阅读说明：本技术 确定质量粒子束的波前的方法和设备 (Method and device for determining the wave front of a mass particle beam ) 是由 J.韦尔特 M.鲍尔 于 2020-04-01 设计创作，主要内容包括：本申请涉及一种用于确定质量粒子束(225、510、1910)的波前(550)的方法(3300)和设备(200),其包括下列步骤：(a)在不同记录条件(315、325)下使用质量粒子束(225、510)记录(3320)参考结构(130)的两个或多个图像(310-390)；(b)以参考结构(130)的修改的参考图像(480)产生(3330)两个或多个记录的图像(310-390)的点扩散函数(1750)；以及(c)为了确定所述波前(550),基于产生的点扩散函数(1750)和不同记录条件(315、325)执行(3340)质量粒子束(225、510)的相位重建。(The application relates to a method (3300) and a device (200) for determining a wavefront (550) of a mass particle beam (225, 510, 1910), comprising the following steps: (a) recording (3320) two or more images (310-390) of the reference structure (130) using the mass particle beam (225, 510) under different recording conditions (315, 325); (b) generating (3330) a point spread function (1750) for two or more recorded images (310) with the modified reference image (480) of the reference structure (130); and (c) performing (3340) a phase reconstruction of the mass particle beam (225, 510) based on the generated point spread function (1750) and the different recording conditions (315, 325) for determining the wavefront (550).)

1. Method (3300) for determining a wavefront (550) of a mass particle beam (225, 510, 1910), the method (3300) comprising the steps of:

a. recording (3320) two or more images (310-390) of a reference structure (130) using the mass particle beam (225, 510) under different recording conditions (315, 325);

b. generating (3330) a point spread function (1750) for two or more recorded images (310) with the modified reference image (480) of the reference structure (130); and

c. to determine the wavefront (550), a phase reconstruction of the mass particle beam (225, 510) is performed (3340) on the basis of the generated point spread function (1750) and the different recording conditions (315, 325).

2. The method (3300) of claim 1, wherein the different recording conditions (315, 325) comprise different parameter settings of the source (205) of the mass particle beam (225, 510) and/or of the imaging system (220) and/or of a detection device (230, 240) which records the image (310) and 390).

3. The method (3300) according to any of the preceding claims, wherein the different recording conditions (315, 325) comprise different focus settings (315, 325) of the mass particle beam (225, 510) when recording the two or more images (310-390).

4. The method (3300) according to any of the preceding claims, wherein a reference image (450) of the reference structure (130) represents at least one recording of the reference structure (130) with the mass particle beam (225, 510), wherein the reference structure (130) is arranged in a focal spot of the mass particle beam (225, 510).

5. The method (3300) of any of the preceding claims, wherein the modified reference image (480) of the reference structure (130) substantially corrects the artifacts (430, 440, 460) when recording the two or more images (310-390) of the reference structure (130).

6. The method (3300) of claim 5, wherein the artifact (430, 440, 460) is caused by: when imaging the reference structure (130) with the mass particle beam (225, 510), electrostatic charging of the reference structure (130) by the mass particle beam (225, 510) and/or at least one edge effect (460) of at least one edge (470) of the reference structure (130) in the two or more recorded images (310-390).

7. The method (3300) of claim 5 or 6, wherein correcting the at least one artifact (430, 440, 460) comprises: determining an effect of electrostatic charging of the reference structure (130) and/or an effect of at least one edge effect (460) in the two or more recorded images (310- & 390).

8. The method (3300) of claim 7, wherein determining the effect of the electrostatic charging (420) of the reference structure (130) and/or the effect of the at least one edge effect (460) comprises: simulating an electrostatic charging (420) and/or the at least one edge effect (460) of the reference structure (130).

9. A method (3300) according to any of claims 5-8, wherein correcting the at least one artifact (430, 440, 460) comprises modifying a reference image (450) of the reference structure (130).

10. The method (3300) of any preceding claim, wherein generating the point spread function (1750) comprises: deconvolving the two or more recorded images (310-390) with the modified reference image (480) of the reference structure (130).

11. The method (3300) of any of claims 1-4, wherein the modified reference picture (480) corresponds to an unmodified reference picture (450).

12. The method (3300) according to any preceding claim, further comprising the steps of: providing the reference structure (130) and/or characterizing the reference structure (130) using the mass particle beam (225, 510).

13. The method (3300) according to any of the preceding claims, wherein the reference structure (130) comprises at least one acicular material arrangement, which is arranged on a substrate (120), and wherein the acicular material arrangement and the substrate (120) have different material compositions.

14. The method (3300) according to any of the preceding claims, wherein the reference structure (130) comprises at least one sharp edge (470) and/or at least one defined side wall angle.

15. The method (3300) of any of claims 1-8, wherein generating the modified reference image (480) comprises: two or more reference images (450) are recorded at different kinetic energies of the mass particle beam (225, 510) under best possible recording conditions, and the modified reference image (480) is replaced by a combination of reference images (450) recorded at different energies.

16. The method (3300) according to any preceding claim, further comprising the steps of: modifying the determined wavefront (550) of the mass particle beam (225, 510) such that the altered wavefront (650) substantially corresponds to the prescribed wavefront (540).

17. Apparatus (200) for determining a wavefront (550) of a mass particle beam (225, 510), comprising:

a. recording means for recording (255) two or more images (310-;

b. generating means for generating (265, 270) a point spread function (1750) for two or more recorded images (310-390) from the modified reference image (480) of the reference structure (130); and

c. execution means for performing (270) a phase reconstruction of the mass particle beam (225, 510, 1910) based on the generated point spread function (1750) and the different recording conditions (315, 325) for determining the wavefront (550).

18. The apparatus (200) according to claim 17, wherein the apparatus (200) is implemented to record (255) a reference image (450) of the reference structure (130).

19. The apparatus (200) according to claim 17 or 18, wherein the apparatus (200) comprises an adjustment option for adapting the determined wavefront (550) of the mass particle beam (225, 510) to a specific wavefront (540).

20. A computer program comprising instructions which, when executed by a computer system (250) of an apparatus (200) according to any of claims 17 to 19, cause the computer system (250) to carry out the method steps according to any of claims 1 to 16.

Technical Field

The present invention relates to a method and apparatus for determining the wavefront of a mass particle beam, such as an electron beam.

Background

Advances in nanotechnology have allowed the production of smaller and smaller components of structural elements. In order to process and display nanostructures, tools are needed that can image these structures, so that a true image of such structures can be generated from the measurement data.

Microscopes are powerful tools for imaging nanostructures. In a microscope, a particle beam typically interacts with a sample to be analyzed or processed. Microscopes can be divided into two categories. Optical or light microscopy images a sample with photons. This type of microscope is used to image microstructures in a number of different ways. The resolution of an optical microscope is limited by the wavelength of the light source used to expose the sample to be examined and the numerical aperture of the optical elements that image the sample due to diffraction effects. In the deep ultraviolet wavelength range, especially for even shorter wavelengths, the generation of light sources is very complex.

Microscopes that use a mass particle beam to image nanostructures, such as electron microscopes, have significant advantages in view of the resolution exceeding optical microscopes due to the short de Broglie wavelength of electrons for imaging purposes. Similar to the situation of optical microscopes, for example, the diffraction limit of an electron microscope is linearly proportional to the de broglie wavelength of an electron and inversely proportional to the aperture angle of the electron beam used. Therefore, the diffraction limit of the electron beam can be reduced by accelerating the electrons of the electron beam to a greater kinetic energy.

However, as the energy of the electrons incident on the sample increases, the energy of the electrons or more generally the mass particles introduced into the sample also increases. However, it is generally undesirable to inject a large amount of energy caused by high velocity electrons or mass particles into the sample to be examined, given the potential for damage to sensitive samples. Reducing the kinetic energy of the electrons to minimize the potential for damage thereto, rather than increasing the aperture angle of the mass particle beam incident on the sample as much as possible, can solve this problem.

Generally, producing a low aberration electron beam, or more generally, a mass particle beam is more difficult than producing a low aberration beam. The aberration problem increases exponentially as the aperture angle of the electron beam increases, in particular since the spherical aberration increases substantially, so that there is a risk that the usable resolution of the electron beam microscope is not determined by the diffraction limit of the electron beam but by the wavefront aberration of the latter (electron beam).

The present invention thus solves the problem of specifying a method and an apparatus to at least partly avoid the above-mentioned dilemma.

Disclosure of Invention

This problem is at least partially solved day by the independent claims of the present application, according to an exemplary embodiment of the present invention. Exemplary embodiments are described in the dependent claims.

In one embodiment, a method of determining a mass particle beam wavefront comprises the steps of: (a) recording two or more images of the reference structure with the mass particle beam under different recording conditions; (b) generating point spread functions for the two or more recorded images with the modified reference image of the reference structure; and (c) performing a phase reconstruction of the mass beam in accordance with the generated point spread function and different recording conditions in order to determine the wavefront.

To generate the point spread function of the image of the reference structure, a modified reference image is used instead of the reference image. This can largely prevent the point spread function generated from the recorded image from containing artefacts contained in the image of the reference structure generated by the mass particle beam. Thus, the characteristics of the mass particle beam or its detection process reflected from the recorded image do not substantially affect the wavefront determined from the point spread function.

Thus, deviations of the determined wavefront of the mass particle beam from a specific wavefront (e.g. a spherical wavefront) can be corrected in a systematic way. Thus, a microscope using a mass particle beam can be operated with low kinetic energy of the mass particle beam, and at the same time with a large aperture angle of the mass particle beam, without its diffraction limited resolution being limited by wavefront aberrations of the particle beam.

In this description, by systematic it is meant that the aberrations exhibited by the wavefront of the quality particle beam are not optimized phenomenologically, for example image contrast, in terms of quantity, but are corrected systematically to the extent possible, i.e. to include all known aberrations.

In the present application, mass particles represent particles whose resting mass is greater than zero (m)₀> 0).

Here and elsewhere in this specification, the word "substantially" is used to denote an indication of a measured quantity within a measurement uncertainty if a measuring device according to the prior art is used to measure the corresponding quantity.

The different recording conditions may comprise different parameter settings of the source of the mass particle beam and/or of the imaging system and/or of the detection apparatus used for recording the images.

Different parameter settings of the source and/or imaging system may include: the kinetic energy of the mass beam, the diameter of the mass beam focal spot, the aperture angle of the mass beam, and the dispersion compensator (Stigmator) settings. Different parameter settings of the detection device may include: acceleration voltage of the detector, energy filter of the detector, and detector type.

When recording two or more images, the different recording conditions may comprise different focus settings of the mass particle beam. The two or more recorded images may comprise images of at least one focus stack of the reference structure. The focus stack may include two or more images. A single recorded image may already provide useful or helpful information for correcting artifacts if the distance to the focus at which the recorded image is located can be determined very accurately. However, a plurality of images are usually recorded in order to determine the focal position, from which possible errors can be determined and corrected.

Furthermore, the different recording conditions may include recording two or more images at different angles of incidence of the mass particle beam on the reference structure. As an alternative to recording the focus stack, images of the reference structure can be recorded from different perspectives, and a point spread function can be generated on the basis of these images, which function is then used to further perform phase reconstruction (phase recovery). For this purpose, it is necessary to compute modified reference images for reference structures recorded from different perspectives or different angles. The entire wavefront to be determined can be reconstructed from the wavefront determined in the sectional view.

In optics, the image of the focusing stack is typically used as an input to reconstruct the phase or wavefront of the particle beam, e.g., a photon beam. For a mass beam, this principle can be used as well to reconstruct its wavefront.

The different recording conditions may comprise different types of reference structures. The reference structure may have different geometric forms. For example, reference structures may include cubes, cuboids, pyramids, and cylinders. The one or more reference structures may comprise a material composition different from the material of the substrate to which the reference structures are applied. Due to the difference in material composition, a material contrast is also produced in addition to the topological contrast when the image is recorded. This maximizes the contrast of the recorded image of the reference structure.

The mass particle beam may have a particular energy. In addition, the mass particle beam may have an energy distribution with a predetermined width. Typically, the width of the (e.g. electron beam) energy distribution is determined by the energy resolution achievable by the electron source and is currently at about 0.5eV (electron volts). Thus, the magnitude is extremely independent of the potential through which the electrons pass for acceleration, and therefore of the energy of the beam electrons.

The reference image of the reference structure may represent at least one recording of the reference structure using the mass particle beam, wherein the reference structure is arranged in a focal spot of the mass particle beam. The reference image may also comprise a combination of a plurality of images recorded in the focal spot with the mass particle beam, for example in the sense of an average value. Furthermore, the reference image of the reference structure may be corrected in terms of artifacts.

When recording two or more images of the reference structure, the modified reference image of the reference structure may substantially correct for artifacts.

When generating the point spread function for two or more recorded images, the modified reference image of the reference structure may substantially correct for artifacts in the two or more recorded images. This makes it possible to substantially prevent the influence of these artifacts on the point spread function generated. This prevents these artifacts from affecting the determination of the wavefront of the quality particle beam.

Artifacts may be caused by the following reasons: when the reference structure is imaged with the mass particles, the reference structure is subjected to electrostatic charging by the mass particle beam and/or at least one edge effect of at least one edge of the reference structure in two or more recorded images.

Typically, the mass particles originating from one half of the space above the sample have a certain lateral dimension, which is used for the mass particle beam interacting with the sample during the detection process. If the sample has sharp edges, the space above the sample from which mass particles for detection purposes originate may be increased or decreased. This is indicated by an increase (an edge increases half the space and appears brighter than the surroundings) or decrease (a corner decreases half the space and appears darker than the surroundings) in the signal detected along the edge of the reference structure.

When recording an image using a mass particle beam, the particle beam will image the reference structure or a part of the reference structure in a distorted manner if the reference structure or a part of the reference structure is electrostatically charged. If these effects of the mass particle beam or its detection process in the reference image (by means of which the point spread function is generated) are not corrected, they are reflected in the generated point spread function and thus ultimately in the wavefront determined for the mass particle beam before the point spread function is generated from the recorded image.

Correcting the at least one artifact may include: an effect of electrostatic charging of the reference structure and/or an effect of at least one edge effect in the two or more recorded images is determined.

Determining the effect of the electrostatic charging and/or the effect of the at least one edge effect of the reference structure may comprise: simulating electrostatic charging and/or at least one edge effect of the reference structure. The simulation of the electrostatic charging and/or the at least one edge effect of the reference structure may be achieved by a model which simulates the interaction between the mass particle beam and the reference structure and/or between mass particles generated in the reference structure by the mass particle beam and the reference structure. Correcting the at least one artifact may include: removing the determined effect of the electrostatic charge and/or the determined effect of the at least one edge effect from the reference image.

Correcting the at least one artifact may include modifying a reference image of the reference structure. Modifying the reference picture of the reference structure may include image processing of the reference picture. The image processing may be performed based on simulation data of electrostatic charging of the reference structure and/or based on simulation data of at least one edge effect.

Generating the point spread function may include: the two or more recorded images are deconvolved with the modified reference image of the reference structure. The point spread function generated may include the intensity distribution of the mass particle beam in the plane of the two or more images.

The modified reference picture may correspond to an unmodified reference picture.

The method according to the invention can also be performed with an unmodified reference image. This may be advantageous, for example, if there is substantially no electrostatic charge on the surface of the reference structure and/or if the mass particle beam causes only small scale edge effects at the edges of the reference structure.

In an alternative embodiment, the two or more recorded images may be used to generate a correction image that is corrected for artifacts contained in the two or more recorded images due to the mass particle beam. To this end, the two or more recorded images may be corrected using the measured point spread function. Artifacts can be removed by computation from the corrected image.

The method of determining the wavefront of a mass particle beam may further comprise the steps of: providing a reference structure and/or characterizing the reference structure using a mass particle beam. Providing the reference structure may include providing reference structure design data. Characterizing the reference structure may include determining a reference image of the reference structure from the design data. Characterizing the reference structure may include recording a reference image using the mass particle beam. The characterization of the reference structure may include recording at least one image of the reference structure with the mass particle beam, wherein the reference structure is configured in a focal point of the mass particle beam.

The design data, the reference image of the reference structure, and/or the reference structure modified reference image may be provided in a non-volatile memory.

The reference structure may comprise at least one acicular material arrangement arranged on the substrate, and the acicular material arrangement and the substrate should have different material compositions. The needle-like material arrangement may comprise a metal or a metal compound, and the substrate may comprise quartz or carbon. The reference structure and the substrate to which it is applied should have the same material composition, the reference structure producing only a weak topological contrast in the recorded image, and therefore the reference structure only protrudes slightly, or not at all, from the substrate in the recorded image.

The acicular material configuration may have a lateral dimension in the range of 0.1nm to 10 μm, preferably 0.2nm to 500nm, more preferably 0.5nm to 100nm, and most preferably 1nm to 50 nm. The acicular material configuration may have a height in the range of 1nm to 1000nm, preferably 2nm to 200nm, more preferably 3nm to 100nm, most preferably 4nm to 20 nm. The needle-like material arrangement may comprise a cylindrical structure.

The reference structure may comprise at least one sharp edge and/or at least one defined sidewall angle. The sharp edge may have a value less than 10^-3mm radius of curvature, preferably less than 10^-4mm, more preferably less than 10^-5mm, preferably less than 10^-6mm. The sidewall angle may have an angle greater than 45 °, preferably greater than 80 °, more preferably greater than 85 °, and most preferably greater than 89 °. For example, if secondary electrons are used to detect a mass particle beam when recording two or more images, a thin reference structure (i.e., a reference structure with a low height) may reduce the topological effect. The lower bound of the reference structure height is provided by the fact that: when irradiating the reference structure with a mass particle beam, substantially no secondary particles of the substrate located below the reference structure should reach the detection system for recording an image of the reference structure, thereby avoiding tampering with the measurement of the reference structure.

Generating the modified reference picture may include: two or more reference images are recorded with different kinetic energies of the mass particle beam under recording conditions which are as optimal as possible, and the modified reference images are replaced by a combination of reference images recorded at different energies.

If it is possible to record a plurality of reference images with different kinetic energies of the mass particle beam using an imaging system which is as close as possible to the ideal imaging system, the modified reference images (whose correction values are usually calculated) can be replaced with the measurement images produced by the measurement in good approximation.

Further, a modified reference image may be generated from a combination of the calculated reference image and the one or more measured reference images.

The method according to the invention may further comprise the steps of: modifying the wavefront of the determined mass particle beam such that the modified wavefront substantially corresponds to the specified wavefront.

Correction of mass beam wavefront aberrations is not achieved by "blind" reduction of aberrations with respect to a selected parameter and therefore by maximizing a particular metric. In contrast, the methods described in the present application allow for the consideration of the effects of various known imaging aberrations when correcting a determined wavefront. Therefore, the degree of correction of aberrations and the efficiency of correction of the wavefront of the mass particle beam can be significantly improved. The high efficiency of wavefront correction means that only a few iterations are required to obtain a wavefront corresponding to a given wavefront within a defined deviation.

In one embodiment, an apparatus for determining a mass particle beam wavefront comprises: (a) recording means for recording two or more images of the reference structure using the mass particle beam under different recording conditions; (b) generating means for generating a point spread function of the two or more recorded images with the modified reference image of the reference structure; and (c) an execution means for executing phase reconstruction of the mass particle beam based on the generated point spread function and different recording conditions in order to determine the wavefront.

The apparatus may be implemented to record a reference image of a reference structure.

The apparatus may comprise an adjustment option for adapting the determined wavefront of the mass particle beam to the specified wavefront.

The aperture angle of the mass beam may be in the range of 0.1mrad to 1000mrad, preferably 0.2mrad to 700mrad, more preferably 0.5mrad to 500mrad, and most preferably 1mrad to 200 mrad. In this specification, the abbreviation "mrad" stands for milliradian.

The computer program may comprise instructions which, when executed by a computer system of the apparatus, cause the computer system to perform the method steps of one of the methods described above.

Drawings

The following detailed description describes presently preferred exemplary embodiments of the invention, reference being made to the accompanying drawings in which:

FIG. 1 depicts a schematic of a focused particle beam having a spot diameter and its aperture angle;

FIG. 2 shows a schematic cross-sectional view of some important components of an apparatus for determining the wavefront of a mass particle beam;

FIG. 3 schematically depicts a focusing stack for reproducing an image of a reference structure recorded with a mass particle beam of the apparatus of FIG. 2;

FIG. 4 shows a schematic plan view of a reference image of a reference structure in an upper partial image and a modified reference image of the reference structure after correction of an artifact of the upper partial image in a lower partial image;

FIG. 5 schematically shows a cross-section through a focused electron beam having a wavefront that causes aberrations;

FIG. 6 shows the focused electron beam of FIG. 5 after correction of the wavefront;

FIG. 7 schematically illustrates a schematic plan view of a simulated image of a cylindrical reference structure;

fig. 8 presents a schematic plan view for a simulated electron beam core whose intensity distribution at the focus is circular and uniform, and 4nm in diameter.

Fig. 9 presents a convolution of the image shown in fig. 7 with the kernel shown in fig. 8.

FIG. 10 shows a schematic plan view of a simulated reference image of the cylindrical reference structure of FIG. 7.

FIG. 11 is a schematic representation of a kernel or point spread function generated by deconvolving the image of FIG. 9 with the reference image of FIG. 10;

FIG. 12 reproduces a schematic cross-sectional view of the nucleus shown in FIG. 11;

FIG. 13 reproduces FIG. 7;

FIG. 14 repeats FIG. 8;

FIG. 15 reproduces FIG. 9 again;

FIG. 16 shows a schematic plan view of a modified simulated reference image of the cylindrical reference structure of FIGS. 7 and 13;

FIG. 17 schematically represents a point spread function generated by deconvolving the image of FIG. 15 or FIG. 9 with the modified reference image of FIG. 16;

FIG. 18 shows a schematic cross-sectional view of the point spread function of FIG. 17;

FIG. 19 shows a wavefront of an electron beam having a wavefront aberration;

FIG. 20 shows a focusing stack of the electron beam intensity distribution of FIG. 19;

FIG. 21 reproduces the convolution kernel of a cylindrical reference element, which has bright edges due to edge effects;

FIG. 22 reproduces a focus stack image of a convolution of the electron beam of FIGS. 19 and 20 with the convolution kernel of FIG. 21, wherein noise is added to the image of the cylindrical reference structure;

FIG. 23 shows a deconvolution kernel without accounting for edge effects;

FIG. 24 reproduces the deconvolution of the focus stack image of the cylindrical reference structure using the deconvolution kernel of FIG. 23;

FIG. 25 shows an image of the reconstructed focal stack of FIG. 24 after phase reconstruction;

FIG. 26 illustrates a reconstructed wavefront of the electron beam shown in FIG. 19;

FIG. 27 reproduces the difference between the wavefront of FIG. 19 and the reconstructed wavefront of FIG. 26 at five times magnification;

FIG. 28 shows a deconvolution kernel that takes into account edge effects when recording a cylindrical reference structure;

FIG. 29 reproduces the deconvolution of the focused stack image of the cylindrical reference structure of FIG. 22 using the deconvolution kernel of FIG. 28;

FIG. 30 shows the reconstructed focal stack image of FIG. 24 after phase reconstruction;

FIG. 31 illustrates a reconstructed wavefront of the electron beam of FIG. 19 determined based on a point spread function of the electron beam of FIG. 30;

FIG. 32 shows the difference between the wavefront of FIG. 19 and the reconstructed wavefront of FIG. 31 at five times magnification; and

fig. 33 specifies a flow chart of a method for determining a wavefront of a mass particle beam.

Detailed Description

The following explains the presently preferred embodiments of the method according to the invention and the apparatus according to the invention. The apparatus according to the invention is illustrated by way of example in a Scanning Electron Microscope (SEM). However, the method according to the invention and the apparatus according to the invention are not limited to mass particle beams in the form of electron beams. Instead, this can be used for any particle beam whose particles have a static mass different from zero. Furthermore, this can be used for microscopes that record images using a scanning focused mass particle beam, or for wide field microscopes. Mass and particle beams are used in the same sense hereinafter.

Some explanations regarding the resolution of scanning electron microscopes can be found in the preceding paragraph of the fifth part of the description. The de broglie wavelength of the electrons and electron beam is given by:

where λ represents the wavelength, h represents the planck constant, e represents the elementary charge, U represents the acceleration voltage for the electron to pass through and m represents the mass of the electron. The voltage for accelerating electrons in a microscope, the mass of an electron being approximately equal to its static mass m₀(m＝m₀)。

The following mathematical relationships are taken from the publications J.E.Barth, P.Kruit "adding different contributions to the charged particle probe size (Addition of differences to the charged particle probe size), Optik, Vol.101, p.101-. The diffraction limit of an electron beam is approximately described by the following equation:

herein, R is_DRepresenting the radius of the spot diameter D of the electron beam or charged particle beam at its focus. The angle α characterizes the aperture angle of the electron beam measured relative to the beam axis. Fig. 1 illustrates these relationships. Electron beam apparatusThere is a beam distribution or intensity distribution corresponding to or at least similar to a gaussian distribution. Focal point R_DThe specification of 50% of the spot radius means that the aperture angle a specifies the angle at which the electron beam intensity falls to half the maximum (HWHM, half width half maximum). As explained in the introductory part, the de broglie wavelength of a mass particle decreases with increasing energy thereof, and therefore the diffraction limit is shifted to a smaller radius or spot diameter. As can be derived from equation (2), the aperture angle α of the mass particle beam affects the diffraction limit in an inversely proportional manner. Thus, increasing the aperture angle α allows to increase the resolution of the electron beam, or more generally of the mass particle beam, while the kinetic energy of its particles remains unchanged.

In addition to the diffraction of the electron beam (or more generally, the mass particle beam), the limited brightness B of the electron beam and the limited area of the beam source also have an effect on the resolution of the electron beam or the mass particle beam. This contribution can be expressed by the following equation:

where I describes the current of the electron beam, B describes its brightness and E describes the kinetic energy of the electrons. The finite brightness or its contribution to the spot radius in the focal spot increases with increasing beam current I. The higher kinetic energy of the particles or electrons reduces this contribution. Similar to diffraction limit R_DFinite brightness R_IThe spot radius of (a) varies inversely with the electron beam, or more generally, the aperture angle a of the mass particle beam.

As shown in the introductory part, aberrations of the components of the mass particle beam imaging system may also have a significant influence on the resolution of the mass particle beam. The reason is that the chromatic aberration effect of the finite energy distribution Δ E of the particles within the particle beam can be expressed by the following equation:

wherein Δ E represents the Full width at half maximum (Full Wi) of the energy distribution of the particles of the mass particle beamdth at halfMaximum, FWHM), and C_CIs a design specific constant. As previously mentioned, the width of the electron beam energy distribution is determined primarily by the energy distribution of the electrons produced by the source. Currently available electron sources typically have an energy distribution AE of 0.5eV to 0.7 eV. First, by narrowing the energy distribution of the particles of the mass particle beam, or by increasing the average energy of the particle beam in the case where the energy distribution is constant, the imaging aberration caused by chromatic aberration can be reduced. Secondly, however, the contribution of the chromatic aberration increases linearly with increasing aperture angle α of the particle beam.

Spherical aberration is caused by imaging caused by aberration in a region of a component or element of the beam optics unit that is away from the axis. The contribution of spherical aberration to the resolution of a microscope based on using a mass particle beam can be expressed as follows:

R_S(50％)≈0.0884·C_S·α³， (5)

wherein, C_SAgain, a specific constant is designed. Particle beam R caused by spherical aberration_SIs proportional to the third power of the electron beam aperture angle alpha. Thus, spherical aberration quickly becomes increasingly important as the aperture angle of the particle beam is increased, and must be taken into account when analyzing the resolution of a particle beam-based microscope using a mass particle beam, as is the case with conventional optical systems.

When considering resolution, other imaging aberrations, such as astigmatism, coma, particle beam vibration, vibrations of the microscope using the mass particle beam can also be taken into account if desired. These effects are ignored in the following examples.

Resolution R of particle beam-based microscopes_totalIs determined by the sum of the squared values of the various effects affecting the resolution limit of the beam:

fig. 2 schematically shows in a cross-sectional view some components of an apparatus 200 for determining a mass particle beam wavefront. The exemplary apparatus 200 of fig. 2 is implemented in the form of a Scanning Electron Microscope (SEM) 210. The latter is disposed in a vacuum chamber 202. The scanning particle microscope 210 is composed of a particle emitter 205 and a column 215, in which a beam optical unit 220 in the form of an electron optical unit 220, for example of the SEM210, is arranged. The particle emitter 205 generates a mass particle beam 225, and an electron or beam optical unit 220 focuses the beam 225 and directs the latter onto the sample 110 at the output of the column 215.

Sample 110 may include a substrate 120 having a reference structure 130 disposed on a surface 115 thereof. The substrate 120 may include a plurality of reference structures 130, which may be implemented in the form of various geometries (not shown in FIG. 2). The substrate 120 may include a quartz substrate and/or a substrate having a Low Thermal Expansion coefficient (LTE). However, the substrate 120 may also include a carbon substrate or a carbon-containing substrate. In general, any material that produces a good material contrast between the substrate and the reference structure 130 can be used as the substrate for the reference structure.

The reference structure 130 may include a metal or a metal compound. For example, the reference structure 130 may include chromium or tantalum or a chromium or tantalum containing compound. The reference structure 130 may include one or more geometric figures. Thus, the reference structure may be implemented in the form of, for example, a cube, a cuboid or a cylinder. Further, the reference structure 130 may include at least one sharp edge and/or at least one steep sidewall angle. Since the substrate 120 and the reference structure 130 have different material compositions, the reference structure 130 is particularly highlighted by the material contrast in the image recorded by the mass particle beam in addition to its topological contrast.

For example, in the apparatus 200, the substrate 120 may comprise a substrate of a lithographic mask or a substrate of a template from nanoimprint lithography. The photolithographic mask may comprise a transmissive photomask or a reflective photomask. The photolithographic mask may include any mask type, such as a binary mask, a phase shift mask, a molybdenum silicide (MoSi) mask, or a mask for two or more exposures. One or more reference structures 130 may be disposed on a photolithographic mask or template for nanoimprint lithography.

The sample 110 is placed on the sample stage 230 or the sample holder 230. The sample stage 230 is also referred to in the art as a "stage". As indicated by the arrows in fig. 2, the sample stage 230 is movable in three spatial directions relative to the cylinder 215 of the SEM210, for example by means of a micromanipulator not shown in fig. 2. Particle beam 225 strikes sample 110 at measurement point 235. Thus, the sample stage 230 facilitates recording of the image focus stack of the reference structure 130 by its displacement along the beam axis of the particle beam 225, i.e. in the z-direction. Furthermore, the six-axis sample stage 230, by virtue of its tilt and/or rotation, allows the reference structure 130 to be recorded from various angles or perspectives. The respective positions of the respective axes of the sample stage 230 are measured interferometrically (not reproduced in fig. 2). In an alternative embodiment, the focus setting of the particle beam 225 may be set or changed by means of the electron optical unit 220 of the SEM 210. Combined adjustment by moving the sample stage 230 and setting the electron optical unit 220 is also possible.

As previously mentioned, the apparatus 200 in the exemplary embodiment illustrated in fig. 2 includes a SEM 210. An advantage of the electron beam 225 as a mass particle beam 225 is that the latter can be generated and formed relatively easily. However, an ion beam, an atomic beam, or a molecular beam (not shown in FIG. 2) may also be used in the apparatus 200. Generally, the apparatus 200 may use a particle beam 225, the particles of which have a stationary mass different from zero.

Further, the apparatus 200 of fig. 2 may include one or more scanning probe microscopes, for example in the form of an Atomic Force Microscope (AFM) (not shown in fig. 2), which may be used to analyze and/or process the sample 110.

A detector 230 arranged in the column 215 of the scanning particle microscope 210 converts the secondary electrons generated by the electron beam 225 at the measurement point 235 and/or the backscattered electrons from the sample 110 into electrical measurement signals and forwards the latter to an evaluation unit 265 of a computer system 250 of the apparatus 200. The detector 230 may include a filter or filtering system to discriminate electrons by energy and/or solid angle (not reproduced in fig. 2).

SEM210 of apparatus 200 may further include a second detector 240 for detecting secondary and/or backscattered electrons generated at measurement point 235 by incident electron beam 225. For example, detector 240 may comprise an Everhart-Sonley detector (Everhart-Thornley detector).

Furthermore, for the case where the sample 110 and/or the reference structure 130 are electrically isolated or have an electrically isolated surface layer 115, the scanning particle microscope 210 may include an ion source 245 that provides low energy ions in the region of the first measurement point 235.

The device 200 includes a computer system 250. The computer system 250 includes a scanning unit 255 that scans the electron beam 225 over the sample 110 and at least partially over the reference structure 130.

Furthermore, the computer system 250 comprises a setting unit 260, which setting unit 260 is used for setting and controlling various parameters of the scanning particle microscope 210 of the apparatus 200. Parameters of SEM 250 may include: the energy of the particle beam 225 or the electron beam 225, the aperture angle of the particle beam 225, the Stigmator setting of the beam optics 220 of the particle beam 225, the adjustment options for altering the spherical and/or chromatic aberration, coma and astigmatism. In addition, SEM210 of device 200 includes adjustment options for correcting higher order aberrations. Thus, SEM210 may correct for the first two stepsThe Zernike polynomials of (a). For high order correction, adjustment options are needed that can generate fields with tripolar or high order polar properties.

Further, the setting unit 260 of the SEM210 sets parameters of the detectors 230 and 240. In addition, the setup unit 260 of the computer system 250 of the SEM210 controls six axes of the sample stage 230.

Furthermore, the computer system 250 comprises an evaluation unit 265, which analyzes the measurement signals from the detectors 230 and 240 and generates therefrom an image, which can be displayed on the display 280 of the SEM 210. The area of the scanning unit 255 where the electron beam 225 or mass particle beam 225 is scanned across the sample 110 and/or reference structure 130 is displayed on the display 280 of the computer system 250 and is therefore designated as the field of view or FOV of the scanning particle microscope 210. In particular, the evaluation unit 265 is designed to generate an image of the reference structure 130 from measurement data of the detector 230 or of the detectors 230, 240. If the particle beam 225 is in focus with the reference structure 130, the evaluation unit 265 may generate a reference image of the reference structure 130 from the measurement data of the detectors 230, 240 if the mass particle beam 225 scans across the reference structure 130. However, the image generated by the evaluation unit 265 from the measurement data may still be part of the image of the focus stack of the reference structure 130 when the particle beam 225 scans the reference structure 130 in the focal spot.

The evaluation unit 265 likewise processes the measurement signals of the distance measuring devices of the one or more interferometers of the sample stage 230 and can likewise present these graphically and/or digitally on the display 280.

Furthermore, the evaluation unit 265 is designed to take account of the electrostatic charging of the sample 110 or of the reference structure 130 when showing the scanning area on the display 280 and thus when generating a reference image of the reference structure 130. Furthermore, evaluation unit 265 may instruct scan unit 255 to take into account its electrostatic charging when performing a scan of reference structure 130. In addition, the evaluation unit 265 may actuate the computer system 250 to at least partially compensate for the electrostatic charging of the reference structure 130 by locally irradiating the reference structure with low energy ions of the ion source 245.

For recording an image of the focus stack of the reference structure 130, the setup unit 260 moves the sample stage 230 along the beam axis of the particle beam 225, i.e. in the z-direction. Fig. 3 schematically shows images 310 to 390 of the focus stack 300 of the reference structure 130. The exemplary reference structure 130 of fig. 3 has a circular surface and a cylindrical structure. In the example shown in fig. 3, the focus stack includes nine images 310 through 390 of the reference structure 130. The focus stack 300 starts with an image 310 having an over-focusing (plus defocus), i.e., the focus is above or in front of the reference structure 130, and then progresses beyond the image 350 in focus toward an under-focusing (minus-focusing) (-defocus), where the focus is located in the reference structure 130. The evaluation unit 265 generates an image of the focus stack 300 of the reference structure 130 from the measurement data of the at least one detector 230, 240 at the respective defocus positions 315 and 325. The evaluation unit 265 may be implemented in hardware, software, solid state and/or combinations thereof.

Referring back to fig. 2, the computer system 250 of the SEM210 further includes a simulation unit 270. The simulation unit 270 is designed to incorporate the effects of the particle beam 225 produced by the one or more detectors 230, 240 by calculating to determine the specific detection process of the secondary particles produced by the mass particle beam 225. As already explained in the third section of the present specification, electrostatic charging of the reference structure 130 may result in display distortion of the reference image.

In the upper partial image 410, fig. 4 schematically shows a plan view of an image 450 of the reference structure 130 in focus, which is arranged on the substrate 120. The substrate 120 may include the substrate 120 of the sample 110. This means that the image 450 in the upper partial image 410 of fig. 4 is the reference image 450 of the reference structure 130. The exemplary reference structure 130 of fig. 4 has a rectangular form. Under the influence of electrostatic charging 420, reference image 450 has distortion 430 and displacement 440 relative to the design data of reference structure 130.

The simulation unit 270 of the computer system 250 of fig. 2 may contain, for example, a model describing the electrostatic charging 420 and the resulting image distortion based on the material composition of the reference structure 130 and the kinetic energy with which electrons or particles of the particle beam 225 are incident on the reference structure 130.

Furthermore, the simulation unit 270 of the computer system 250 may be designed to computationally determine the effect of the half-space variation from which the secondary particles generating the image originate, due to the specific geometry of the reference structure 130, based on the geometry of the reference structure 130 and the energy of the particle beam 225. The upper partial image 410 of fig. 4 illustrates an edge effect 460 caused by the reference structure 130, which is illustrated by the thick border line 445 in the reference image 450 of the reference structure 130.

The evaluation unit 265 of the computer system 250 of the apparatus 200 of fig. 2 contains one or more algorithms which allow the specific effects 430, 440 and 460 of the particle beam specific image recording process determined by the simulation unit 270 to be corrected in the reference image 450 of the reference structure 130. Thus, evaluation unit 265 may generate modified reference picture 480 from reference picture 450. This is illustrated in the lower partial image 460 of fig. 4. Artifacts caused by electrostatic charging of the reference structure 130 and edge effects 460 at sharp edges 470 are substantially corrected in the modified reference image 480 of the reference structure 130.

The algorithms of the simulation unit 270 may be implemented in hardware, software, or a combination thereof. For example, the authors s.babin et al, j.vac.sci.technol.b 24(6), pp.2056-2959 (11/12 2006) "technique for automatically measuring the diameter and astigmatism of an electron beam: BEAMERER "describes a method of determining the beam size of an electron beam by an automated process. A corresponding simulation program beameter is available from eBeam Technologies, inc.

Furthermore, the simulation unit 270 is designed to determine the convolution kernel or point spread function of the images 310 and 390 of the focus stack 300 of the reference structure 130 from the images 310 and 390 of the focus stack 300 of the reference structure 130 and the modified reference image 480 of the reference structure 130 by performing a deconvolution operation. Finally, based on the point spread function of the focus stack 300, the simulation unit 270, by performing a phase reconstruction, may be used to determine the mass particle kinetic energy of the particle beam 225 and the respective defocus positions 315, 325 when recording the image 310-390 of the focus stack 300 of the reference structure 130, the wavefront of the particle beam 225 when recording the image 310-390 of the focus stack 300.

The simulation unit 270 may perform phase reconstruction by using a known algorithm, such as a Gerchberg-Saxton algorithm, an NLSQ (non-linear least squares) algorithm, a Yang-Gu algorithm, a ping-pong algorithm, or a Ferweda algorithm.

Each of the computer system 250, the evaluation unit 265 and/or the simulation unit 270 may comprise a memory, preferably a non-volatile memory (not shown in fig. 2), containing one or more models for charge charging of the various reference structures 130 and/or various models of edge effects 460, which are caused by the detection process when scanning the reference structures using the mass particle beam 225.

As shown in FIG. 2, the evaluation unit 265 and/or the simulation unit 270 may be integrated into the computer system 250. However, the evaluation unit 265 and/or the simulation unit 270 may also be implemented as a dedicated unit (not shown in fig. 2) inside or outside the device 200. In particular, the evaluation unit 265 and/or the simulation unit 270 of the computer system 250 may be designed to perform some of its tasks using a dedicated hardware implementation.

Computer system 250 may be integrated into apparatus 200 or implemented as a stand-alone device (not shown in FIG. 2). The computer system 250 may be configured using hardware, software, firmware, or a combination thereof.

Unlike that shown in fig. 2, scanning particle microscope 210 of apparatus 200 may comprise a multi-beam scanning particle microscope capable of directing multiple particle beams (not shown in fig. 2) simultaneously on sample 110. Multi-beam scanning particle microscopes include detectors or detector arrangements that can detect secondary particles generated by individual particle beams in parallel. Furthermore, the evaluation unit 265 of the multi-beam scanning particle microscope may be designed to combine partial images produced by the secondary particles of the individual particle beams to form a whole image.

Diagram 500 of fig. 5 schematically shows a cross section of a focused mass particle beam 510 having an aperture angle a with respect to a beam axis 520. The particle beam 510 is focused to a spot diameter 530 in a focal point 560. As previously described, wavefront 550 is deconvolved from images 310-390 of focus stack 300 and modified reference image 480 and determined with the aid of a phase reconstruction algorithm and has a relatively large aberration or aberration 560 relative to wavefront 540 of the incident spherical wave. The aberrations 560 prevent the resolution of the mass particle beam 510, which is limited by the diffraction limit of the particle beam 510 when there are no aberrations, from being fully utilized. This is avoided by the aberrations 560 of the wavefront 550 of the mass particle beam 510.

The determined wavefront 550 allows its aberrations 560 to be systematically corrected. SEM210 of apparatus 200 of fig. 2 has an adjustment option for correcting all substantial aberrations of wavefront 550. The wavefront can be corrected in several ways. First, for various possible adjustment options for SEM210, the sensitivity of the adjustment options for SEM210, such as the coil current in beam optics unit 220, may be measured. Further, the wavefront can be corrected by calculation. For this reason, models suitable for the effects of the various adjustment options of SEM210 are necessary. In addition, it may be helpful if SEM210 has high production accuracy.

Typically, the adjustment options of SEM210 do not act on only a single aberration, i.e., a linearly independent Zernike (Zernike) polynomial. Therefore, a so-called interaction matrix is formulated in a preferred process. The interaction matrix is known from adaptive optics. The interaction matrix may be formulated by calculation (i.e., based on the design of SEM210, or from experimental results). When the interaction matrix is formulated based on experimental results, each adjustment option to SEM210 measures the effect on each linearly independent aberration. The resulting interaction matrix thus converts the vector of adjustment options into a vector of aberrations. Thus, matrix inversion allows for the generation of a matrix that allows for the conversion of wavefront aberrations (expressed as aberrations) into a vector of adjustment options for the SEM210 that are required for correction purposes.

Fig. 6 shows wavefront 550 of fig. 5 after correction for aberration 560. As can be seen from fig. 6, the corrected wavefront 650 has substantially the wavefront 540 of a spherical wave. Thus, the systematic correction of the determined wavefront 550 allows the SEM210 using a mass particle beam 510 with a large aperture angle α to obtain a resolution limited by the diffraction limit of the particle beam 510 as a usable resolution.

Fig. 7-12 show a first example for generating a point spread function for the electron beam 225, 510. The first example described in fig. 7 to 12 and the other examples described below are carried out based on simulation. In the first two examples shown, the reference structure 130 is implemented in the form of a cylinder with a diameter of 10 nm. For example, the cylinder may be fabricated by depositing chromium on a quartz substrate.

FIG. 7 shows a simulated image of the cylindrical reference structure 130 when it is irradiated by an electron beam 225, 510, the electrons of which have a kinetic energy of 600 eV. The lighter outer edge 750 represents the edge effect 460 of the cylindrical reference structure 130.

Fig. 8 shows a blur kernel 850 used to illustrate the effect of the electron beam 225, 510 having a diameter of 4nm on the image recording of the cylindrical reference structure 130 of fig. 7. The core 850 employed has a circular beam with a perfect plane at the focal point. In fig. 8, this is specifically illustrated by the ratio of the standard deviation σ of the intensity to the average intensity ave (σ/ave is 0).

FIG. 9 reproduces the convolution of the cylindrical reference structure 130 of FIG. 7 with the 4nm wide electron beam 225, 510 or kernel 850 of FIG. 8. The bright edge 750 of the image of fig. 7 is as clearly visible as the bright edge 950 in the blurred convolved image 900 of fig. 9. Fig. 9 presents an image of reference structure 130 that is visible on display 280 of SEM210 when reference structure 130 is scanned by electron beams 225, 510.

FIG. 10 shows a reference image 450 of a cylindrical reference structure 130, which is used to deconvolute the image 900 of FIG. 9. The reference image 450 has a uniform intensity distribution over the planar surface of the cylindrical reference structure 130. The reference image 450 is used to generate a convolution kernel 850 or point spread function 850 of the electron beams 225, 510 from the blurred convolved image 900 by deconvolution.

Fig. 11 shows a convolution kernel 1150 or point spread function 1150 generated from the images 900, 1000 of fig. 9 and 10. As is evident from the ratio of the standard deviation σ to the average intensity ave, the point spread function 1150 varies over a region of 2.5% across the surface. This conflicts with the ideal point spread function 850 of fig. 8.

FIG. 12 shows a cross section of the point spread function 1150 of FIG. 11. Unlike what is assumed for the convolution kernel 850 of FIG. 8, the point spread function 1150 has no flat surface. Furthermore, the intensity distribution of the point spread function 1150 exhibits a pronounced rounded edge 1250.

Fig. 13-18 present a second example for generating a point spread function for the electron beam 225, 510. As before, the second example is also carried out based on simulations. Fig. 13, 14 and 15 of the second example correspond to fig. 7, 8 and 9 of the first example.

Unlike the first example, in the second example of fig. 13 to 18, the deconvolution of the image 900 of fig. 15 is not carried out with the reference image 450 of the cylindrical reference structure 130 of fig. 10. In contrast, FIG. 16 presents a modified reference image 480. When the reference image 450 of the cylindrical reference structure 130 is imaged or recorded by the electron beams 225, 510, the modified reference image 480 takes into account the edge effect 460, which is represented by the kernel 850. This is visible in modified reference image 480 by bright edges 1650. Thus, the modified reference image 480 has a non-uniform intensity distribution on the planar surface of the cylindrical reference structure 130. The generation of modified reference image 480 from measured reference image 450 has been explained previously for the case shown in fig. 4.

The modified reference image 480 is used to deconvolute the convolution kernel or generate a point spread function from the blurred convolved image 900 of fig. 15. Fig. 17 shows the convolution kernel 1750 or point spread function 1750 generated from the images 480, 900 of fig. 15 and 16. The resulting convolution kernel 1750 does not reproduce the entirety of the perfect point spread function of FIG. 17. However, the deviation from the perfect convolution kernel 850 (a/ave ratio of 0.4%) is about six times better than the point spread function 1150 generated based on the reference image 450.

Similar to FIG. 12, FIG. 18 shows a cross-sectional view of the point spread function 1750 of FIG. 17. As can be inferred from fig. 14, the point spread function 1750 of fig. 17 closely approximates the cylindrical intensity distribution of the electron beam. The point spread function 1750 generated as illustrated in the second example may be used as an input in a phase reconstruction process for determining the wavefront 550 of the electron beam 225, 510.

Fig. 19-32 below present two examples of wavefronts 550 used to reconstruct an electron beam for which the wavefront is imperfect. Similar to the two examples above, the data discussed below was generated by simulation. For both examples described below, the phase reconstruction is additionally carried out as a further step, so that the deviation of the determined wavefront from the specific wavefront can be determined.

In this simulation, the following two examples assume that the kinetic energy of the electron beam is 400 eV. Further, it is assumed below that the wavefront 550 of the electron beam 1910 is subject to random variations of half a wavelength with respect to the average value (RMS, root mean square) of the wavefront. Further, the following example specifies that the electron beam 1910 has a Numerical Aperture (NA) of 0.04 or 40 milliradians. In both examples, the reference structure 130 is again implemented in the form of a cylinder, which now has a diameter of 16 nm.

Fig. 19 reproduces the wavefront of the electron beam 1910, which was used to image the cylindrical reference structure 130 in the following simulation. As before, the wavefront 550 of the electron beam 1910 has random variations of half de Broglie wavelength. The purpose of the following simulation is to generate a point spread function of the electron beam 1910 from the electron beam 1910 with the wavefront aberration 1920, which facilitates phase reconstruction with the smallest possible wavefront aberration.

Fig. 20 presents five intensity distributions of the electron beam 1910, which correspond to five different focus settings of the electron beam 1910. In the five images of fig. 20, as considered from the top down, the focal point has the following positions relative to the surface of the reference structure 130: 125nm, -62.5nm, 0nm, 62.5nm and 125 nm. The image of fig. 20 shows the point spread function of the electron beam 1910 of fig. 19, which is used as an input quantity in a further simulation process. The wavefront aberration 1920 of the electron beam 1910 of fig. 19 is significantly prominent in the focus stack of the intensity distribution.

Fig. 21 shows a convolution kernel 2100 for a cylindrical reference structure 130. The convolution kernel 2100 of the cylindrical reference structure 130 has an edge effect 460 as previously described that is highlighted in the convolution kernel 2100 of the reference structure 130 by a bright edge 2150.

Fig. 22 reproduces the convolution of the cylindrical reference structure 130 of fig. 21 with the intensity distribution of the electron beam 1910, which is explicitly indicated in the image of the focusing stack of fig. 20. In addition, noise is added to the image of the reference structure 130. Thus, the image of FIG. 22 shows an image of the focus stack of the reference structure 130 that the SEM210 of the apparatus 200 will produce when the reference structure 130 is scanned by the electron beam 1910, and which is displayed on the display 280.

Fig. 23 shows a deconvolution kernel 2350, which in the illustrated third example is used to deconvolve an image of the focus stack of the reference structure 130 of fig. 22. The deconvolution kernel 2350 represents an ideal image without artifacts of the reference structure 130. This means that in the example of fig. 23, the reference image of the reference structure 130 has no additional brightness in the border or edge region of the cylindrical reference elements.

FIG. 24 reproduces an image of the focus stack of the reference structure 130 of FIG. 22 deconvolved by the deconvolution kernel 2350 of FIG. 23. The phase reconstruction is performed in the next step. The phase reconstruction is carried out by modifying the wavefront of the measured image of the focusing stack of figure 24, which results in a focusing stack. The image of fig. 25 reproduces the result of the phase reconstruction performed on the basis of the image of fig. 24.

Fig. 26 shows a reconstructed wavefront 2620 of an electron beam 1910 closest to the focus stack of fig. 25. The reconstructed wavefront 2620 of the electron beam 1910 is adapted by means of Least squares fitting (Least squares fit).

To check the accuracy of the wavefront reconstruction of the electron beam 1910, a difference is formed between the defined wavefront 1920 (input or reference wavefront) of the electron beam 1910 and the reconstructed wavefront 2620. Fig. 27 shows the difference wavefront magnified by five times. The reconstructed wavefront 2620 of the electron beam 1910 has a deviation of 0.133 wavelengths from the plane wavefront. The remaining wavefront aberration 2620 is mainly caused by the remaining astigmatism of the electron beam 1910.

In marchehal approximation, the input wavefront 1920 of the electron beam 1910 has the following Strehl factor: s ═ exp [ - (2 π. 0.500)²]＝5.17·10^-5. In the same approximation, the reconstructed wavefront 2620 has a Strehl factor: s ═ exp [ - (2 pi-0.133)²]0.497. This represents a deconvolution of the image of the focus stack of fig. 22 with the convolution kernel 2350 of fig. 23 resulting in a significant improvement of the reconstructed wavefront of the electron beam 1910 and thus an improvement of the image quality of the recorded image.

The fourth example explained below based on fig. 28 to 32 is based on the initial case described on the basis of fig. 19 to 22. Instead of deconvolution with the deconvolution kernel 2350 of fig. 23, deconvolution of the focus stack of the image of the reference structure 130 of fig. 22 is with the convolution kernel 2850 of fig. 28, which accounts for the edge effect 460 with an excess edge height 2860.

Fig. 29 presents an image deconvolved with the deconvolution kernel 2850 of fig. 28 that is an image of the focus stack of reference structure 130 of fig. 22.

The image of the focal stack of fig. 30 presents the focal stack of fig. 29 reconstructed by phase reconstruction from the image of the focal stack of fig. 22 reference structure 130.

Similar to fig. 26, fig. 31 shows a reconstructed wavefront 3120 of the electron beam 1910, which corresponds to the greatest possible extent to the focus stack of fig. 30, which in turn corresponds to the measured focus stack of fig. 29.

To check the accuracy of the wavefront reconstruction of the electron beam 1910, a difference is formed between the defined wavefront 1920 (input or reference wavefront) of the electron beam 1910 and the reconstructed wavefront 2620. Fig. 27 shows the difference wavefront magnified by five times. The reconstructed wavefront 2620 of the electron beam 1910 has a deviation of 0.133 wavelengths from the plane wavefront. The remaining wavefront aberration 2620 is mainly caused by the remaining astigmatism of the electron beam 1910.

As already explained above in the case of fig. 27, in the Marechal approximation the input wavefront 1920 of the electron beam has the following Strehl factor: s ═ exp [ - (2 π. 0.500)²]＝5.17·10^-5. In the same approximation, the reconstructed wavefront 2620 has a Strehl factor: s ═ exp [ - (2 pi · 0.072)²]0.815. By deconvolving the image of the reference structure of the focus stack of fig. 22 with the deconvolution kernel 2850 of fig. 28 (which takes into account the edge effect 460), the image quality can again be significantly improved compared to fig. 26 and 27.

The methods described in the last two examples may be performed multiple times in succession (iteratively) to minimize the remaining residual error in determining wavefront 550. When the deconvolution kernel 2850 that accounts for the edge effect 460 is used, the method performed iteratively converges to a smaller residual error than the deconvolution kernel 2350 of fig. 23. The methods explained in this application can handle noise in the recorded images of the reference structure 130. However, at a certain level of noise, the latter can compromise the efficiency of the correction.

Finally, the flow chart 3300 of fig. 33 summarizes again the necessary steps of the described method for determining the wavefront 550 of the mass particle beam 225, 510, 1910. The method starts at step 3110. In a next step 3320, two or more images 310-390 of the reference structure 130 are recorded using the mass particle beams 225, 510, 1910 under different recording conditions 315, 325. The two or more images 310-390 may be part of the focus stack 300. The recording of images 310-390 may be performed using scanning unit 255 of SEM210 of apparatus 200. The setup unit 260 may achieve different recording conditions 315, 325 by adjusting the sample stage 230.

In step 3330, a point spread function 1750 for the two or more recorded images 310-390 is generated using the modified reference image 480 of the reference structure 130. A reference image 450 of the reference structure 130 may be recorded with the mass particle beams 225, 510, 1910 under control of the scanning unit 255 of the computer system 250. The changes to be made in the reference image may be determined by means of the simulation unit 270. The evaluation unit 265 of the computer system 250 may perform the changes determined by the simulation unit 270 in the reference image 450, thus being able to generate a modified reference image 480. The evaluation unit 265 deconvolves the recorded images 310-390 with the modified reference image 480 to generate the point spread function 1750 for the images 310-390.

In step 3340, in order to determine wavefront 550, a phase reconstruction of mass particle beams 225, 510 is performed based on the generated point spread function 1750 and the different recording conditions 315, 325. This phase reconstruction may be performed by the simulation unit 270 of the computer system 250. The method ends at step 3350.