered electrons. The image contrast does not always correspond to the scattered electron dose because the elastic scattered electron beam causes a slight change in phase, that is, oscillation of the electron beam intensity due to the defocus of the objective lens 7 and the change in the thickness of the sample 12. On the other hand, since the thermal diffuse scattered electron beam does not interfere, the vibration does not occur, and the image contrast and the scattered electron dose correspond. In general, the relationship between the proportion of elastic scattering and thermal diffuse scattering in the scattered electron beam and the scattering angle is as shown in FIG. The amount of elastic scattered electrons has a peak at a low scattering angle position, and sharply decreases as the scattering angle increases. on the other hand,
The amount of thermally diffuse scattered electrons has little scattering angle dependence. For this reason, the heat diffusion diffused electrons occupy most of the high-angle side. Therefore, if the detection angle range α to β of the scattered electron beam is set on the high angle side, that is, several tens mrad or more so that the heat diffuse scattered electron beam occupies most of the angle, an image contrast corresponding to the scattered electron dose can be obtained. . For example, when an electron beam is incident on a Si single crystal from the [110] direction, if the scattering angle range is set to about 50 mrad or more, it is confirmed that the amount of the scattered electron beam 15 and the image contrast have a substantially linear relationship. I have.

The detection angle range is set so as to satisfy the above conditions, and the sample 12 is moved to a certain tilt axis 14 using the sample tilt mechanism 8.
It is rotated around to obtain a plurality of two-dimensional projection images 13 observed from various directions.

Next, a process of specifying the tilt axis 14 from the plurality of two-dimensional projected images 13 by image analysis will be described. First, a coordinate system shown in FIG. 7 is assumed. An orthogonal coordinate system fixed to the observation system is defined as a coordinate system xyz, and an orthogonal coordinate system fixed to the sample 12 is defined as a coordinate system {η}. Coordinate system xyz and coordinate system ξ
The coordinate conversion with ηζ is expressed by the following equation using the Euler angle θψφ.

[0025]

(Equation 1)

Here, if there is a standard sample having a known shape, the value of ({η}) set in the sample system and the value of (xyz) measured in the observation system when the sample is tilted around the tilt axis. The Euler angle θψφ can be obtained from the correspondence. However, a structure whose shape is known is not generally included in the TEM sample. Therefore, an arbitrary feature point in the sample 12, for example, an Ag atom in a Si crystal is used. Two feature points are selected from the sample and analyzed using a vector connecting the two points. As shown in FIG. 8, the inclined axis 14 is a ζ axis,
The sample 12 is rotated around the ζ axis. A vector u when the sample 12 is not rotated, a vector v when the sample 12 is rotated + α around the ζ axis, and a vector w when the sample 12 is rotated −α are assumed. Considering the coordinate system {η}, the vectors u, v, w form an isosceles triangle on the {η plane. Vector s
= V−w is parallel to the ξ axis, and the vector t = u− (v +
w) / 2 is parallel to the η axis. Since the length ratio between the two can be obtained from the rotation angle α, the unit vector of the coordinate system {η}, the vector s ′ = (k, 0, 0), the vector t ′
= (0, k, 0) can be set. A vector s ′ obtained by measuring the coordinates of the vectors u, v, w in the observation system xyz,
The value of t 'and the vectors s', t' in the {η} coordinate system
, The Euler angle θψφ and the length k of the unit vector can be obtained, and the direction of the ζ axis, which is the inclined axis 14, can be analyzed. Note that the observation system observes the projected image from the z direction, so that the z coordinate cannot be measured. However, since four simultaneous equations can be defined by the x, y coordinates of the vector s ′ and the x, y coordinates of the vector t ′, It is possible to specify two variables.

Through the above steps, the tilt axis 14 can be analyzed by image analysis using feature points in the sample without using a sample having a known shape. In order to avoid a reading error or the like, two points in the sample are selected so that the lengths of the vectors s and t are as large as possible. Further, it is desirable to set the rotation angle δ as large as possible so that the lengths of the vectors s and t are not so different. The feature point is desirably a shape that can specify its position even when observed from any direction, for example, fine particles, the top of a rectangular parallelepiped, or the like. If an appropriate feature point does not exist in the sample, a focused electron beam is applied to the square of the region of interest to create a feature point, for example, a defect or a contamination trace.

Next, a process for setting the tilt axis 14 according to the design specification will be described. First, the design specifications of the tilt axis are shown in FIG. When observing the two-dimensional projection image 13 from the z direction, if the tilt axis 14 is parallel to the xy plane that is the projection image plane, the line profile orthogonal to the tilt axis 14 that appears on the two-dimensional projection image 13 is two-dimensional. The one-dimensional projection data 2 of the cross-sectional image 1 is obtained. The tilt axis 14 is set so as to be parallel to the xy plane by using the sample tilt mechanism 8, and it is confirmed whether the direction of the tilt axis satisfies the design specification by using the tilt axis analysis method. If the specification is not satisfied, the direction of the tilt axis is corrected using a sample tilt mechanism.

Next, steps for constructing each two-dimensional cross-sectional image will be described. The process includes the tilt axis 1 set in the design specification direction.
Plural two-dimensional projection images 13 observed obliquely around 4
Collecting one-dimensional projection data of a two-dimensional cross-sectional image orthogonal to the projection image from
Constructing a two-dimensional reconstructed image;
A step of deleting an area where a two-dimensional cross-sectional image cannot be restored;
A process of performing a missing angle information recovery process on the two-dimensional image to obtain a two-dimensional cross-sectional image.

The process of collecting one-dimensional projection data will be described.
First, the geometric relationship between the one-dimensional projection data 2 and the two-dimensional cross-sectional image 1 is shown in FIG. The sample 12 is a flat plate sample having a thickness of 2d. A region of interest 301 is set in the sample 12, and its width is set to 2a. The maximum inclination angle of the sample 12 is ± γ degrees.
First, consider the case where d is sufficiently smaller than a as shown in FIG. The end data 201 of the one-dimensional projection data 2 at the sample inclination angle γ includes projection data of an area other than the attention area 301. Therefore, only the one-dimensional projection data 2 from which both end data 201 has been deleted is used for image reconstruction. Several methods for determining the range of the end data 201 will be described below. The range to be deleted from the geometric relationship is a (1-cosγ) -2d · sinγ from both ends of the data. Even if the thickness 2d of the sample 12 is not known, if a feature point exists around the attention area 301, it can be determined from the position of the feature point on the projection data. If there is no feature point around the attention area 301, a convergent electron beam is applied to the square of the attention area to create a feature point, for example, a defect or a contamination trace, and the position of the feature point on the projection data is calculated. decide. When d and a have substantially the same length as shown in FIG. 10B, projection data of an area other than the attention area 301 is mixed in the entire range of the one-dimensional projection data 2 at the sample inclination angle γ. Thus, it can be seen that the one-dimensional projection data 2 cannot be used for reconstruction processing. Therefore, the width 2 of the attention area 301
a must be set as wide as possible.

Next, the step of back-projecting the one-dimensional projection data to construct a two-dimensional reconstructed image includes a two-dimensional Fourier transform method, a filter-corrected back projection method, and a superposition method used in ordinary X-ray-CT. This is performed by a method such as an integration method.

Next, a step of deleting a region where a two-dimensional cross-sectional image cannot be restored from the two-dimensional reconstructed image will be described. In the two-dimensional reconstructed image at the stage of back-projecting the one-dimensional projection data 2,
In the region of interest 301 shown in FIG. 10A, only the region 302 is back-projected from the projection data of all sample inclination angles. In an area other than the area 302, the information amount of the projection data included in the image is extremely small. In the electron microscope, since the projection angle range is limited, it is necessary to perform appropriate missing angle information recovery processing using known information. However, if this processing is directly applied to the attention area 301,
Not only cannot the cross-sectional image of the area other than the area 302 be recovered, but also an area other than the area 302 may cause an artifact in the recoverable area 302. Therefore, it is necessary to delete a region other than the region 302 and replace it with an appropriate sample structure, for example, a substrate crystal so that no artifact occurs. Regions other than the region 302 can be specified geometrically as regions outside the region when a straight line is drawn in each projection direction from the end of the target structure.

A process for obtaining a two-dimensional cross-sectional image by performing a missing angle information recovery process on the two-dimensional image 3 subjected to the above-described process will be described.
The missing angle information recovery process is performed based on known information regarding the two-dimensional cross-sectional image, for example, information such as probability information and symmetry.

Next, a process of analyzing a three-dimensional atomic arrangement and atomic species from a three-dimensional image constructed by stacking the two-dimensional sectional images 1 will be described. Each two-dimensional cross-sectional image is sequentially reconstructed along the sample tilt axis, and the three-dimensional structure is constructed by stacking them.
Since the image contrast in the two-dimensional projected image 3 is a contrast depending on the number of atoms existing on the path through which the electron beam has passed and the atomic number Z, it is difficult to specify the atomic number Z only from the two-dimensional projected image 3. Met. By analyzing the sample shape by three-dimensional reconstruction, the atomic number Z of the atoms constituting the sample can be specified.

[0035]

According to the present invention, a three-dimensional atomic arrangement image can be formed with high precision from the electron microscope images observed from various observation directions by using a general image reconstruction theory. Thereby, three atoms of impurity atoms and their clusters in the crystal are obtained.
The dimensional structure can be analyzed at the level of one atom. Therefore, the present invention provides effective information for analyzing the cause of the failure of the ULSI element such as the leak current and the breakdown voltage failure.

[0036]

[Brief description of the drawings]

FIGS. 1A and 1B are explanatory diagrams illustrating the principle of image reconstruction by back projection and FIG.

FIG. 2 is an overall configuration diagram of an apparatus used in an embodiment of the present invention.

FIG. 3 is a flowchart showing steps for constructing a three-dimensional atomic array image.

FIG. 4 is an explanatory diagram showing an angular distribution of a scattered electron dose for atoms having different atomic numbers.

FIG. 5 is an explanatory diagram showing an electron dose scattered in a scattering angle range α to β when an electron beam is incident on each sample position.

FIG. 6 is an explanatory diagram showing the scattering angle dependence of the ratio of elastic scattering and thermal diffuse scattering.

FIG. 7 shows a coordinate system xyz, a coordinate system {η}, and an Euler angle θ.
FIG. 4 is an explanatory diagram showing a relationship of ψφ.

FIG. 8 is an explanatory diagram showing each vector used for coordinate axis analysis.

FIG. 9 is an explanatory diagram showing that one-dimensional projection data can be collected from a two-dimensional projection image when the tilt axis is parallel to the two-dimensional projection image.

FIGS. 10A and 10B are explanatory diagrams showing a relationship between one-dimensional projection data and a two-dimensional cross-sectional image when the sample is thin, and FIG.

[Explanation of symbols]

1 ... two-dimensional cross-sectional image, 2 ... one-dimensional projection data, 201 ... 2
One-dimensional projection data unnecessary for three-dimensional cross-sectional image reconstruction, 3 ... 2
Dimensional reconstructed image, 301 ... attention area in two-dimensional cross-sectional image, 3
02: an area from which a two-dimensional cross-sectional image can be recovered, 4: a field emission electron gun, 5: a condenser lens, 6: a beam deflection scanning coil, 7: an objective lens, 8: a sample fine movement / tilt mechanism, 9
... Electron beam detector, 10 ... Processor for control and image processing, 11 ... Electron beam incident, 12 ... Sample, 13 ... 2D projection image, 14 ... Slope axis of sample, 15 ... Scattered in angle range α-β Electron beam.

──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Kensuke Sekihara 1-280 Higashi Koikekubo, Kokubunji City, Tokyo Inside the Central Research Laboratory, Hitachi, Ltd. (72) Inventor Jun Jun Ikego 2-280 Higashi Koikekubo, Kokubunji City, Tokyo Hitachi, Ltd. (56) References JP-A-4-337236 (JP, A) JP-A-2-15545 (JP, A) JP-A-59-4408 (JP, U) (58) Fields surveyed (Int .Cl. ⁷ , DB name) G01N 23/04 H01J 37/22 501 H01J 37/28

Claims (5)

(57) [Claims] An electron source for emitting an electron beam, and an electron beam for a sample
A first electron optical system for irradiating a predetermined area of a sample holder for holding a sample, a sample tilting device for tilting the sample holder with respect to the electron beam for irradiating a sample on the sample holder, transmitted through the sample A second electron optical system that forms a projected image with an electron beam, an electron beam detecting device that detects the projected image, and a plurality of electron beam detecting devices that detect the projected image according to a rotation angle of the sample holder.
Processing the data of the projected image of the sample, the position of the obtained image and the sample
Data processing for reconstructing a three-dimensional image of the sample from the angle of the holder
An electron microscope comprising: a processing device . 2. The electron microscope according to claim 1, further comprising a tilt axis adjusting device for adjusting a center axis position of the tilt of the sample holder as the sample tilt device. 3. A step of mounting a thinned sample on a sample holder, a step of irradiating the sample with an electron beam from an electron source, and a step of tilting the sample to acquire a transmission electron image at a plurality of tilt angles. Calculating the rotation center position of the sample holder from the feature amount in the image from the image obtained in the obtaining step, and obtaining the plurality of obtained images from the angles of the sample holder and the obtained rotation center at the time of obtaining the plurality of images. Reconstructing a three-dimensional structure of the sample from the position. The method according to claim 4, wherein the feature amount, the three-dimensional atomic arrangement observation method according to claim 3 characterized by using any of Ag or defects or contamination remains in the Si crystal. 5. A three-dimensional atomic arrangement observation method according to claim 3, characterized in that by adding the step of deleting the region in which the structure other than the feature quantity is mixed from the two-dimensional projection image.