NL2006300C2

NL2006300C2 - Electron microscope and method of estimating defocus therefor.

Info

Publication number: NL2006300C2
Application number: NL2006300A
Authority: NL
Inventors: Ruiz Arturo Tejada; Arnold Jan Dekker
Original assignee: Univ Delft Tech
Priority date: 2011-02-25
Filing date: 2011-02-25
Publication date: 2012-08-28

Description

P30630NL00/MVE

Electron microscope and method of estimating defocus therefor

The invention relates to an electron microscope configured to produce an image of an amorphous specimen by illuminating said specimen with a particle beam of electrons and detecting electrons scattered through the specimen. The invention further relates to a method for estimating the defocus of an image of an amorphous specimen in an electron 5 microscope according to the invention.

Electron microscopes use an electron source to produce a particle beam of electrons, i.e. an electron beam, which illuminates the specimen. Upon illumination the particle beam of electrons will interact with the specimen, wherein the result of the interaction can be 10 detected to produce an image. The main advantage of an electron microscope over light-powered optical microscopes is the greater resolving power due to the smaller wavelengths of the electrons compared to visible light.

Because of this greater resolving power, electron microscopes are very popular tools for 15 academic research in material sciences, nanotechnology, and biology. They are also highly valued in, among others, the semiconductor industry, where they are used for production monitoring, control and troubleshooting.

When a specimen is illuminated by an electron beam, various types of interaction may 20 occur.

In a first type of interaction, the specimen allows at least partially the transmission of the electron beam, e.g. because the specimen is thin enough. During passing of the electron beam through the specimen, a portion of the electrons is scattered by the specimen, so that 25 the electron beam after passing carries information about the structure of the specimen which can be detected to form an image.

In a second type of interaction, at least a portion of the electron beam is reflected of the specimen. The so-called reflected beam of elastically scattered electrons can be detected to 30 produce an image.

-2-

In a third type of interaction, electrons of the electron beam lose energy during passing of the electron beam which energy is converted into for example heat, emission of secondary electrons, light emission or x-ray emission. By detecting one or more of these converted energy forms, an image can be produced.

5

Although the different types of interaction are described in isolation, it is known to a person skilled in the art of electron microscopy that one or more types of interaction may occur at the same time, but that depending on the energy of the electrons in the electron beam and the material properties and dimensions of the specimen, one type of interaction may 10 dominate and is therefore a suitable candidate to produce an image from.

Electron microscopes come in different types which are at least partially related to the different types of interaction described above.

15 One type of electron microscope is the transmission electron microscope (TEM) in which an electron beam is used to illuminate an entire imaging area on the specimen at once and detect the transmitted electron beam by projecting the electron image onto a fluorescent viewing screen, the image of which can be recorded by for instance a CCD camera or other suitable device. A TEM is used mainly to produce bright-field images.

20

Another type of electron microscope is the scanning transmission electron microscope (STEM) in which the electron beam is focused into a small probe at the specimen. The probe is scanned along the imaging area on the specimen and at each spatial location, the transmitted electrons are detected. This type of microscope allows to distinguish between 25 electrons that are scattered at small angles which form the basis of a bright-field image and electrons that are scattered at large angles which form the basis of a dark-field image.

Other types of electron microscopes are the scanning electron microscope (SEM) which scans the imaging area on the specimen in a similar way as a STEM, but in most cases uses 30 secondary electrons emitted by the specimen to produce an image, and the reflection electron microscope (REM) which uses elastically scattered electrons that are reflected from the specimen to produce an image.

The invention relates to electron microscopes which detect the transmission of electrons 35 through a specimen to produce images, such as the TEM and STEM. From now on, if reference is made to an electron microscope, reference is made to an electron microscope configured to produce an image of an amorphous specimen by illuminating said specimen -3- with a particle beam of electrons and detecting the electrons scattered through said specimen unless specifically stated otherwise.

It is noted that it is known to a person skilled in the art of electron microscopes that it is 5 possible that an electron microscope can be used in different ways and thus may be used as a TEM or STEM without having to use a different device, but merely by changing the operating mode of the electron microscope.

Common to all electron microscopes (in fact this also applies to optical microscopes) is that 10 the microscope’s optical properties need to be adjusted to ensure a certain level of image quality. The optical properties include spherical aberration, astigmatism and defocus. Since these properties can only be observed from image data, their manual calibration requires the electron microscope’s operators to repeatedly acquire and visually interpret images and to adjust the electron microscope’s controls accordingly.

15

Automated calibration has been proposed to speed up the process of adjusting the optical properties of electron microscopes. An example of such a calibration method is for instance disclosed in US patent application US 2010/0032565 A1.

20 In a standard defocus estimation approach, a 2D bright-field image of a thin amorphous specimen is transformed from the 2D spatial domain to the 2D frequency domain by computing the modulus squared of its Fourier transform. The radial profile of the resulting image (in frequency domain) is estimated through radial averaging. By finding the position of the local minima in the estimated radial profile, i.e. fitting the radial profile to a theoretical 25 model of the optics of the electron microscope having defocus as a parameter, the defocus can be estimated. The estimated defocus can subsequently be used to adjust the optics of the electron microscope in order to set the defocus to a value that, for example, enhances the image quality or that maximizes the image information content.

30 A disadvantage of the disclosed method is that the transformation from the 2D spatial domain to the 2D frequency domain and subsequently fitting the radial profile to the theoretical model is challenging for instance due to noise present in the image data. Another disadvantage may be that computing the radial average may take long as interpolation is needed to cope with the fact that the image data is usually arranged in a rectangular array of 35 pixels.

-4-

An object of the invention is therefore to provide an electron microscope configured to produce an image of an amorphous specimen by illuminating said specimen with a particle beam of electrons and detecting the electrons scattered through said specimen, wherein said electron microscope has an improved estimator to estimate the defocus.

5

This object is achieved by an electron microscope for producing an image of an amorphous specimen by illuminating said specimen with a particle beam of electrons and detecting the electrons scattered through said specimen, comprising: - an electron source to produce a particle beam of electrons; 10 - a detector to detect said electrons scattered through the specimen; - a display device; - a processing unit configured to gather image data from an output of the detector to display a corresponding image of the specimen on the display device; 15 - an electron optical system arranged along an optical axis extending between the electron source and the detector to manipulate the particle beam of electrons for illuminating said specimen with the particle beam of electrons and directing the electrons scattered through said specimen towards the detector, wherein the electron optical system comprises at least one electron optical 20 component to focus the image and a stigmator to correct for astigmatism in the electron optical system; - a specimen holder arranged between the electron source and the detector to hold said specimen; and - an estimator to estimate the amount of defocus of the image, said estimator 25 comprising a theoretical optics model based on said at least one electron optical component, wherein the theoretical optics model is dependent on the defocus, and wherein said estimator is configured to: a. obtain bright-field image data of said specimen from the processing unit along multiple radial image lines around the optical axis, wherein 30 each radial image line has a different angular orientation relative to the optical axis, b. autocorrelate the image data of each radial image line, c. average the autocorrelated image data of the radial image lines, and d. fit the theoretical optics model to the averaged autocorrelated image 35 data in order to estimate the defocus of the image.

-5-

Autocorrelating the image data has the advantage that it automatically minimizes the influence of noise present in the image data, which makes the subsequent fitting process less challenging and thus easier.

5 To obtain the theoretical optics model, use can be made of the radial profile of the Fourier transform of the point spread function of the at least one electron optical component to focus the image. Said radial profile can subsequently be squared, and by taking the inverse-zero-order Hankel transform of the squared radial profile the result is suitable to be compared to the obtained averaged autocorrelated image data.

10

Said radial profile is usually a function of at least two variables including spatial frequency and defocus. By determining the radial profile for different values of the defocus and subsequently squaring and transforming each radial profile as described above, a collection of 1D functions is created which can easily be compared to the 1D averaged autocorrelated 15 image data. The value of the defocus is in such a case determined by checking which 1D function comes closest to the averaged autocorrelated image data.

To check which 1D function comes closest to the averaged autocorrelated image data, a known technique such as the least squares method can be used.

20

In the above described method of obtaining the theoretical optics model, it is assumed that astigmatism in the electron optics system is removed beforehand using the stigmator, either manually or using an automated calibration method.

25 The abovementioned radial profile may also be a function of spherical aberration, so that for obtaining the collection of 1D functions representing the theoretical optics model, the spherical aberration may first be determined and used as input to the theoretical optics model. However, it is also possible to determine different radial profiles for different combinations of defocus and spherical aberration and subsequently square and transform 30 the radial profiles as described above to form a collection of 1D functions, wherein each function belongs to a different combination of defocus and spherical aberration which functions can be compared to the averaged autocorrelated image data to estimate both the defocus and the spherical aberration at the same time.

35 In an embodiment, the electron microscope is a STEM, wherein the at least one electron optical component is configured to focus the particle beam of electrons into a small probe at the specimen, and wherein the electron optical system further comprises deflector -6- components drivable by the processing unit to deflect the particle beam of electrons in order to scan the small probe along the specimen, and wherein the estimator is configured to obtain the bright-field image data along the multiple radial image lines by letting the processing unit scan the small probe along radial lines around the optical axis at the 5 specimen. As a result no full image is required, which saves a lot of time compared to prior art estimators which require the full images. As an example, obtaining a full image of 512 by 512 pixels, with a typical pixel-dwell time of 20 microseconds takes about 5.2 seconds. If this can be reduced to 256 pixels per radial image line and e.g. 100 radial image lines, the image data is obtained about ten times faster. When less radial image lines or less pixels per radial 10 image line are used, the advantage can be further increased.

Another advantage is that no interpolation is required to obtain the bright-field image data along the multiple radial image lines which is required when a rectangular scan pattern is used. A further advantage is that both bright-field images and dark-field images can be 15 obtained substantially without having to alter the optical properties of the electron optical system. This allows for instance to alternately obtain a bright-field image and a dark-field image with a relatively high frequency, wherein the bright-field images are used to estimate the defocus and the dark-field images are used by the operator in a normal way for analysis reasons. It may even be possible to obtain a bright-field image and a dark-field image at the 20 same time.

In an embodiment, the electron optical system of the electron microscope comprises an aperture that limits the amount of electrons in the particle beam of electrons that is directed towards the specimen, wherein a smaller aperture corresponds to less electrons in the 25 particle beam of electrons. The aperture then acts as a filter. It has been found that smaller apertures limit the radial image lines’ bandwidth which in turn leads to less oscillating line autocorrelation functions which can be more easily estimated and subsequently more easily compared with the theoretical optics model. In a STEM, the aperture can be set to the same values used to obtain dark-field images.

30

In an embodiment, the electron microscope is a TEM. When the TEM has a detection device such as a CCD camera outputting a nxm array of pixels, the bright-field image data along the multiple radial image lines will have to be computed using interpolation as not all pixels will be exactly on a specific radial image line.

35 -7-

To minimize the interpolation required in a TEM for obtaining the image data along the radial image lines, a detection device can be used in which the pixels are oriented in a radial manner.

5 The invention also relates to a method to estimate the defocus of an image of an amorphous specimen in an electron microscope configured to produce said image by illuminating said specimen with a particle beam of electrons and detecting the electrons scattered through said specimen, said method comprising the following steps: a. removing astigmatism in the image by adapting an electron optical system of 10 the electron microscope; b. subsequently obtaining bright-field image data of said specimen along multiple radial image lines around the optical axis, wherein each radial image line has a different angular orientation relative to the optical axis; c. autocorrelating the image data of each radial image line; 15 d. averaging the autocorrelated image data of all radial image lines; e. providing a theoretical optics model based on the electron optical system, wherein the theoretical optics model is dependent on the defocus; f. fitting the theoretical optics model to the averaged autocorrelated image data to estimate the defocus of the image.

20

In an embodiment, the bright-field image data along said multiple radial image lines are obtained by scanning a small probe of a STEM along said multiple radial image lines.

In an embodiment, the bright-field image data along said multiple radial image lines are 25 obtained by obtaining bright field image data in a rectangular pattern and subsequently interpolating said image data to obtain the image data along said multiple radial image lines.

In an embodiment, the theoretical optics model may also be depending on the spherical aberration in the electron optical system and fitting the theoretical optics model to the 30 averaged autocorrelated image data thus also allows to estimate the spherical aberration at the same time as the defocus. Alternatively, the spherical aberration may be determined or estimated separately and be used as input to the theoretical optics model such that the theoretical optics model is only depending on the defocus.

35 In an embodiment, an aperture of the electron optical system configured to set the amount of electrons in the particle beam of electrons is set to a value which limits the bandwidth of -8- the radial image lines. Preferably, the aperture is set to the same value as preferred to obtain dark-field images.

The invention will now be described in a non-limiting way by reference to the accompanying 5 drawings in which like parts are indicated by like reference numerals, and in which:

Fig. 1 depicts a schematic cross section of a STEM according to an embodiment of the invention;

Fig. 2 depicts a flow chart of a method to estimate the defocus according to an embodiment of the invention; 10 Fig. 3 depicts schematically image data including the multiple radial image lines for which image data needs to be obtained for the estimator of an electron microscope according to the invention;

Fig. 4 depicts a possible scan pattern for a STEM according to the invention.

15 Fig. 1 depicts schematically a cross section of a STEM according to an embodiment of the invention which comprises an electron source ES to produce a particle beam of electrons, i.e. an electron beam PB, and a detector DE to detect electrons scattered through an amorphous specimen SP which is held in a specimen holder SH arranged between the electron source ES and the detector DE.

20

Between the electron source ES and the detector DE an optical axis OA can be defined along which an electron optical system is arranged to manipulate the electron beam PB of electrons for illuminating said specimen SP with the electron beam PB and directing the electrons scattered through said specimen SP towards the detector DE.

25

In this embodiment, said electron optical system comprises an aperture AP to determine the amount of electrons in the electron beam PB, a stigmator AS to correct for astigmatism in the electron optical system, deflector components DC to deflect the electron beam PB, a first electron optical component FC1 to focus the electron beam PB into a small probe at the 30 specimen SP, a second electron optical component FC2 to collect the electron beam after passing through the specimen, and further electron optical components EOC to direct the electron beam towards the detector DE. The deflector components DC are driven by a deflector drive unit DU, and the first and second electron optical components FC1, FC2 are driven by a focus unit FU. The deflector components DC, the first and second electron 35 optical components FC1, FC2, and the further electron optical components EOC may be provided in the form of electrostatic or electromagnetic components. It will be apparent to a person skilled in the art that additional electron optical components need to be provided for -9- a proper functioning of the electron microscope, but these components are not shown because they are not relevant for a clear understanding of the invention and well known to the skilled person.

5 In Fig. 1, the shown electron beam PB corresponds to electrons that are emitted by the electron source ES, pass through the specimen and are not scattered by it, i.e. scattered with zero angle, or scattered by a small angle. Not shown are the electrons that are absorbed by the specimen, reflected by the specimen or scattered by the specimen at a larger angle.

10

The detector DE comprises two sensor parts, namely a bright-field sensor part BFD and a dark-field sensor part DFD. The electrons that are scattered at small angles by the specimen are detected by the bright-field sensor part BFD, which usually has a disc-shape, and the electrons that are scattered at relatively large angles are detected by the dark-field sensor 15 part DFD which usually has a ring-shape as a result of which two portions can be seen in the cross section of Fig. 1.

The outputs of the bright-field and dark-field sensor parts BFD,DFD are connected to a processing unit PU which gathers the image data to display a dark-field image or a bright-20 field image on a display device DD.

In Fig. 1 it is shown that the first electron optical component FC1 focuses the electron beam PB into a small probe at the specimen SP, but that a corresponding focal point FP is not correctly located at the specimen SP. This means that the image obtained in this situation 25 will be out of focus and the defocus is indicated by DF. In order to increase the image quality, the focal point FP should be located closer to the specimen SP.

In the situation of Fig. 1, only image data can be obtained at the location where the small probe is located relative to the specimen SP. By driving the deflector components DC via the 30 deflector drive unit DU which communicates with the processing unit PU, the small probe can be scanned along an entire imaging area of the specimen SP and the image data obtained therefrom can be correctly linked to the corresponding location of the small probe, thereby producing a full image of the specimen.

35 The STEM further comprises an astigmatism remover AR which is configured to receive image data from the processing unit PU, estimate the amount of astigmatism in the electron optical system, and subsequently drive the stigmatorto remove the astigmatism, here - 10- indicated by an arrow between the astigmatism remover AR and the stigmator AS. Alternatively or additionally, the stigmator may also be manually driven, so that an operator or user is able to manually remove the astigmatism. The advantage of the astigmatism remover is that the astigmatism can be removed in an automated way instead of manually.

5 A part of the functioning of an electron microscope according to the invention, for example the STEM according to Fig. 1, can be visualized by a flow diagram as depicted in Fig. 2. The flow starts with step 101 in which image data is obtained for estimating the astigmatism in the electron optical system. This is done using the processing unit PU of Fig. 1 as described 10 above. Subsequently, the amount of astigmatism is estimated in step 102 by the astigmatism remover AR or by a user. When the amount of astigmatism is known, the astigmatism remover or a user is able to determine the settings of the stigmator AS to remove the astigmatism. Actual removal (step 103) is performed by driving the stigmator according to the determined settings, either manually or using the astigmatism remover. It 15 may happen that the astigmatism is not removed at once, but that steps 101 to 103 may have to be repeated until the astigmatism is below a predetermined value.

As astigmatism is a second order optical property of the electron optical system, no further adaptations are required when subsequently using the electron microscope within for 20 instance an hour. If necessary, the steps 101 to 103 can be performed once in a while in order to keep the astigmatism at a minimal level. It is recommended to perform the steps 101 to 103 at least at the start of use of the electron microscope as indicated by the flow diagram of Fig. 2.

25 After removing the astigmatism, bright-field image data is obtained along multiple radial image lines (step 104). This can be done in different ways. Fig. 3 shows an image field obtained using a rectangular scan pattern or in case the microscope is a TEM using a detector, e.g. a CCD camera, with pixels arranged in a rectangular array. The pixels in Fig. 3 are depicted as square blocks. A virtual optical axis OA can be drawn in this image field 30 corresponding to the actual location of the optical axis with respect to the image taken.

The radial image lines RIL are also indicated in Fig. 3 and are straight lines each having a different angular orientation with respect to the optical axis OA. A radial image line RIL starts at the optical axis and extends away from the optical axis.

35

As can be clearly seen in Fig. 3, not all pixels are located exactly on a radial image line RIL, so that for obtaining image data along a radial image line, interpolation is required to obtain the image data at the location of the radial image line. For computational reasons, it is also convenient if the amount of image data per radial image line is the same for each radial image line.

- 11 - 5 When the electron microscope is a STEM or operating in STEM mode as in Fig. 1, it is also possible to obtain the image data along the radial image lines by scanning along the radial image lines, thereby avoiding the interpolation as all image data is directly obtained at the location of the radial image lines. A possible scan pattern is shown in Fig. 4.

10 In the scan pattern of Fig. 4, the scan starts with scanning along radial image line ΙΘ1, running through the optical axis OA and continuing to scan along radial image line ΙΘ2, which makes it easy from control point of view as scanning along a straight line through the optical axis thus obtains image data along two radial image lines. The small probe of the STEM is then angularly displaced as indicated by arrow A1 and a second scan is made along a 15 straight line through the optical axis, thereby obtaining image data along radial image lines ΙΘ3 and ΙΘ4. The small probe is further angularly displaced as indicated by arrow A2 and image data is obtained along radial image lines ΙΘ5 and ΙΘ6. The small probe undergoes a third angular displacement as indicated by arrow A3 and in a similar fashion image data is obtained along radial image lines ΙΘ7 and ΙΘ8. It will be apparent to the person skilled in the 20 art of STEMs that this scan pattern can be executed as long as necessary thereby continuously obtaining image data along radial image lines.

An advantage of the scan pattern of Fig. 4 may be that the specimen is less exposed to radiation of the electron beam and thus damage to the specimen may be reduced compared 25 to obtaining full images for defocus estimation as used in prior art calibration methods.

Referring to Fig. 1 again, the image data measured by the detector DE is send to the processing unit PU. The bright-field image data is transmitted to a estimator DES. In case the image data is obtained by scanning along radial image lines as shown in Fig. 4, the 30 estimator can continue to perform the steps 105 to 108. In case image data is provided in the form as shown in Fig. 3, the estimator DES first has to interpolate the data in order to obtain the image data along the multiple radial image lines before continuing with the steps 105 to 108.

35 In step 105, the autocorrelation function of the image data of each individual radial image line is estimated, usually referred to as autocorrelating the image data. The individual - 12- autocorrelated image data of the radial image lines are combined to be averaged over the multiple radial image lines in step 106.

Provided to the estimator DES is a theoretical optics model. This can for instance be done 5 by estimating the point spread function of the first electron optical component FC1 of the electron microscope of Fig. 1, subject this function to the Fourier transform and determine the radial profile therefrom, which can easily be done as due to the absence of astigmatism, the Fourier transform of the point spread function of the first electron optical component is substantially circular symmetric. It is possible to obtain different radial profiles, wherein each 10 radial profile belongs to a particular value of defocus, which radial profiles can be squared and subsequently transformed using the inverse, zero-order Hankel transform to obtain 1D functions which can be compared to the averaged autocorrelated image data.

By fitting the theoretical optics model to the averaged autocorrelated image data in step 107, 15 e.g. checking which 1D function comes closest to the averaged autocorrelated image data using the least squares method, an estimate of the defocus can be extracted from the fitting in step 108. The estimate of the defocus in turn can be used to accordingly drive the focus unit (see Fig. 1) to adjust the focus in step 109 and for instance minimize the defocus thereby increasing the image quality. After adjustment, the estimation can start again at step 20 104.

In obtaining the theoretical optics model, the spherical aberration may separately be determined and be used as input to the model. However, it is also possible to obtain a theoretical model which is both dependent on defocus and spherical aberration, so that 25 fitting the model to the averaged autocorrelated image data also allows to estimate the spherical aberration at the same time as the defocus, which estimation of the spherical aberration in turn can also be used to improve the image quality.

Parallel to the steps 104 to 109 or in between step 109 and 101, so in between two 30 estimations, a dark-field image can be obtained to be used for analysis purposes as is common to electron microscopes.

It is noted that any number of radial image lines can be used to estimate the defocus, e.g. 20, 100 or 200, but it has been found that the more radial image lines are used, the more 35 accurate the estimation of the defocus becomes.

-13-

It is possible to obtain image data along for instance 20 radial image lines, estimate the defocus, and subsequently obtain new image data along the 20 radial image lines for a next estimation. When the image data is obtained by a scan pattern as indicated in Fig. 4, it is possible to increase the frequency of estimation by estimating the defocus after each new 5 scan along a radial image line.

Suppose the defocus is estimated using image data along four radial image lines using a scan pattern according to Fig. 4. A first estimation can then be done after scanning radial image line ΙΘ4. Subsequently image data is obtained of radial image line ΙΘ5. It is now 10 possible to estimate the defocus again using the image data of radial image lines ΙΘ2 to ΙΘ5 or to increase the accuracy of the earlier estimation by using the image data of radial image lines ΙΘ1 to ΙΘ5. The same can be done after obtaining the image data along radial image line ΙΘ6. A new estimation may use the image data long radial image lines ΙΘ3 to ΙΘ6 or a more accurate estimation can be made using the image data along radial image lines ΙΘ1 to 15 ΙΘ6. In this example, a new estimation is made after each next scan of a radial image line, but it could well be done after each two or more scans.

The method according to the invention as for instance depicted in Fig. 2 may in part be implemented in software designed to execute the steps of the method and/or may in part be 20 implemented in hardware. It is not essential for the invention whether method steps are carried out in hardware or in software.

Claims

An electron microscope for taking an image of an amorphous sample by illuminating the sample with a beam of electron particles and detecting the electrons scattered by the sample, comprising: - an electron source for producing a beam of electron particles ; 10. a detector for detecting electrons scattered by the sample; - a screen; - a processing unit adapted to collect image data from an output of the detector to display a corresponding image of the sample on the screen; - an electron-optical system arranged along an optical axis extending between the electron source and the detector for manipulating the beam of electron particles for illuminating the sample with the beam of electron particles and guiding the electrons scattered by the sample 20 towards the detector, wherein the electron-optical system comprises at least one electron-optical component for focusing the image and a stigmator to compensate for astigmatism in the electron-optical system; - a sample holder arranged between the electron source and the detector 25 for holding the sample; and - an estimator for estimating the amount of blur of the image, the estimator having a theoretical optical model based on the at least one electron-optical component for focusing the image, the theoretical optical model 30 depending on the blur, and wherein the estimator is arranged for: a) obtaining bright-field image data from the processing unit sample along a plurality of radial image lines around the optical axis, each image line having a different angular orientation with respect to the optical axis, b ) autocorrelating the image data of each radial image line, c) averaging the autocorrelated image data of the radial image lines, and d) fitting the theoretical optical model to the average autocorrelated image data to estimate the blur of the image.

An electron microscope according to claim 1, wherein the electron microscope 5 is a STEM, wherein the at least one electron-optical component is arranged to focus the beam of electron particles in a small probe at the sample, and wherein the electron-optical system further comprises deflection components that control can be deflected by the processing unit around the beam of electron particles for scanning the small probe along the sample, the estimator being arranged to obtain the bright-field image data along the multiple radial image lines through the processing unit at the small probe have the sample scanned along radial lines around the optical axis.

An electron microscope according to claim 1, wherein the electron microscope 15 is a TEM.

An electron microscope according to claim 3, wherein the pixels of the detector are arranged radially around the optical axis.

5. An electron microscope according to any one of claims 1-4, wherein the theoretical optical model is dependent on the blur and spherical aberration, and wherein the estimator is adapted to fit the theoretical optical model to the average autocorrelated image data to simultaneously to estimate both blur and spherical aberration.

6. A method for estimating the blurring of an image of an amorphous sample in an electron microscope adapted to take the image by illuminating the sample with a beam of electron particles and detecting the electrons scattered by the sample, the method comprising the steps of: a. removing astigmatism in the image by adjusting an electron-optical system from the electron microscope; 30 b. subsequently obtaining bright-field image data from the sample along a plurality of radial image lines around the optical axis, each radial image line having a different angular orientation relative to the optical axis; c. autocorrelating the image data of each radial image line; 35 d. averaging the autocorrelated image data of all radial image lines; e. providing a theoretical optical model based on the electron-optical system, wherein the theoretical optical model is dependent on the blur; f. fitting the theoretical optical model to the average 5 autocorrelated image data to estimate the blurriness of the image.

A method according to claim 6, wherein the bright-field image data along the multiple radial image line is obtained by scanning a small probe of a STEM along the multiple radial image lines.

A method according to claim 6, wherein the bright-field image data along the plurality of radial image lines are obtained by obtaining bright-field image data in a rectangular pattern and then interpolating said image data to obtain the image data along the plurality of radial image lines .

A method according to any one of claims 6-8, wherein step e) is replaced by providing a theoretical optical model based on the electron-optical system, wherein the theoretical model is dependent on the blur and spherical aberration of the electron-optical system, and wherein step f) is replaced by fitting the theoretical optical model 20 to the average autocorrelated image data to simultaneously estimate both the blur and the spherical aberration.