Method and apparatus for generating a multidimensional XCT measurement data set of a sample and computer-implemented method for reconstruction

Abstract

A method and apparatus for generating a multidimensional XCT measurement data set (I) of a sample (P) using spectral multi-structured scatter imaging are disclosed. A plurality of scatter images (19) are generated with a measurement process comprising the steps of: a) structuring a spectrum of an EUV / X-ray radiation (6) as a structured frequency range for providing an EUV / X-ray measuring beam (12) with a measuring beam spectrum which comprises narrow-band structured frequency ranges, wherein at least one axial frequency space coordinate (Kz) in the frequency space (Kx, Ky, Kz) is assigned to each of the narrow-band structured frequency ranges; b) irradiating the EUV / X-ray measuring beam (12) onto a spatially laterally extended surface section of the sample (P); c) detecting scattered radiation, which is scattered back from the sample (P), using a detector (17) which outputs intensity values for the structured frequency range belonging to the measurement process, to each of which at least one lateral frequency space coordinate (Kx, Ky) in the frequency space (Kx, Ky, Kz) is assigned, wherein the intensity values form the scatter image (19) for the associated measurement process. The intensity values can be summarised to form the multidimensional XCT measurement data set (I), wherein the positions of the intensity values in the frequency space (Kx, Ky, Kz) are given by at least one lateral frequency space coordinate (Kx, Ky) and one axial frequency space coordinate (Kz). Also disclosed are computer-implemented methods for reconstructing a multidimensional structure of a sample from a multidimensional XCT measurement data set.

Description

Method and device for generating a multidimensional XCT measurement data set of a sample and computer-implemented method for reconstruction [1] The present invention relates to a method for generating a multidimensional (in particular three- or two-dimensional) EUV coherence tomography measurement data set of a sample by means of multispectral, in particular spectrally multistructured, scattering imaging, a computer-implemented method for reconstructing a multidimensional structure from a multidimensional EUV coherence tomography measurement data set and a device for obtaining multidimensional EUV coherence tomography measurement data of a sample. [2] The so-called XCT - optical coherence tomography in the spectral range of EUV radiation or X-ray (X-ray) radiation - is based, like optical coherence tomography in the visible spectral range (optical coherence tomography - OCT), on the backscattering of incident electromagnetic radiation at an interface of a sample to be examined. By evaluating the interference of the radiation (back)scattered at several interfaces, information about the sample to be examined can be obtained. A prerequisite for this is, in addition to spatial superposition, a low temporal coherence of the superimposed radiation. For a basic introduction to OCT technology, see e.g.: "Optical coherence tomography - principles and applications AF. Fercher et al., Rep. Prog. Phys. 66 239 (2003) or "Optical Coherence Tomography", W. Drexler et al., 2nd edition.Springer (2015), ISBN 978-3-319-06419-2 (eBook), ISBN 978-3-319-06418-5 (hardcover), http: / / dx.doi.org / 10.1007 / 978-3-319-06419-2. [3] XCT implements the concepts of OCT examinations for radiation in the UV range from 5 eV to 25 eV (or 50 nm to 250 nm) or EUV (extreme ultraviolet) range from 5 eV to 250 eV (or 5 nm to 250 nm) and in the (soft) X-ray range from 250 eV to >1000 eV (or 1.23 nm to 5 nm). Deviating from the generally accepted use of the abbreviation EUV for the aforementioned (energy / wavelength) range, the term EUV is used here for the common spectral range of UV (ultraviolet) and EUV for reasons of clarity (unless otherwise stated). XCT uses, for example, synchrotron-based or laser-based sources to generate EUV or X-ray radiation. Radiation sources. XCT measures spectrally resolved broadband radiation, e.g., EUV radiation, reflected from an irradiated area of ​​a sample and, averaged over the irradiated area, determines the deep structure of the sample using Fourier transformation. See, for example, US 7,656,538 B2 ("frequency wave length coherence tomography") for laser-based XCT. Due to the large usable frequency bandwidths in, for example, EUV, XCT enables axial resolution in the range of, for example, a few tens of nanometers, although more recent developments aim for axial resolution in the nanometer range. The high axial resolution, as can be achieved, for example, in laser-based XCT, is accompanied by a - by comparison, much coarser - lateral resolution in the range of a few tens of pm. [4] In XCT, the lateral resolution of a sample is determined by, among other factors, the extent of the irradiated area, the irradiation direction, the irradiation distribution, and the detector-side imaging. For a planar XCT analysis of a sample, the sample can be scanned laterally, usually by moving the sample relative to the incident radiation. [5] For the broadband EUV or X-ray radiation used, optics with a low numerical aperture (NA) are typically used. This low NA determines the extent of the irradiated area and limits the lateral resolution. However, due to the broadband nature of the radiation used and the low dispersion and low refractive index of various materials in the underlying wavelength range – combined with often high absorption – it is difficult to implement a high NA in the optical setup. [6] To avoid unwanted absorption, reflection optics are almost exclusively used in the EUV wavelength range. This applies both to the irradiation of the broadband EUV radiation and to the detection of the backscattered radiation. [7] In the scientific publication "Laboratory setup for extreme ultraviolet coherence tomography driven by a high-harmonic source," J. Nathanael et al., Rev. Sci. Instrum. 90, 113702 (2019), an exemplary setup for detecting XCT spectra using a nanostructured sample as an example, as well as exemplary data processing of the XCT spectra, are described. Spectroscopic XCT is further explained in "Coherence tomography with broad bandwidth extreme ultraviolet and soft X-ray radiation," S. Skruszewicz et al., Applied Physics B (2021) 127:55. [8] How an exemplary XCT evaluation in combination with a one-dimensional phase retrieval (PR) algorithm can derive the deep structure of a sample from an autocorrelation signal is explained in detail in the scientific publication “Optical coherence tomography with nanoscale axial resolution using a laser-driven high-harmonic source”, S. Fuchs et al., Vol. 4, No. 8 / August 2017 / Optica and the associated supplement (“Supplementary information”). [9] The increasing importance of structural investigations of complex 3D semiconductor structures using X-ray radiation is shown, for example, in the scientific publication “Soft x-ray: novel metrology for 3D profilometry and device pitch overlay”, C. Porter et al., Proc, of SPIE Vol. 12496 1249611-9, SPIE Advanced Lithography + Patterning, 2023, San Jose, California, United States using the example of “gate all around (GAA) devices”.

[0010] Also known are lensless imaging techniques or coherent diffraction imaging techniques, which are typically referred to as "coherent diffraction imaging" (CDI) or ptychography (see, among others, Rodenburg, "Ptychography and Related Diffractive Imaging Methods", Advances in Imaging and Electron Physics 150 (2008)). These techniques enable high lateral resolution by using spatially coherent radiation and large-area detectors or line detectors. In lensless imaging, a diffraction image of the sample is acquired with the detector, whereby imaging optics can be dispensed with.

[0011] From the intensity data of these diffraction images, the phase of the detected light is calculated algorithmically in an iterative calculation to derive lateral sample information. Examples of iterative CDI algorithms are known, for example, from Marchesini, "Invited Article: A unified evaluation of iterative projection algorithms for phase retrieval," Rev. Sci. Instrum. 78, 011301 (2007).

[0012] One aspect of this disclosure is based on the object of specifying methods that can capture and resolve nanostructures of a sample in a multidimensional (e.g., three- or two-dimensional) manner within the scope of an EUV or X-ray radiation-based examination. In particular, comparable resolutions in the depth direction and the lateral directions are of interest. A further object is to specify a device and a method with which data for a multidimensional Structural analysis of a sample. A further task is to provide methods for evaluating such data.

[0013] At least one of these objects is achieved by a method according to claim 1, a computer-implemented method according to claim 13 and a device according to claim 21. Further developments are specified in the subclaims.

[0014] In one aspect, a method for generating a multidimensional XCT measurement data set £>(Kx, Ky, Kz) of a sample using spectral multistructured scattering imaging comprises the following steps: - Providing EUV / X-ray radiation with a broadband spectrum in the EUV to X-ray spectral range, - Generating a plurality of scattering images, each of the plurality of scattering images being generated using a measuring process comprising the steps of: a) structuring the broadband spectrum of the EUV / X-ray radiation as a structured frequency range to provide an EUV / X-ray measuring beam with a measuring beam spectrum comprising at least one of a plurality of narrowband structured frequency ranges, wherein at least one axial frequency space coordinate in the frequency space is assigned to each of the narrowband structured frequency ranges, b) irradiating the EUV / X-ray measuring beam onto a spatially laterally extended surface section of the sample, c) detecting scattered radiation that is backscattered by the sample with a detector that outputs intensity values ​​for the structured frequency range associated with the measuring process, each of which is assigned at least one lateral frequency space coordinate in the frequency space,where the intensity values ​​form the scatter pattern for the corresponding measurement process; and, - Combining the intensity values ​​to form a multidimensional XCT measurement data set £>(Kx, Ky, Kz), where the positions of the intensity values ​​in the frequency space are given by at least one lateral frequency space coordinate and one axial frequency space coordinate.

[0015] In a further aspect, a device for obtaining a multidimensional XCT measurement data set £>(Kx, Ky, Kz) of a sample comprises an EUV / X-ray radiation source for generating broadband EUV / X-ray radiation and an optical device for structuring the broadband spectrum of the EUV / X-ray radiation as a structured frequency range and outputting an EUV / X-ray measurement beam with a measurement beam spectrum, which comprises at least one of a plurality of narrowband structured frequency ranges for a measuring process, wherein at least one axial frequency space coordinate in the frequency space is assigned to the narrowband structured frequency ranges. Furthermore, the device comprises a sample holder for supporting the sample at an angle of incidence of the EUV / X-ray measuring beam onto a surface portion of the sample, and a focusing element configured to form a spatially lateral beam expansion parameter of the EUV / X-ray measuring beam on the surface portion of the sample, wherein the beam expansion parameter lies in particular in a range from one micrometer to several hundred micrometers, measured based on the FWHM intensity of the EUV / X-ray measuring beam.Furthermore, the device comprises a detector arranged to record intensity values ​​in a far field from the sample, and a control device connected to the optical device for structuring the broadband spectrum for setting the structured frequency range for a plurality of measurement processes and to the detector for reading out the intensity values ​​and designed to compile the multidimensional XCT measurement data set £>(Kx, Ky, Kz) from the intensity values.

[0016] In a further aspect, a computer-implemented method for reconstructing a multidimensional structure of a sample from a multidimensional XCT measurement data set £>(Kx, Ky, Kz) comprises the steps: - Reading, into a processor, intensity values ​​of a multidimensional XCT measurement data set £>(Kx, Ky, Kz), where positions of the intensity values ​​in the frequency space are given by at least one lateral frequency space coordinate and one axial frequency space coordinate, and - with the processor, executing a Fourier transform algorithm which converts the multidimensional XCT measurement data set £>(Kx, Ky, Kz) into a sample structure data set O(x, y, z), wherein the sample structure data set O(x, y, z) comprises in a spatial space a depth direction coordinate of the sample P which is assigned to the axial frequency space coordinate, and at least one lateral coordinate of the sample which is assigned to the at least one lateral frequency space coordinate.

[0017] In some further developments, the broadband spectrum can be structured by using a monochromator to diffract a narrowband frequency range from the EUV / X-ray radiation, which propagates as the EUV / X-ray measuring beam along an optical beam path to the sample.

[0018] In some further developments, the scattered radiation can be detected by detecting the scattered radiation of an associated structured frequency range reflected by the sample in one lateral dimension with a line detector or in two lateral dimensions with a planar detector and outputting it as a one-dimensional or two-dimensional scattering image.

[0019] In some embodiments, the at least one lateral frequency-space coordinate can be determined by parameters of the EUV / X-ray measurement beam and parameters of the measurement process. Additionally or alternatively, the axial frequency-space coordinate can be given by the photon energy of the radiation detected at a lateral frequency-space coordinate.

[0020] In some developments, the intensity values ​​of a scatter pattern from a measurement process can be projected onto an axial frequency-space coordinate assigned to the measurement process, in particular onto an average axial frequency-space coordinate assigned to the respective measurement process. This can be given, for example, by: Kz = (2 sin a) k; where k is a wave vector associated with a narrowband structured frequency range, and a is an angle of incidence of the EUV / X-ray measurement beam to the surface section of the sample.

[0021] In some developments, each of the intensity values ​​of a scatter pattern from a measurement process can be projected onto an axial frequency-space coordinate of a plurality of axial frequency-space coordinates, in particular depending on the at least one lateral frequency-space coordinate of each of the intensity values. Additionally or alternatively, axially uniform scatter patterns can be compiled based on projected axial frequency-space coordinates, which comprise intensity values ​​from scatter patterns acquired in different measurement processes at the same axial frequency-space coordinate.

[0022] In some developments, several axial frequency space coordinates can be assigned to the associated structured frequency range for at least one measurement process, and in one measurement process, intensity values ​​for several scattering patterns can be obtained by assigning the intensity values ​​to respectively associated axial frequency space coordinates of a narrow-band structured frequency range.

[0023] In some developments, a first subset of measurement processes can be carried out at a first relative position between the incident EUV / X-ray measurement beam and the spatially laterally extended surface section of the sample and a second subset of measurement processes at a second relative position between the incident EUV / X-ray measurement beam and the spatially laterally extended surface section of the sample, and furthermore, the multidimensional XCT measurement data set £>(Kx, Ky, Kz) are formed from the intensity values ​​of the first subgroup and the second subgroup in the frequency space, in particular taking into account the axial frequency space coordinates.

[0024] In some further developments, the EUV / X-ray measuring beam can be irradiated onto a plurality of spatially laterally extended, adjacent or partially overlapping surface sections of the sample and the recorded intensity values ​​of the scattering images assigned to the surface sections can be combined to form the three-dimensional XCT measurement data set £>(Kx, Ky, Kz).

[0025] In some developments, the scattering patterns can be detected in a plurality of measurement processes with different structured frequency ranges. The plurality of scattering patterns can be recorded accordingly for a plurality of narrowband structured frequency ranges, in particular for a plurality of central wavelengths.

[0026] In some further developments, the EUV / X-ray measuring beam can be spectrally structured differently for N measuring processes, in particular it can be adjusted for several narrowband structured frequency ranges in order to carry out N measuring processes with different spectral resolution.

[0027] In some further developments, the EUV can combine the majority of scattering images as a two-dimensional Fourier measurement data set or as a three-dimensional Fourier measurement data set to form the multi-dimensional XCT measurement data set.

[0028] In some further developments of the computer-implemented method, this may also include: - a scaling step in which, for each of the plurality of axial frequency space coordinates of the frequency space, intensity values ​​in the at least one lateral frequency coordinate are scaled based on a common photon energy, so that in particular the increments in the at least one lateral frequency coordinate underlying the multidimensional XCT measurement data set £>(Kx, Ky, Kz) for all axial frequency space coordinates are matched to each other, and / or - a field calculation step in which the square root of each intensity value is taken.

[0029] In some developments of the computer-implemented method, the Fourier transformation algorithm can be configured to determine a phase of the detected scattered light by means of Fourier transformations, wherein in particular the phase is reconstructed iteratively by repeatedly executing a loop with at least one Fourier transformation, at least one inverse Fourier transformation and setting boundary conditions in the Fourier or position space for the Fourier transformation and / or the inverse Fourier transformation.

[0030] In some further developments of the computer-implemented method, the Fourier transform algorithm may include: - an XCT reconstruction step for one-dimensional phase reconstruction in a depth-direction coordinate of the spatial space, which derives a partially reconstructed sample structure O'(Kx', Ky', z) based on the axial frequency-space coordinate for lateral positions, comprising a plurality of depth-resolved diffractograms, wherein the XCT reconstruction step comprises in particular an ID phase reconstruction XCT algorithm, and - a lateral reconstruction step for one- or two-dimensional phase reconstruction in at least one lateral coordinate, which derives the sample structure data set O(x, y, z) based on the at least one lateral frequency space coordinate and at least one subset of the plurality of depth-resolved diffractograms.

[0031] In some further developments of the computer-implemented method, this may further include - a signal evaluation step in which cumulative signal contributions of the depth-resolved diffractograms are calculated to identify depth positions contributing to the sample structure data set O(x, y, z), and wherein the subset of the plurality of depth-resolved diffractograms for the lateral reconstruction step may comprise the depth-resolved diffractograms of the contributing depth positions.

[0032] In some developments of the computer-implemented method, the Fourier transform algorithm can include a 3D reconstruction step for multidimensional Phase reconstruction in a depth-direction coordinate of the sample structure data set O(x, y, z) and at least one lateral coordinate of the sample structure data set O(x, y, z) based on the axial frequency-space coordinate and the at least one lateral frequency-space coordinate. In particular, a region of direct reflection can be excluded from the joint reconstruction.

[0033] In some embodiments of the computer-implemented method, the multidimensional XCT measurement data set £>(Kx, Ky, Kz) can be generated according to one of the methods disclosed herein.

[0034] In some developments of the computer-implemented method, a plurality of scattering images can be fed to the computer-implemented method as a two-dimensional Fourier measurement data set or as a three-dimensional Fourier measurement data set, wherein the scattering images were detected in a plurality of measurement processes with different structured frequency ranges.

[0035] In some embodiments of the device, the EUV / X-ray radiation source can be a synchrotron-based or laser-based radiation source. Additionally or alternatively, the sample holder can be configured: - for storing the sample to adjust the angle of incidence of the EUV / X-ray measuring beam and / or - for storing the sample in focus or in front of the focus of the focusing element and / or - for scanning the sample by laterally moving the sample with respect to the EUV / X-ray measuring beam.

[0036] In some developments of the device, the optical device can be designed as a monochromator or multichromator, in particular a grating-based, grating-pair-based, or multi-slit-based monochromator or multichromator. In particular, a grating-based or grating-pair-based monochromator or multichromator can be rotatably mounted to adjust a central frequency of the structured frequency range. Additionally or alternatively, a multi-slit-based monochromator or multichromator can be adjustable to adjust a central frequency of the structured frequency range with respect to a multi-slit configuration, such as slit width. Furthermore, additionally or alternatively, the focusing element can be arranged upstream or downstream of the optical device for structuring the broadband spectrum.

[0037] In some developments of the device, the focusing element can be designed to form a spatially lateral beam expansion parameter of the EUV / X-ray measuring beam on a surface section of the sample, which defines a lateral dimension of the sample and accordingly a lateral frequency component and is in particular in a range from 1 pm to 1 mm, in particular from 10 pm to 500 pm, measured based on the FWHM intensity of the EUV / X-ray measuring beam.

[0038] In some developments of the device, the detector for recording one-dimensional scatter patterns can be configured as a line detector for detecting a series of intensity values, or for recording two-dimensional scatter patterns, as a planar detector for detecting an array of intensity values, and / or can be rotatably mounted for aligning a detected lateral dimension. Furthermore, the intensity values ​​detected by the detector during a measurement process can each form a scatter pattern, or an intensity value detected by the detector can be assigned to one of a plurality of scatter patterns depending on an axial frequency space coordinate assigned to the intensity value.

[0039] In some embodiments of the device, the control device may comprise a processor configured to execute at least one of the methods disclosed herein.

[0040] In some developments, radiation in the range from 5 eV to 250 eV and / or soft X-ray radiation in the range from 250 eV to 1000 eV can be provided as (spatially coherent) EUV / X-ray radiation. In particular, EUV radiation with a bandwidth in the range from 25 eV to 100 eV and / or soft X-ray radiation with a bandwidth in the range from 50 eV to 150 eV can be provided, wherein the bandwidths are selected in particular such that a number of axial frequency space coordinates required for resolution in the depth direction, and thus of narrowband structured frequency ranges, can be sampled in the measurement processes.

[0041] In some further developments, a spatially laterally extended surface section of the sample can extend laterally in at least one direction over one or a few micrometers up to several hundred micrometers, measured based on the FWHM (Full Width at Half Maximum) intensity of the EUV / X-ray measuring beam 12.

[0042] In some developments, the detector, which is particularly designed as a line detector or as a linearly movable detector, can be used to acquire a one-dimensional scatter pattern in which the intensity values ​​are arranged along a lateral frequency-space coordinate (e.g., Kx or Ky). A two-dimensional XCT measurement data set with a lateral frequency-space coordinate (Kx or Ky) and the axial frequency-space coordinate (Kz) can be generated as a multi-dimensional XCT measurement data set (Kx or Ky, Kz). If multiple Kz values ​​are acquired per measurement process due to the measurement geometry or the measurement beam spectrum, reassignment is possible to obtain one-dimensional scatter patterns or, in general, the two-dimensional XCT measurement data set.In some developments, the detector, which is designed in particular as an area detector or as a detector that can be moved to a two-dimensional position grid, can be used to acquire a two-dimensional scatter pattern in which the intensity values ​​are arranged in a grid defined by two lateral frequency-space coordinates Kx and Ky. A three-dimensional XCT measurement data set with the two lateral frequency-space coordinates Kx and Ky and the axial frequency-space coordinate Kz can be generated as a multidimensional XCT measurement data set £>(Kx, Ky, Kz). If multiple Kz values ​​are acquired per measurement process due to the measurement geometry or the measurement beam spectrum, reassignment is possible to obtain two-dimensional scatter patterns or, in general, the three-dimensional XCT measurement data set.

[0043] In some developments of the device, the optical device can be designed to set different structured frequency ranges, in particular a plurality of narrow-band structured frequency ranges, in particular with a plurality of central wavelengths, for detecting the scattering images (19) in a plurality of measuring processes.

[0044] In further developments of the device, the device for obtaining a multidimensional XCT measurement data set can be configured to structure the EUV / X-ray measurement beam differently spectrally for N measurement processes, in particular to adjust it for several narrow-band structured frequency ranges in order to carry out N measurement processes with different spectral resolutions.

[0045] In some developments of the method for generating a multidimensional XCT measurement data set or of the computer-implemented method for reconstruction, when combining the intensity values ​​to form the multidimensional XCT measurement data set, at least one lateral frequency space coordinate is scaled depending on the axial frequency space coordinate.

[0046] In general, a broadband spectrum in the UV / EUV to X-ray spectral range is understood herein to mean, for example, a spectrum that comprises a plurality of narrowband frequency ranges and, in particular, enables the selection or extraction of several, e.g., 5 to 10, 10 to 100, or 10 to 1000 or more, narrowband frequency ranges from the broadband spectrum. "Narrowband" refers to the fact that the narrowband frequency range—possibly depending on the lateral frequency coordinate—can be assigned an axial frequency coordinate in the frequency space. Narrowband includes, for example, spectral widths in the range of X / δZ (here as the center wavelength of the narrowband frequency range and 5k as the width, e.g., the FWHM width, of the narrowband frequency range): 5 (or 10 or 20 or 50) to 10000 (or 5000 or 2000 or 1000 or 500), in particular from 50 to 200, whereby the spectral width obviously also depends on the respective position of X in the spectral range.

[0047] Furthermore, for example, broadband spectra of the provided EUV / X-ray radiation extend in the spectral range from 5 eV to >1000 eV, e.g. in the spectral range from 5 eV (or from 10 eV or 20 eV) to 1000 eV (or to 500 eV or 100 eV), in particular in the UV range from 5 eV to 25 eV (or 50 nm to 250 nm), in the UV / EUV (extreme ultraviolet) range from 5 eV to 250 eV (or 5 nm to 250 nm) and in the (soft) X-ray range from 250 eV to >1000 eV (or 1.23 nm to 5 nm) or within one or more of these ranges belonging to the spectral range. Broadband includes, for example, spectral widths in the range from 5 eV to 25 eV - especially in the UV range - or spectral widths in the range from 5 eV (or from 10 eV or 25 eV or 30 eV) to 70 eV (or up to 100 eV or 250 eV or 500 eV or 1000 eV) - especially in the EUV / X-ray range, and in particular spectral widths in the range from, for example, 30 eV to 100 eV in the UV / EUV / X-ray range. Area.

[0048] The concepts described herein can, among other things, offer the following advantages in the structural analysis of a sample compared to state-of-the-art methods such as electron microscopy or avoid corresponding disadvantages of the state of the art:

[0049] The invention enables laterally extended deep structure imaging of multidimensional nanostructures in a spatially extended sample, such as a nanometer-layer coating of an optical element (e.g., lithograph e-masks) or a nanostructured semiconductor element (e.g., microprocessors, memory elements). The laterally extended deep structure imaging proposed herein is performed using a novel concept that combines XCT with aspects of lensless imaging, referred to herein as diffractive XCT (DXCT for short) for "diffractive XUV or X-ray coherence tomography" (diffractive EUV / X-ray coherence tomography).

[0050] The inventors have recognized and demonstrated that a combination of aspects of XCT with aspects of coherent diffraction imaging is feasible, whereby the advantages of the high NA of coherent diffraction imaging, in particular the increased lateral resolution, become accessible within the framework of 2D or 3D nanoscopy. In other words, the proposed laterally extended deep structure imaging combines the advantages of both methods and can thus simultaneously achieve high axial resolution and high lateral resolution in the multidimensional reconstruction of a sample's deep structure based on scattering images.In the context of the DXCT concepts disclosed herein, "multidimensional" means, when generating a multidimensional XCT measurement data set and reconstructing a structure of a sample from a multidimensional XCT measurement data set, that an analysis in the axial direction is also performed at the same time as an analysis in at least one lateral direction.

[0051] The inventors recognized that laterally extended deep structure imaging for sample reconstruction neither requires phase knowledge nor a complete scan of the Fourier space. Although the measured, incompletely acquired Fourier space results in a lower resolution, this is acceptable given the high resolution potential of the proposed laterally extended deep structure imaging in multidimensional space for many structures, especially since resolutions in the range of a few nanometers can be achieved.

[0052] The 2D or 3D deep structure imaging proposed herein can be applied to (large) samples where the two- or three-dimensional Fourier space of the sample can be sufficiently captured within a DXCT measurement at many wavelengths with a correspondingly large detector. The size of the detector, i.e., the scanning / measurement signal acquisition in at least one lateral direction, is important for the information acquisition in this lateral direction and influences the lateral Resolution. The spectral bandwidth or the nature of the wavelength scan are important for the axial resolution.

[0053] The method proposed here also demonstrates advantages for samples that, due to their dimensions, weight, and shape, can only be aligned and moved in space with very high technical effort. Thus, the concepts disclosed herein can enable high multidimensional resolution of the sample without the need for controlled and precise rotation, as is common in rotational or tomographic measurement techniques such as computed tomography (CT).

[0054] In addition, an additional rotation of the sample and / or the illumination and measurement direction can also help with the concepts disclosed herein to further increase the spatial resolution by additional information of the two- or three-dimensional Fourier space of the sample.

[0055] The concepts proposed herein can enable high multidimensional resolution even with detectors with a low NA. Furthermore, a high NA can also make a larger region in the multidimensional Fourier space of the sample accessible for measurement, thus further increasing the spatial resolution.

[0056] The inventive concepts disclosed herein for 2D or 3D imaging ("2D / 3D nanoscopy") with a resolution in the nanometer range in at least one lateral dimension and the depth dimension (for example, in all three spatial dimensions) give rise to new applications such as the non-destructive investigation of layered samples in reflection geometry (e.g., EUV mask metrology (layer-based photomasks in EUV lithography), of (multilayer) coatings such as optical or EUV mirrors, of functional structures in solar cells or semiconductors (graphene-based electronics), and of biological membrane layer structures. Furthermore, they enable the diagnosis of periodic and non-periodic structures. In particular, the 2D or 3D imaging concepts disclosed herein enable two- or three-dimensional imaging of near-surface structures of thick samples.For example, in the investigation of (structured) semiconductors, they represent an interesting alternative or complement to scanning electron microscopy (SEM), X-ray reflectometry (XRR) or transmission electron microscopy (TEM), since the latter usually require destructive or invasive preparation of the sample.

[0057] Disclosed herein are concepts that allow aspects of the prior art to be improved, at least in part. In particular, further features and their usefulness will become apparent from the following description of embodiments with reference to the figures. The figures show: Fig. 1 shows a schematic Kz-Kx section through an Ewald sphere to explain the measurement of the Fourier space within a backscattering-based DXCT measurement method, Fig. 2 is a schematic representation of an exemplary DXCT measurement setup, Fig. 3A-3C are sketches to illustrate an exemplary DXCT measurement method for obtaining photon energy-dependent scattering patterns, Fig. 4 is a sketch illustrating the generation of a stack of scaled scatter images as a three-dimensional DXCT measurement data set, Fig. 5 is a flow chart illustrating an exemplary serial reconstruction of a three-dimensional structure of a sample from a three-dimensional DXCT measurement data set, Fig. 6 an overview sketch of the serial reconstruction Fig. 7 is a flow chart illustrating an exemplary parallel reconstruction of a three-dimensional structure of a sample from a three-dimensional DXCT measurement data set, Fig. 8 an overview sketch of the parallel reconstruction and Fig. 9 is a sketch illustrating the extension of DXCT measurement data by lateral scanning of a sample.

[0058] Essential for at least some of the concepts described herein is that, according to the invention (for the first time), multidimensional and three-dimensional DXCT measurement data in the form of spectrally resolved scattering images (depending on the structure to be examined, also diffraction images) are acquired for a volume range of a sample and compiled for evaluation. The multidimensional DXCT measurement data generated and reconstructed according to the concepts disclosed herein comprise, in addition to an axial frequency space coordinate (Kz), at least one of the lateral frequency space coordinates (Kx, Ky). With new reconstruction methods proposed herein, a 3D structure of the sample can be determined, for example, from a stack of two-dimensional scattering images within the framework of a 3D phase reconstruction. In general, new reconstruction strategies for reconstructing a 3D object / 3D sample from the DXCT measurement data are also developed, in particular for The proposed approach is to obtain two-dimensional or three-dimensional structural information with high 2D or 3D resolution. The reconstruction strategies proposed herein differ fundamentally from known reconstruction algorithms based on one-dimensional axial or two-dimensional lateral approaches. In addition to multidimensional DXCT measurement data, the new reconstruction strategy can also be used to reconstruct data from modeled 2D or 3D structures, for example, as part of a model comparison.

[0059] The diffractive XCT proposed herein is a multispectral imaging technique in which a set of acquired scattering images provides information about the sample along two frequency axes (Kx or Ky), Kz, or along three frequency axes Kx, Ky, Kz. The scattering images are measured at different wavelengths in reflections and do not contain any phase information. Diffractive XCT forms a specific (2D or 3D) Fourier spatial distribution of the sample from the scattering images. Starting from this distribution, the phase information along the (two or three) frequency axes and thus along the (two or three) spatial axes (here (x or y), z, or x, y, z) can be derived – for multiple axes sequentially or simultaneously – using phase recovery algorithms. The spatial structure of the three-dimensional sample can thus be reconstructed.

[0060] To explain the physical background of diffractive XCT, the underlying backscattering is first considered generically (in 3D) (Fig. 1). Subsequently, an exemplary DXCT measuring device (Fig. 2), exemplary DXCT measurement methods, the generation of a 3D Fourier data set £>(Kx, Ky, Kz) (Figures 3 and 4), and exemplary evaluation methods (Figures 5-8) are described. Furthermore, a 2D conversion and sampling within the framework of DXCT (Fig. 9) are explained.

[0061] It is known that an electromagnetic field scattered by a sample in Fourier space corresponds, for example, to a first Born approximation of a 3D Fourier transform of the spatial scattering potential of the sample. Conversely, with knowledge of the scattered field in Fourier space (given by the (detector pixel) values ​​Kx, Ky, Kz as (spatial) frequency coordinates; accordingly, the Fourier space is also referred to herein as frequency space or K-space), a spatial structure of the "scattering" sample can be reconstructed; i.e., the spatial structure can be calculated from the reflectively sampled Fourier space. Reconstruction is possible, for example, using Fourier transforms. The basis for the reconstruction methods proposed here Strategies involve measuring DXCT data in Fourier space. The measurement is performed, for example, with a pixelated (e.g., line or area) detector in backscatter.

[0062] Fig. 1 illustrates the backscattering of spatially coherent radiation using an Ewald sphere in a Kz-Kx section. Fig. 1 illustrates which region in Fourier space can be detected with diffractive XCT. A vector k(i) represents the incident radiation, whose wavelength X can be adjusted within the scope of the DXCT measurement and which specifies the direction of illumination onto the sample. A normal vector n refers to the surface of the sample. A vector k(s) illustrates the backscattered radiation. The difference vector K = k(s) - k(i) and the change in k(i) yields the sample information in Fourier space. The measurement of the sample (DXCT measurement points) in the far field proposed here using a planar and pixelated detector provides access to the multidimensional Fourier space of the sample and thus to the (lateral) structure and the deep structure of the sample.

[0063] DXCT measurement points in 3D Fourier space correspond to measurement points with (spatial) frequency coordinates Kx, Kz in the sectional view shown in Fig. 1. The frequency coordinate Kx and also the frequency coordinate Ky (not shown in Fig. 1) are determined by the illumination geometry (parameters of the illuminating EUV / X-ray radiation, in particular of the measurement beam such as beam parameters, spatial coherence, . . .) and the measurement geometry (parameters of the measurement process such as angle of incidence, spatial distance, arrangement and geometry of the detector). The frequency coordinate Kz is essentially determined by the photon energy or energy (“central” frequency m, wavelength X or wavenumber k) of the detected radiation. The following approximately applies: Kz = (2m cos a) / c = (4ir cos <z) / X = (2 cos a) k, wobei a der Einfallswinkel durch die Beleuchtungsrichtung relativ zum Oberflächenlot der Probe bzw. den Normalenvektor n ist. Für die Betrachtung in Bezug auf das Oberfläche der Probe würde die Kosinusfunktion durch eine Sinusfunktion ersetzt. Mit anderen Worten können die Intensitätswerte eines Streubilds eines Messvorgangs auf eine dem Messvorgang zugeordnete axiale Frequenzraumkoordinate projiziert werden, insbesondere auf eine mittlere, dem jeweiligen Messvorgang zugeordnete axiale Frequenzraumkoordinate. Der beschriebene Zusammenhang zwischen Kz und Wellenzahl k gilt insbesondere für Messungen, bei denen Detektoren verwendet werden, die nur einen kleinen Raumwinkelanteil des rückgestreuten Feldes abdecken und / oder wenn der Fourierraum der Probe relativ grob diskretisiert ist.

[0064] For large detectors or for a more precise discretization of the sample's Fourier space, the curvature of the Ewald sphere can be taken into account when calculating Kz. In other words, each of the intensity values ​​of a scattering pattern from a measurement process can be projected onto an axial frequency-space coordinate Kz of a plurality of axial frequency-space coordinates, in particular depending on the at least one lateral frequency-space coordinate of each of the intensity values. In particular, based on the projected axial frequency-space coordinates Kz, axial uniform scattering patterns can be compiled, comprising intensity values ​​from scattering patterns acquired in different measurement processes for the same axial frequency-space coordinate.

[0065] Furthermore, for at least one measurement process, two narrowband frequency ranges that are spectrally separated from each other and thus different axial frequency space coordinates can be assigned to the associated (narrowband) structured frequency range, so that several scattering patterns can be obtained in one measurement process by assigning the intensity values ​​of different pixels to different axial frequency space coordinates.

[0066] Three circle segments shown in dashed lines in Fig. 1 represent the DXCT measurement points in Fourier space that can be detected with a detector for one of the wavelengths I, X2 and X3. It can be seen that both the position and the length of a circle segment, i.e. the DXCT measurement points in Fourier space that can be detected for a wavelength, change when the measurements (in backscattering) are carried out with a changed (central) wavelength. The range in Fourier space that can be detected by means of diffractive XCT results from the adjustable wavelength range and the spatial geometry of the measurement setup. As an example, Fig. 1 shows a detectable range Bl of Fourier space in the Kz-Kx section through the Ewald sphere.

[0067] In Fig. 1, an overly large detector NA is schematically assumed to illustrate the DXCT measurement. Such a large NA is not typically present in the DXCT measurement setups and methods described below. For example, a (laser-based) diffractive XCT also functions with significantly smaller NA values.

[0068] For example, in one measurement series, a detector angle acceptance of ±0.3° and an NA of ~ 0.006 were used. In this case, a single measurement with a single wavelength results in essentially linearly arranged measurement points (Kx, Ky) at a fixed Kz value. As already mentioned, these measurement points (Kx, Ky) can also be address different Kz values, for example, if larger solid angles or a larger NA and / or a finer spectral discretization or spectral resolution are present. In such cases, the individual measurement points (Kx, Ky) must be assigned to the corresponding Kz values ​​in the 3D Fourier data set £>(Kx, Ky, Kz).

[0069] For the laser-based DXCT setup described below, the wavelength can be varied, for example, in the range of approximately 7 nm to 60 nm (or 20 eV to 80 eV). The DXCT proposed here can also be implemented with appropriate light sources and optics, for example, for longer-wavelength spectral ranges >60 nm (<20 eV) as well as for shorter-wavelength spectral ranges <7 nm (>180 eV). The so-called water window between 2.3 nm and 4.4 nm (282 eV–533 eV) is particularly relevant for organic samples.

[0070] If many measurements are made with central wavenumbers ko in the range 0 < ko, min < ko < ko ma, the shape of the detectable area Bl in the 3D Fourier space approximately corresponds to a truncated pyramid for an exemplary rectangular pixel array as a (planar) detector. The (non-measurable) tip of the pyramid lies at the coordinate origin of the Fourier space. The height of the pyramid is determined by the complete bandwidth of the wavelengths used and can be understood as a measure of the axial resolution. The aperture angle of the truncated pyramid corresponds to an angular acceptance of the detector, ie, the numerical aperture NA of the detector, and can be understood as a measure of the lateral resolution.

[0071] In the DXCT measurement method described in more detail below, scattering patterns of a sample are acquired for a plurality of central wavelengths (photon energies). As previously explained, the photon energy-dependent scattering patterns can be assigned to the lateral frequency coordination Kx and Ky. To reconstruct the sample, the photon energy underlying a scattering pattern is transferred to one or more Kz values ​​of the energy-dependent frequency coordinate according to the above projection. A stack of scattering patterns can be summarized as a two-dimensional Fourier measurement data set £>((Kx or Ky), Kz) or as a three-dimensional Fourier measurement data set £>(Kx, Ky, Kz) and submitted to evaluation as a "measurement result". For example, a three-dimensional Fourier measurement data set £>(Kx, Ky, Kz) can be evaluated to reconstruct a three-dimensional sample structure O(x, y, z) in spatial space.

[0072] In tomographic measurement techniques, the sample is measured from different illumination directions to ensure high three-dimensional resolution. In the diffractive XCT disclosed herein, a single illumination direction is sufficient to enable high two-dimensional resolution. Thus, DXCT can achieve lateral resolutions of <100 nm combined with an axial resolution of a few tens of nm.

[0073] Furthermore, in the case of diffractive XCT, the detectable area B1 in Fourier space can be further expanded by applying tomographic methods, such as rotating the illumination direction of the sample and the detector. Fig. 1 shows this schematically for the case of a second detectable area B2 in Fourier space with a changed illumination direction. For clarity, the illumination direction is rotated by -35° here. In other words, after rotating the sample into a new orientation with respect to the usually fixed beam path and tracking the detector, another DXCT measurement can be performed, thus filling another area in Fourier space with DXCT measurement points.In other words, a first subgroup of measurement processes can be carried out at a first relative position between the incident EUV / X-ray measurement beam and the spatially laterally extended surface section of the sample P and a second subgroup of measurement processes can be carried out at a second relative position between the incident EUV / X-ray measurement beam and the spatially laterally extended surface section of the sample P, such that the intensity values ​​of the first subgroup and the second subgroup complement each other in the frequency domain (in particular taking into account the axial frequency domain coordinates). Within the scope of laser-based diffractive XCT, for example, changes in the angle of incidence of a few tenths of a degree, over a few degrees up to a full rotation can be implemented, so that the detected areas in the Fourier space can, for example, partially overlap or be adjacent to one another. For example, to expand the Fourier space to be analyzed, the detector can optionallyB. can be adjusted relative to the sample using a controllable detector holder (see Fig. 2).

[0074] The explanations in connection with the 2D section of the Ewald sphere shown in Fig. 1 already show that the scattering analysis concepts disclosed herein are also applicable to two-dimensional XCT measurement data that include an axial frequency space coordinate Kz and a lateral frequency space coordinate. Such two-dimensional XCT measurement data can, for example, be used for a two-dimensionally structured sample with a Line detector according to the concepts disclosed herein and subsequently to Reconstruction of a two-dimensional structure of the sample.

[0075] The following description refers, by way of example, to the analysis of a sample in three dimensions, see in particular the description with reference to two-dimensional scattering patterns (Kx, Ky), such as those underlying Figures 3-8.

[0076] An exemplary setup of a laser-based DXCT measurement setup 1 designed to acquire DXCT measurement data is shown schematically in Fig. 2.

[0077] The exemplary DXCT measurement setup 1 comprises a (laser) radiation source 3. The radiation source 3 is based, for example, on a femtosecond laser system for generating pulsed laser radiation with a central wavelength of, for example, 800 nm and pulse durations in the range of several tens of femtoseconds at repetition rates in the kHz range. The radiation source 3 further comprises an optical parametric amplifier, which can shift the wavelength, for example, into the infrared or into the range around 1300 nm and generate laser pulses with a pulse energy in the mJ range. The radiation source 3 further comprises a source chamber 5. The laser pulses are focused into the source chamber 5, which is filled, for example, with argon gas, to generate high harmonics (High Harmonic Generation HHG). In addition to a classic, strongly modulated HHG spectrum, a quasi-continuous spectrum, for example, can be generated by changing the signal wavelength in the optical parametric amplifier by time averaging.in the EUV range. EUV / X-ray radiation 6 generated in this way propagates along an optical beam path 7 in the DXCT measurement setup 1.

[0078] In addition to the DXCT measurement setup presented as an example, the DXCT method can also be performed with other radiation sources or modified laser-based setups that can provide (spatially coherent) radiation for a variety of wavelengths. For example, accelerator-based radiation sources such as synchrotron or plasma-based radiation sources, as well as other laser radiation sources such as fiber lasers, disk lasers, or rod lasers, can be used.

[0079] For the high harmonic generation process, OPCPA systems can also be used instead of OPA for wavelength matching of the fundamentals. Furthermore, for the generation of broadband EUV radiation by high harmonic generation, so-called gating approaches can also be used to reduce the number of The aim is to limit the time points at which the high harmonics are generated as much as possible. A more general approach is to use laser pulses with a few optical cycles, whereby, ideally, the high harmonics are emitted at only one time point.

[0080] With powerful laser beam sources, coherent broadband radiation can be provided for the ultraviolet spectral range (UV, > 5 eV or < 250 nm), the extreme ultraviolet spectral range (EUV, > 25 eV or < 50 nm), and the soft X-ray range (SXR, > 250 eV or < 5 nm) by generating high harmonics. This includes spectral ranges with low absorption, such as the silicon transmission window in the EUV range (30-99 eV; 12-41 nm) for semiconductor-based samples and the water transmission window in the X-ray range (285-531 eV; 4.4-2.3 nm) for organic samples.

[0081] The maximum (theoretical) axial resolution is determined by the coherence length l c « (2 In 2) / n * (Ä 2 / AA) and thus by the broadband nature of the light source used. In these spectral ranges, axial resolutions of ~11 nm for the silicon transmission window and ~3 nm for the water transmission window are theoretically possible with diffractive XCT. With the DXCT concepts disclosed herein, lateral resolutions of <100 nm can be achieved with a single illumination geometry (i.e., without sample rotation). In general, spectral windows can be addressed depending on the sample material, existing absorption, and resolution.

[0082] In DXCT, the relevant spectral range can be formed from (several narrow-band) structured frequency ranges for a sufficient number of measurement processes (scattering images) with the measuring beam, for example 50, 100, 200, 500 or more measurement processes, whose central photon energies or wavelengths can be varied by, for example, 0.1 eV, 0.2 eV, 0.5 eV, 1 eV or more apart.

[0083] Regarding the coherence of the radiation, coherence in the lateral direction (spatial coherence) plays a particularly important role. This is preferably achieved across the measuring beam in such a way that the generation of scattering patterns with the highest possible contrast is enabled for lateral examination of the sample.

[0084] In general, and especially in comparison with a synchrotron, laser-based radiation sources based on the generation of high harmonics have the advantage of compact structure. Furthermore, laser-based radiation sources can ensure a high photon flux with high spatial coherence. Furthermore, in the spectral range accessible by high-energy spectroscopy, the reflectivity of solid samples can be sufficiently high, so that laser-based radiation sources can be used to investigate, for example, lithography masks and nanostructured solid samples in materials science, particularly those with buried 2D materials. The sensitivity with which structures can be measured depends on factors such as the scattering cross section, the size, and the roughness of the structure. At least in the case of a layered structure, for which a reflectivity can also be defined, structures with reflectivities in the range of approx. 0.001% up to 100% can be measured using DXCT.

[0085] For a good reconstruction of a 3D Fourier data set £>(Kx, Ky, Kz), for example, a good contrast of (different) diffraction or interference structures in the scattering images is important. Among other things, a high spatial coherence of the (e.g. EU V-) radiation across the entire beam cross-section, as can be found in laser-based EUV sources, has a positive effect on the contrast in the scattering images.

[0086] Following EUV / X-ray generation in the source chamber 5, the optical beam path 7 of the EUV / X-ray radiation runs in a vacuum to a sample chamber 9, in which a sample P is arranged in the beam path 7 by means of a sample holder 10. For the DXCT analysis of the sample P, a pressure in the range of 1 * 10 8 mbar can be reached.

[0087] The EUV / X-ray radiation 6 generated in the source chamber 5 is spectrally structured (generally an optical device for structuring the broadband spectrum) by a spectrum structuring unit (shown in Fig. 2 as an example as a filter unit 11). The structuring unit can act on EUV / X-ray radiation with a spectrally modulated HHG spectrum as well as with a quasi-continuous spectrum.

[0088] In the exemplary embodiment of Fig. 2, the filter unit 11 is arranged in a filter chamber 11A in the beam path 7 and spectrally forms an EUV / X-ray measuring beam 12 from the EUV / X-ray radiation 6. The EUV / X-ray measuring beam 12 comprises at least one, e.g., narrowband frequency component and / or another structured frequency component. The frequency component is, e.g., in its central frequency in the frequency range of the incoming EUV / X-ray radiation 6 is adjustable and is set for a measurement process of the diffractive XCT, ie for the generation of a scattering image.

[0089] For example, the filter unit 11 is designed as a mono- or multi-chromator (e.g. grating-based, grating pair-based, multi-slit-based) with adjustable spectral transmission. A monochromator can spectrally limit the incident broadband EUV / X-ray radiation 6 and transmits, for example, a narrowband frequency component with a spectral narrowband of up to -10000 (X / 5Z). Larger bandwidths can be easily implemented. The filter unit 11 can be controlled with regard to the specific selection of the frequency component to be transmitted; for example, a grating-based monochromator can be adjusted by rotating the central frequency of the transmitted structured frequency range. For example, for a coherent X-ray spectrum, the frequency component can be adjusted in e.g. B. 1 eV steps of 310-430 eV (2.9-4.0 nm) in the range of the water transmission window and for a coherent EUV spectrum the frequency component in e.g.The intensity can be varied in 0.1 eV steps of 50-100 eV (25-50 nm) within the silicon transmission window. A multichromator can spectrally limit the incoming broadband EUV / X-ray radiation 6 to several spectral narrowband frequency components and transmits, for example, each narrowband frequency component with a spectral narrowband of up to -10000 (X / 5X) to the sample under investigation.

[0090] Infrared radiation components may be present in the EUV / X-ray radiation 6 and in the EUV / X-ray measurement beam 12. In the exemplary embodiment shown in Fig. 2, these are filtered from the EUV / X-ray measurement beam 12 (or from the EUV / X-ray radiation 6) using one or more thin filter foils 13 or membranes, including those made of aluminum or zirconium. Filter foils made of other materials can be provided for the respective underlying spectral window. The infrared radiation components can be filtered alternatively and / or additionally using spatial filter apertures and / or optimized multilayer mirrors.

[0091] Fig. 2 further shows a focusing chamber 15A, which is arranged downstream of the filter chamber 11A, by way of example. Alternatively, the focusing chamber 15A can be arranged upstream of the filter chamber (i.e., in the beam path 7 in the region of the EUV / X-ray radiation 6). In the focusing chamber 15A, one (or more) mirrors (generally one (or more) focusing elements) for focusing the EUV / X-ray measuring beam 12 onto the sample P can be located in the beam path 7; exemplary shapes of the mirror typically include paraboloids, ellipsoids, and toroids. In the focusing chamber 15A, For example, a background pressure of approximately 5* 10 7 mbar. In general, a focusing element can be arranged upstream or downstream of the optical device for structuring the broadband spectrum.

[0092] By way of example, in Fig. 2, the EUV / X-ray measuring beam 12 is directed onto the sample P in the focusing chamber 15A by a toroidal mirror 15. An exemplary focal length is f = 1 m with a grazing incidence of 10°. An exemplary numerical aperture (NA) of the toroidal mirror can be in the range of, for example, 0.01-0.1, for example, 0.03. The focal diameter (FWHM) - as an example of a spatially lateral beam expansion parameter of the EUV / X-ray beam - can typically be in the range of approximately 1 pm to one millimeter or, for example, in the range of approximately 10 pm to 500 pm, for example, 10 pm or 100 pm. Thus, under measurement conditions, a surface section of the sample P of, for example, approximately 100 pm 2 until 10000 pm 2irradiated with the spectrally selected, spatially coherent radiation of the EUV / X-ray measuring beam 12. The spatially lateral beam expansion parameter defines a lateral dimension and, accordingly, a lateral frequency component.

[0093] By positioning the irradiated area of ​​the sample in focus, the wavefront with a constant k(i) can be used as the basis for the reconstruction. Alternatively, the position of the irradiated area of ​​the sample can be positioned in front of the focus, for example, to laterally enlarge the irradiated area and reduce the fluence. This can then be considered performing a DXCT with the sample out of focus, which can be taken into account in the reconstruction.

[0094] Taking into account the generally low reflectivity in the EUV / X-ray range, the sample P is preferably irradiated at an oblique angle. For example, the angle of incidence α to the normal of the surface of the sample P can be in the range of 5° to 45°, e.g., 10° or 15°. The angle between the incident EUV / X-ray measuring beam 12 and the backscattered radiation (scattered radiation 14) is 2α and determines the geometric arrangement of a detection chamber 75 A, particularly with respect to the components forming the EUV / X-ray measuring beam 12. The angle of incidence also affects the maximum achievable axial resolution (in the depth direction), since the axial resolution is limited at oblique incidence by the projection and additional refraction upon entering the sample P. Illuminating the sample P at an angle also helps separate the incoming EUV radiation from the outgoing scattered radiation 14.

[0095] The radiation of the EUV / X-ray measuring beam 12 penetrating the sample P is scattered near the surface, for example down to depths of a few micrometers, at the internal structures of the sample P, depending on the absorption in the material. Depending on the absorption, a depth analysis for a near-surface volume fraction of the sample P in the range of, for example, (10x10x2) pm can be carried out within the scope of the diffractive XCT. 3 until (100x100x5) pm 3 be performed.

[0096] To expand the volume fraction to be analyzed, the sample P and the beam path 7 can be adjusted relative to each other using a controllable sample holder (illustrated by the rotation vectors u, v, w indicated in the sample chamber 9 in Fig. 2). With regard to DXCT analysis, the controllable sample holder can be configured to adjust the direction of incidence of the EUV / X-ray measurement beam 12 with respect to a normal of the surface section of the sample P such that the same surface section is irradiated at a different angle of incidence a'. This makes it possible to perform measurements under different reflection conditions, thereby expanding the measurement data acquired and evaluated in Fourier space. (See the explanations for areas B1 and B2 of the Fourier space in Fig. 1.)

[0097] Furthermore, in some embodiments, the sample P can be moved in a (lateral) direction relative to the beam path 7, e.g., orthogonal to the normal of the sample surface or orthogonal to the direction of incidence of the EUV / X-ray measurement beam (illustrated by the x, y, z axes of the sample indicated in the sample chamber 9 in Fig. 2). The EUV / X-ray measurement beam 12 thus impinges on the surface of the sample P at different positions and enters the sample P. This enables DXCT measurements on adjacent, preferably overlapping, lateral regions of the sample P, thereby enabling large-area scanning of a sample. (See also Fig. 9 and the associated description.)

[0098] The scattered radiation 14 reflected by the sample P is detected by a planar detector 17. The detector 17, e.g., a CCD camera with 2048x2048 or 4096x4096 pixels, is arranged in the detection chamber 17A. The detector 17 is designed for measuring photons, e.g., in the energy range from approximately 1 eV to 30 keV, particularly in the EUV and / or X-ray range. For example, an in-vacuum CCD camera MTE 2048B from Princeton Instruments with a large-area, back-illuminated sensor and a quantum efficiency greater than 95% can be used.

[0099] The detector 17 can, for example, be arranged at a distance of, for example, 100 mm from the sample P with a lateral dimension of, for example, 27.6 mm x 27.6 mm and thus cover a solid angle range of approximately 0.076 sr. The detector 17 is positioned, for example, in the far field of the beam geometry of the scattered radiation and detects pixel-related intensity values ​​of the incident scattered radiation 14 during the DXCT measurement. The intensity values ​​are transmitted to an evaluation unit 21 as a scatter pattern for the respectively set, narrow-band structured EUV / X-ray measuring beam 12. A schematic exemplary scatter pattern 19 is shown in Fig. 2 on an input and output device 21A of the evaluation unit 21.

[0100] Additional components of the DXCT measurement setup 1 may include components for beam inspection such as spectrometer modules, mirrors and focus diagnostics, which can be inserted into the beam path as required, e.g., for adjustment.

[0101] The detector 17 provides two-dimensional (pixel-resolved) intensity values ​​(pixel size e.g. 13.5 pm x 13.5 pm). The intensity values ​​represent the lateral spatial frequency components Kx and Ky in the frequency domain of the sample P. If, for example, the EUV / X-ray measuring beam 12 is relatively roughly restricted to a narrowband frequency range of the EUV / X-ray radiation 6 within a selected spectral range or frequency range and / or a detector with a low NA is used, a specific energy component, the frequency coordinate Kz, can typically be assigned to the scattering pattern 19. Otherwise, pixel-specific frequency coordinates Kz can be assigned to the scattering pattern 19, which result from the respective structure and the respective structured frequency range for a pixel during a measurement process. In the scattering pattern 19 indicated in Fig. 2, for example, there are interferences due to the reflected signal from regions of the sample P at different depths, such as a layer structure defined in the nanometer range.

[0102] As explained below, a joint evaluation of a plurality of scattering patterns 19 can be performed in the evaluation unit 21, wherein the scattering patterns 19 were detected during a plurality of measurement processes with different (narrowband) structured frequency ranges; in the example, each of the scattering patterns 19 is assigned an energy component and thus frequency coordinates Kz. From the scattering patterns 19 - as explained in more detail below - a multidimensional characterization of the sample P can be derived and, for example, a layer structure of the sample can be analyzed.

[0103] When evaluating the acquired spectrum, especially in the EUV range, the inevitable absorption of the EUV radiation propagating in the sample P plays a crucial role. The strength of the absorption (and thus the transmission of the radiation to a deeper layer transition) depends on the frequency of the radiation and the materials present in a sample P and must therefore be taken into account during the evaluation.

[0104] The evaluation unit 21 may comprise a central processing unit (CPU or GPU) programmed to execute calculations related to adapting model data to DXCT measurement data by executing instructions stored in program code. The CPU may comprise one or more microprocessors in conjunction with one or more memory elements 21C, as well as the aforementioned input and output device 21A (e.g., a touchscreen or a screen). The memory element 21C may store one or more microprocessor-readable instructions (program code), DXCT measurement data, and model data and provide them to the microprocessor 21B for data processing.For example, the microprocessor 21B can perform a Fourier transform of the DXCT measurement data (in particular, iteratively determine the phase of the detected scattered light, and thus the sample structure in spatial space), derive layers, layer thicknesses, and layer transitions, in particular their lateral dimensions and their positions in the depth direction, extract spectrally resolved reflectivity coefficients, initialize and update model data, model spectrally resolved model reflectivity coefficients based on at least one parameter in a layer transition-specific manner, and optimize model data based on values ​​determined for the parameters. In particular, the microprocessor 21B can, for example, be configured to perform (complex-valued) Fourier transforms in 1D, 2D, and / or 3D in order to reconstruct the sample from stacks of scatter images 19.In addition, the evaluation unit 21 can comprise a controller for controlling various components of the DXCT measurement setup 1 in order to carry out desired actions during the DXCT measurement, such as setting parameters of the EUV / X-ray radiation 6 or the EUV / X-ray measurement beam 12, such as in particular frequency structuring / filtering of the EUV / X-ray radiation 6, in particular setting spectrally structured (narrowband) frequency ranges for the respective scattering images 19, the size selection and position selection of an irradiated surface section, the exposure time of the surface and the selection of the angle of incidence onto the sample P. Furthermore, the evaluation unit 21 can read out, access and / or send measurement data sets of the detector 15 (intensity values ​​of the scattering images) and process the. scatter patterns to a scaled 3D measurement data set / scatter pattern set, etc. By way of example, control connections 23 are shown in Fig. 2, which connect the evaluation unit 21 and the radiation source 3, the filter chamber 11 A, the detector 17, and the sample holder 10 for the exchange of control data.

[0105] For further details of an exemplary DXCT measurement setup 1, reference is made to the aforementioned publications on XCT measurement setups, whose teachings are applicable for a technical implementation of various aspects of a DXCT measurement setup 1. Regarding the newly introduced approach of spectral structuring and the implementation of scattering imaging, reference is made to the explanations provided therein.

[0106] For the rotational extension of the Fourier space to be analyzed, rotation can be carried out using a spatially adjustable sample holder (indicated in Fig. 2 by the rotation vectors u, v, w) and a spatially adjustable detector holder (indicated in Fig. 2 by the rotation vectors a).

[0107] Figures 3A-3C illustrate the DXCT measurement principle using three exemplary measurement processes performed with measurement beams 12(1), 12(2), 12(3). The measurement beams 12(1), 12(2), 12(3) are each limited to different narrowband structured frequency ranges in the optical device 11 for structuring the broadband spectrum of the spatially coherent EUV / X-ray radiation 6. In this example, the spectral structure of the coherent radiation, the beam geometry, and the detection geometry are selected such that an energy-dependent frequency coordinate Kz(1), Kz(2), Kz(3) can be assigned to each of the measurement beams 12(1), 12(2), 12(3). In other implementations, the measuring beams 12(1), 12(2), 12(3) can each be assigned to several energy-dependent frequency coordinates Kz, which must be taken into account when calculating the sample structure.

[0108] As illustrated in Figures 3A-3C, the measuring beams 12(1), 12(2), 12(3) can be a correspondingly spectrally narrow, pulsed laser radiation with different central wavelengths I, 2, X3, which are generated using the HHG and an adjustable filter.

[0109] To generate the corresponding scattering images 19(1), 19(2), 19(3), the measuring beams 12(1), 12(2), 12(3) are directed onto a surface section 31 of the schematically illustrated sample P. The irradiated surface section 31 of the sample P and the illumination direction (Vector k(i), direction of incidence a with respect to normal vector n of the surface section 31) are - as shown in Figures 3A-3C - the same in all three measurement processes, only the structuring of the broadband, spatially coherent EUV / X-ray radiation 6 is adapted to different central wavelengths I, 2, X3.

[0110] According to the DXCT measurement principle, scattering patterns of a sample P are acquired for a plurality of (narrowband) structured frequency ranges (wavelengths / photon energies). Thus, with the detector 17 in Figures 3A-3C, corresponding photon energy-dependent intensity distributions of the backscattered scattered radiation (vector k(s)) are recorded in the far field and output as scattering patterns 19(1), 19(2), 19(3) for further processing. The scattering patterns 19(1), 19(2), 19(3) are an example of a two-dimensional arrangement of scattered radiation intensity values ​​acquired with a detector in the far field. To reconstruct the sample, the photon energy of each scattering pattern can be transferred into Fourier space, e.g., according to the aforementioned projection Kz = (2 cos er) k. As already mentioned, this simplified projection applies, as an example, to the case where the curvature of the Ewald sphere is neglected for the detector surface.If such a simplified projection of a detected scattering pattern ("detector image") onto a respective axial frequency space coordinate Kz is not possible, corresponding Kz values ​​can also be assigned to the recorded intensity values, specific to Kx and Ky. From the intensity values ​​recorded in the measurement processes (i.e., the originally recorded scattering patterns) and the assigned Kz values, Kz-adjusted scattering patterns can be created, in which the intensity values ​​are now each assigned to a Kz value (corresponding to the scattering patterns 19(1), 19(2), 19(3) of the simplified projection). It is noted that the assignment of Kz values ​​can be implemented both as part of generating a multidimensional XCT measurement data set £>(Kx, Ky, Kz) and as a preparatory step in the reconstruction of a multidimensional structure of the sample.Furthermore, the person skilled in the art will recognize that, alternatively, the processing can be carried out based on the detected intensity values ​​taking into account pixel-specific axial frequency space coordinates Kz, i.e., without explicitly forming Kz-adapted scatter patterns, and can be implemented algorithmically.

[0111] To simplify the explanation of the reconstruction of a multidimensional structure of a sample (P) from a multidimensional XCT measurement data set, scattering patterns 19(1), 19(2), 19(3) with a specifically assigned axial frequency space coordinate Kz are assumed below.

[0112] Fig. 4 illustrates the preparation of the photon energy-dependent discrete intensity distributions for further reconstruction. The goal is to combine the scattering patterns 19(1), 19(2), and 19(3) into an XCT measurement data set £>(Kx, Ky, Kz) – also referred to herein as scaled scattering pattern stack 33. The scaled scattering pattern stack 33 is prepared for sample reconstruction via Fourier transformation, which uses an identical discretization of the measurement data (here, the intensity values ​​of the scattering patterns) in the Kx and Ky directions for all Kz values.

[0113] As an intermediate step, Fig. 4 shows an unscaled scatter pattern stack 35 in which the measured scatter patterns 19(1), 19(2), and 19(3) are combined along the energy-frequency component Kz, thus filling the three frequency axes with intensity values. The Kx and Ky values ​​correspond, pixel-dependently, to the intensity values ​​acquired with the detector 17.

[0114] As part of a scaling step 36, the photon energy-dependent scattering patterns in the scaled scattering pattern stack 33 of Fig. 4 are scaled based on a common photon energy in the lateral frequency coordinates Kx, Ky. Fig. 4 shows scattering patterns 19(1)', 19(2)', 19(3)' scaled accordingly in the lateral frequency coordinates Kx, Ky. The scaling is based on the fact that the scattering (diffraction) at a sample structure is directly proportional to the illumination wavelength (central wavelength 1, 2, 3). In the scaled scattering patterns 19(1)', 19(2)', 19(3)', the increments in the Kx and Ky directions underlying the intensity values ​​are adjusted to each other for all Kz values ​​(scattering patterns). This means that the scaled resolution 5K_x;y for the lateral spatial frequency Kx or Ky is adjusted for the scattering patterns 19(1)', 19(2)', and 19(3)'. Scaling results in a reduction of the scattering patterns at low photon energies (Kz values). The scattering patterns are, for example,in an energy range from 310 eV to 430 eV and scaled in each case such that: 5K_x;y (310 eV) = 5K_x;y (430 eV). In this case, all scattering patterns are scaled to the highest photon energy (430 eV), as shown in Fig. 4. Thus, in the example of Fig. 4, the scattering pattern 19(1)' remains unchanged at the highest energy and corresponds to the scattering pattern 19(1) of the unscaled scattering pattern stack 35.

[0115] With a constant discretization in the rescaled scattering images, the spatial frequencies { K_xy} = pixel —> qm 1 be energetically calibrated. For the calibration, 5K_x;y = p / (zk) applies. For the example case of a detector 17 with a pixel array, the pixelation (pitch p) of the detector 17 is, for example, p = 13.5 pm. With a distance of the detector 17 to the sample P of 230 mm and a wavelength of the measuring beam 12 of = 2.88 nm (430 eV) a detector pixel has a lateral resolution of the spatial frequency of 5K_x;y = 0.0204 qm 1 assigned. An exemplary pixel array of 201 ^201 pixels allows a bandwidth of AK xy = 4.08 qm 1 (full width, FW). The addressable spatial frequencies in the example case are accordingly { K_xy} = [-2.04; 2.04] qm 1 Due to the high photon energies, the depth resolution is 5K_z = 0.8066 qm 1 and the bandwidth Ak_z = 96.8 qm '. The measurable spatial frequencies in the depth direction are { K_z} = [250.0; 346.8] qm '. Although the spectral resolution in the axial direction 5K_z differs from the spectral resolution in the lateral direction 5K_xy, this does not affect subsequent reconstruction, because a different axis scaling ~Ki = cKi in Fourier space merely causes an inverse scaling ~xi = xi / c of ​​the corresponding axis i in position space due to the Fourier transform, where c is a factor. The inverse scaling is advantageous for reconstruction, since an identical discretization of the lateral and axial components 5k_xy == 5k_z requires the use of very large 3D matrices (entries in #Kx x #Ky x #Kz > 201 x 201 x 4745) and can significantly increase the computational effort.

[0116] As explained in connection with Fig. 1, with a low NA of the detector of, for example, 0.01 and a coarse discretization of the photon energies (Kz discretization / increment), the consideration of a spherical curvature of the “cuts” in the Fourier space can be dispensed with even in the case of using a planar detector 17.

[0117] The result of the scaling step is a compilation of measured intensity values ​​with uniform Kx / Ky / Kz increments in the three dimensions of the frequency components Kx, Ky, Kz. These values ​​can be read as an XCT measurement data set £>(Kx, Ky, Kz) by a computer to execute a Fourier transform algorithm that converts the multidimensional XCT measurement data set £>(Kx, Ky, Kz) into a sample structure data set O(x, y, z). In other words, a three-dimensional distribution or "cloud" of intensity values ​​is present in three-dimensional Fourier space, which can be used as a result of the XCT measurement to reconstruct the three-dimensional depth structure.

[0118] For a three-dimensional Fourier data set acquired using diffractive XCT (XCT measurement data set £>(Kx, Ky, Kz)), the evaluation includes an algorithmic derivation of the phase of the backscattered radiation. The derived phase components for the three Fourier axes / frequency domain coordinates Kx, Ky, Kz result in a sample structure O(x, y, z) along the spatial axes x, y, z in the spatial domain.

[0119] The algorithmic implementation of the reconstruction of the sample structure 0(x, y, z) includes phase extraction and Fourier transformation(s): £>(Kx, Ky, Kz) —> Phase extraction / Fourier transformation(s) —> 0(x, y, z)

[0120] As recognized by the inventors, the reconstruction can be performed sequentially, e.g., first for the axial Fourier axis (frequency domain coordinate Kz) and then for the lateral Fourier axis(es) (frequency domain coordinates Kx and Ky), or simultaneously, i.e., in parallel, for all three Fourier axes (frequency domain coordinates Kx, Ky, Kz). An exemplary sequential reconstruction is described below in connection with Figures 5 and 6. An exemplary simultaneous reconstruction of all three Fourier axes is described below in connection with Figures 7 and 8.

[0121] In the flowchart of Fig. 5, the previously explained measurement processes and data processing steps of the measurement data are summarized in dashed boxes 41 and 43. Furthermore, an exemplary sequential 1D-2D evaluation of the measurement data £>(Kx, Ky, Kz) - first one-dimensional in the depth direction and then two-dimensional in the lateral directions - is illustrated in the dashed boxes 45A and 45B, whereby the measurement data £>(Kx, Ky, Kz) are based on scattering patterns obtained by means of diffractive XCT using spectrally structured EUV / X-ray measurement beams 12.

[0122] In Fig. 5, a cycle of DXCT measurement processes 49 for obtaining DXCT scattering images is illustrated in the bottom left corner of box 41. For a plurality of specifically spectrally structured EUV / X-ray measurement beams 12 from a radiation source 3, one- or two-dimensional intensity distributions of the EUV / X-ray radiation 14 backscattered by the sample into the solid angle of the detector 17 are recorded in 1 to N measurement processes with the detector 17.

[0123] In Fig. 5, a sample is shown schematically in perspective in a three-dimensional layer structure 51 in spatial space (x, y, z). The scattering of the EUV / X-ray measuring beam 12 emanating from the radiation source 3 is illustrated using an exemplary sectional view 53 of the sample. The spatially coherent EUV / X-ray measuring beam 12 can be spectrally structured for the N measuring processes, so that the EUV / X-ray measuring beam 12 can be adjusted for several narrowband structured frequency ranges, e.g., o = o 1 to o = oN (step 55) and in this way N spectrally resolved DXCT measuring processes 49 can be carried out (iteratively). For the exemplary sectional view 53, the EUV / X-ray measuring beam 12 is scattered by two layer-like structures 53A, 53B in the sample.

[0124] Complex reflectivities of the layered structures 53A, 53B underlie the scattering processes in the sample and lead to the illumination of the laterally resolving detector 17 (1D / 2D pixel arrangement) arranged in the far field with the backscattered radiation (scattered radiation 14). The complex reflectivities can be described in the form of spectrally resolved reflectivity coefficients r(Kx, Ky; CD) and assigned to the detector pixels. For each measurement process (for co = co 1 to CD = CDN), the detector 17 measures an intensity scattering pattern, i.e., squares of the magnitude of the scattered field, which in the example of Fig. 5 each represent a two-dimensional distribution of the measured intensity with intensity values ​​R(Kx, Ky; CD 1 . . . CDN).In other words, detector 17 detects intensities of the diffracted EUV / X-ray radiation (scattered radiation 14) at a plurality of positions in the far field, whereby the spectral phase at the respective measurement location (detector pixel) is unknown. The scattering images 19 are shown in Fig. 5 in exemplary sectional views as pixel arrays of various gray tones. The size of detector 17 corresponds to the maximum numerical capacity NA, which in turn determines the maximum lateral resolution.

[0125] As discussed in connection with Fig. 4 and illustrated in box 43 of Fig. 5, the scattering patterns can be combined into an unscaled scattering pattern stack 35. In preparation for reconstruction, the roots of the measured intensity values ​​- data points (R(Kx, Ky; CD1 . . . CDN)) - can be extracted (step 57). Furthermore, in step 59, the energy values ​​(CD1 to CDN) of the scattering patterns are converted into Kz values ​​(Kzl to KzN) and a uniform scaling in kx and ky is introduced for the Fourier transform - data points (R(Kx, Ky; Kzl . . .KzN)). Thus, XCT measurement data £>(Kx, Ky, Kz) are available for reconstruction, which correspond to the spectral (energy-dependent) reflectivities r in the frequency domain without phase information.

[0126] Although the known intensity reflectivity R(K) = |r(K)| 2determine an amplitude of the field reflectivity |r(K)| = (R(Kx, Ky; Kzl . . .KzN)), however, for the XCT analysis (derivation of the deep structure) without knowledge of the phase information, the interferences between the buried layers and those relative to the sample surface remain indistinguishable, with the corresponding presence of artifacts.

[0127] To reconstruct the sample structure, the phase e (in Fourier space) or e 10(z) (in the spatial space), whereby the reflectivities in the spatial space r(z) e 10(z) and in Fourier space r(k) e are completely (especially artifact-free).

[0128] Phase reconstruction methods typically work iteratively by repeatedly executing a loop consisting of Fourier transformation, inverse Fourier transformation, and specifying the necessary associated constraints in Fourier space and position space. These constraints are chosen specifically to effect / accelerate the convergence of the phase reconstruction. For example, algorithms such as the Gerchberg-Saxton algorithm and / or the hybrid input-output feedback algorithm can be used. The above basic principle of convergence generally applies to phase reconstruction in ID, 2D, or 3D. For XCT, particularly for a three-stage algorithm for one-dimensional phase reconstruction in XCT, reference is made to the scientific publications mentioned above for examples of phase reconstruction and derivation of the deep structure.

[0129] In the serial approach shown in Fig. 5, the sample structure O(x, y, z) is reconstructed step by step. The reconstruction, explained in detail below (Boxes 45A, 45B), takes advantage of the fact that a pixel-specific intensity reflectivity "R(o)" can be derived from a plurality of scatter patterns for a pixel (given by Kx and Ky). The serial approach thus initially reconstructs the axial structure of the scatter patterns in a sub-space (Kx, Ky, z) pixel by pixel from the XCT measurement data £>(Kx, Ky, Kz), which are given in 3D Fourier space (Box 45A). In this case, the well-known PR-XCT algorithm, for example, can be applied point by point, i.e., individually for each point (Kx, Ky) in the scatter patterns.In a first step, a one-dimensional phase extraction can be applied to the stack of scaled scattering images for a single lateral data point to obtain phase information of the third Fourier component (axial frequency-space coordinate Kz). The result of the calculation is depth-resolved scattering images / diffractograms (20(1), 20(2)), i.e., partially reconstructed scattering images that can be combined to form a partially reconstructed sample structure O'(Kx', Ky', z) in the subspace space (Kx, Ky, z). In a second step, the lateral structure of the sample structures O(x, y, z) can be reconstructed from the depth-resolved scattering images / diffractograms (20(1), 20(2)) using a PR-2D algorithm for each depth z individually (Box 45B).

[0130] An advantage of the serial reconstruction approach can be the independent use of two established algorithm approaches and the decomposition of the 3D reconstruction into lower-dimensional problems. The serial reconstruction approach can be particularly applied when layered samples are present, allowing separate calculation of axial and lateral structure, as they are self-referencing and provide sufficient information about the axial structure at the individual points of the detector 17. However, the serial reconstruction approach does not take advantage of the redundant information present in the 3D Fourier space, which can reduce the convergence of the reconstruction and increase susceptibility to noise.

[0131] The first step of the serial reconstruction approach (XCT reconstruction step), implemented as an example in the dashed box 45 A, can be represented as follows: £>(Kx, Ky, Kz) XCT-PR / axiale Structure O(Kx, Ky, Kz) et^ ^Kx ,Ky)}

[0132] In Box 45A, the repeated execution of Fourier transformations and inverse Fourier transformations, and thus the switching between Fourier space KS (Fourier space reflectivities KS-r) and sub-location space OS' (sub-location space reflectivities OS'-r), is illustrated by arrows 61. The starting point is the scaled XCT measurement data set £>(Kx, Ky, Kz) - in Box 45A: KS-r, to which a random phase 63 for Kz is initially applied for the Fourier transformation into the sub-location space OS' (result: OS'-r (Kx, Ky, z) 65). In the sub-location space OS', for example, constraints 67 for z are introduced for the inverse transformation into the Fourier space KS (intermediate result: OS'-r' (Kx, Ky, z) 69). In the Fourier space, a new KS-r 71 results, to which, for example, constraints 73 for Kz are introduced for the next transformation into the subspace OS' (intermediate result KS-r' (Kx, Ky, Kz) 75).After m runs, one-dimensionally reconstructed depth-resolved scattering patterns / diffractograms O'(Kx, Ky, z) 69A are obtained in the subspace OS'. These represent the input of the subsequent two-dimensional reconstruction in box 45B, from which the sample structure O(x, y, z) 89A results.

[0133] Example parameters of the one-dimensional phase recovery algorithm include 2 x 120 iterations of HIO in the first step, 500 iterations of HIO in the second step, and 500 iterations of GS in the third step. An example discretization of the scattering patterns is 0.2 pm per pixel.

[0134] With regard to the one-dimensional evaluation step of the measurement data £>(Kx, Ky, Kz) outlined in Box 45 A, it should be noted that OCT / XCT evaluation methods are known that allow a precise reconstruction of the axial layer structure for complex interferences (see, for example, the publication by S. Fuchs et al. mentioned above). Using so-called "phase retrieval" (PR)-XCT algorithms, the axial structure of the sample and the complete spectral phase are derived from XCT measurement data generated with broadband EUV radiation by reconstructing the scattering process. The phase contains, among other things, information about the absorption and dispersion of the materials present in the sample, as well as about the incidence geometry and the refractive index of the dominant material in a layer.In order to derive the axial layer structure of the sample from the reflected spectrum, the XCT evaluation method uses a multi-stage algorithm, for example, to perform a one-dimensional Fourier transformation (frequency domain coordinates Kz) of the acquired spectrum into the spatial domain (here the depth direction z).

[0135] The second step of the serial reconstruction approach (lateral reconstruction step), implemented as an example in the dashed box 45B, can be represented as follows: O'(Kx, Ky, z) “CDI” PR / lateral structure FT / Kx Ky^ O(x, y, z) e ielateral «A z))

[0136] In the second step, a two-dimensional phase calculation, e.g., similar to a CDI phase reconstruction, can be performed separately for each z to reconstruct the lateral structure.

[0137] In Box 45B, the repeated execution of Fourier transformations and inverse Fourier transformations, and thus the change between Fourier space KS (Fourier space reflectivities KS-r) and position space OS (position space reflectivities OS-r), is illustrated by arrows 81. The starting point for the reconstruction is the reconstructed depth-resolved scattering patterns / diffractograms O'(Kx, Ky, z) 69A located in the sub-position space OS'-r (see Box 45A). These diffractograms are fed to the two-dimensional phase calculation (Box 45B in Fig. 5) as KS-r (form: KS-r (Kx, Ky, z) 95). Initially, a random phase 83 for Kx and Ky is generated for the Fourier transformation into the position space OS (result: OS-r (x, y, z) 85). In the position space, constraints 87 for x and y are introduced for the inverse transformation into the Fourier space KS (intermediate result: OS-r' (x, y, z) 89). In the Fourier space orSublocal space results in a new KS-r (form: KS-r (Kx, Ky, z) 91), to which constraints 93 for Kx and Ky are introduced for the next Fourier transformation into the local space OS. (Intermediate result: KS-r' (Kx, Ky, Kz) 95). As mentioned, algorithms for two-dimensional reconstruction in the second step can be derived, for example, from CDI algorithms. After n iterations, a three-dimensional reconstruction of the structure O(x, y, z) 89A of the three-dimensional sample is obtained in the position space OS.

[0138] Algorithms for phase reconstruction, as known from XCT for a one-dimensional data set and outlined above, can be extended analogously to two- or three-dimensional data sets. These variants are referred to herein as phase reconstruction in 2D (PR2D) or in 3D (PR3D), whereby phase reconstruction in ID (PR1D) refers to the well-known XCT analysis. These algorithms differ fundamentally in the dimensionality of the Fourier transform: ID, 2D, or 3D. Otherwise, they are almost identical in the implementation of the boundary conditions (constraints). Although reconstruction for three-dimensional samples may be more complex because the system of equations is larger due to the cube, in principle better results (faster) can be achieved due to additional redundant information. Depending on the sample, different algorithms can be used for phase reconstruction, such as:The Gerchberg-Saxton (GS) or Error Reduction (ER) algorithm, the Hybrid Input-Output (HIO) algorithm, the Hybrid Projection-Reflection (HPR) algorithm, or the Relaxed Averaged Alternating Reflection (RAAR) algorithm, a combination of these, or even KL algorithms, can be used. Example parameters of the two-dimensional phase recovery algorithm include 3000 iterations of a feedback RAAR algorithm.

[0139] Fig. 6 summarizes the sequential reconstruction of the sample structure. Starting from the originally measured (Fourier) data set £>(Kx, Ky, Kz) of the scattering images, a one-dimensional phase reconstruction (PR-OCT algorithm) is performed for each lateral data point (Kx, Ky) (XCT reconstruction step 97A). This results in depth-resolved scattering images / diffractograms O'(Kx, Ky, z), in other words, partially (i.e., in Kz) reconstructed scattering images. Similar to a two-dimensional CDI phase recovery algorithm, a lateral reconstruction of the depth-resolved scattering images / diffractograms O'(Kx, Ky, z) can then be performed to calculate the sample structure O(x, y, z) (lateral reconstruction step 97B). The depth-resolved diffractograms (20(1), 20(2)) shown in Fig. 6 represent, for example, diffractograms of the depth position z contributing to the sample structure data set O(x, y, z) identified according to a signal evaluation step.

[0140] As the inventors have recognized, certain samples may have advantages that simplify and / or accelerate the second step of reconstruction. Due to the sequence of reconstruction – initially one-dimensional, subsequently two-dimensional – axial structures are reconstructed first. This allows spectral modulations to be analyzed to obtain deep structure information. Since the depth profile was obtained for each lateral position (Ky, Ky), the signal strength corresponding to a z-value (depth value) is an indication of the presence of lateral structures of the sample in this z-range (slice information). In conclusion, there is no lateral structure (slice information) at depth values ​​that have no significant or low signal contribution or signal noise.This allows limiting the lateral reconstruction to the depth values ​​with signal (significant signal contribution) and thus leads to a faster calculation of the object structure.

[0141] Furthermore, axial phase reconstruction automatically shifts the modulation-free signal to the surface at z = 0. This frees the depth-resolved scattering patterns / diffractograms O'(Kx, Ky, z) from the intense incident and low-modulation radiation. At the same time, noise, which also lacks a specific spectral modulation frequency, is shifted to the surface at z = 0. These aspects can significantly increase the contrast in the depth-resolved scattering patterns / diffractograms O'(Kx, Ky, z) and simplify the lateral reconstruction of the sample in the second step.

[0142] For the sake of completeness, it is mentioned that it is also possible to swap the order of reconstruction using PR-XCT and PR2D and this can be advantageous, for example, for particularly laterally structured samples.

[0143] The parallel reconstruction approach uses parallel computation in all three dimensions instead of serial computation. In this approach, the axial and lateral structures are determined simultaneously using a PR3D algorithm, rather than sequentially. This reconstruction approach is particularly applicable to samples with unknown structures, as well as stamp-like or self-referencing structures.

[0144] To reconstruct the sample structure, a combination of a RAAR algorithm and a HIO algorithm with a total of 3000 iterations can be used. For example, the three-dimensional data set £>(Kx, Ky, Kz) of the scattering images can be initially processed with 128 A3 entries using “zero padding” e.g. to 256 A 3 entries The feedback in the HIO algorithm, for example, is ß = 0.75. The threshold for a so-called shrink wrap of an applied aperture is, for example, 6%. The remaining parameters can be based on those of the serial reconstruction approach.

[0145] The data can be further cleaned for parallel reconstruction. The illumination beam k(i) reflected from the surface of the sample superimposes the actual scattering images (Kx, Ky, Kz) as a specular reflection at (Kx, Ky) ~ 0 and can thus interfere with the measured signal of the actual diffraction or scattering on the sample. Using a circular mask, for example, an inner region of the measured data at (Kx, Ky) ~ 0 can be set to zero, for example, which avoids reconstruction of the dominant illumination function or the illumination beam k(i). (Please note that such a correction is not necessary for serial reconstruction, since the axial PR-XCT algorithm automatically shifts modulation-free components of the depth reconstruction into the DC component or into the surface at z = 0.)Depth-resolved scattering images / diffractograms O'(Kx, Ky, z) are cleaned of the illumination beam k(i) and freed from the background in the serial approach.

[0146] In box 99 of Fig. 7, the repeated execution of Fourier transforms and inverse Fourier transforms, and thus the switching between Fourier space KS (Fourier space reflectivities KS-r) and position space OS (position space reflectivities OS-r), is illustrated by arrows 101 for the parallel reconstruction approach. As with serial reconstruction, the starting point is the scaled XCT measurement data set £>(Kx, Ky, Kz). The scaled XCT measurement data set £>(Kx, Ky, Kz) can be obtained, for example, through a plurality of measurement processes, as shown in boxes 41 and 43 in conjunction with Fig. 5. For the corresponding description, reference is made to the preceding description of Fig. 5. Box 99 shows the reflexivity in the Fourier space KS-r, to which a random phase 103 for Kx, Ky and Kz is initially applied during the Fourier transformation into the position space OS (result: OS-r(x, y, z) 105).In the position space, constraints 107 for x, y, and z are introduced for the inverse transformation into the Fourier space KS (intermediate result: OS-r'(x, y, z) 109). In the Fourier space, a new KS-r 111 results, onto which constraints 113 for Kx, Ky, and Kz are introduced for the next transformation into the position space OS (intermediate result: KS-r'(Kx, Ky, Kz) 115). As mentioned, algorithms for three-dimensional reconstruction can be derived, for example, starting from the one-dimensional PRXCT algorithm or the two-dimensional CDI algorithm. n runs result in a three-dimensional reconstruction of the structure O(x, y, z) of the three-dimensional sample in the position space OS.

[0147] Fig. 8 summarizes the parallel reconstruction. Starting from the original measured (Fourier) data set £>(Kx, Ky, Kz) of the scattering images, a three-dimensional phase reconstruction (PR3D algorithm) is performed for each data point (Kx, Ky, Kz), resulting in the sample structure O(x, y, z). The parallel approach directly reconstructs the 3D structure O(x, y, z) of the sample from the known 3D Fourier space O(Kx, Ky, Kz) without any intermediate step using the PR3D algorithm (3D reconstruction step 117). The parallel approach has the advantage that even more complex, especially non-layered, samples can be reconstructed. Furthermore, the parallel approach can exploit redundancies for sample reconstruction within the 3D Fourier space. A prerequisite for parallel reconstruction is an evaluation unit that can handle the high computational effort required by 3D calculation and the use of 3D matrices.

[0148] In the comparison between serial and parallel reconstruction, for example, a larger area of ​​the lateral structure, i.e. a larger field of view, can be reconstructed with the serial approach if there is a higher or better diffraction contrast of the depth-resolved scattering images / diffractograms O'(Kx, Ky, z) in the serial reconstruction compared to the original 3D Fourier space O(Kx, Ky, Kz) in the parallel reconstruction. A better diffraction contrast and the resulting improved spectral resolution 5K_x or 5K_y can result in a larger field of view Ax ' ~ l / 5K_x or Ay ' ~ l / 5K_y in the spatial domain for the serial approach. The original measurement data O(Kx, Ky, Kz) are automatically cleaned up in the serial approach by applying the PR-XCT algorithm.A false correlation or a noisy background, which cannot exhibit a specific spectral modulation, is shifted by the PR-XCT algorithm into the DC peak by its Fourier transform, whereby the depth-resolved scattering images / diffractograms O'(Kx Ky, z) are corrected compared to the original 3D Fourier space O(Kx , Ky , Kz) and the diffraction contrast increases.

[0149] Previously, the concepts disclosed herein were explained in detail for three-dimensional XCT measurement data sets. As mentioned at the beginning, the concepts disclosed herein can also be applied to the generation and evaluation of two-dimensional XCT measurement data sets. In this case, the positions of the intensity values ​​are specified in a two-dimensional frequency space (Kx or Ky, Kz), i.e., they are determined by a lateral frequency-space coordinate (Kx or Ky) and, on the other hand, an axial frequency-space coordinate (Kz). For two-dimensionally structured samples, for example, the acquisition of a one-dimensional scattering pattern with a line detector can provide sufficient lateral structural information, which can be derived together with the depth information. The description of the reconstruction, as explained by way of example in connection with Figures 5 to 8, can be adapted and implemented accordingly with regard to the lateral frequency-space coordinates for the reconstruction of a single lateral frequency-space coordinate (a one-dimensional scattering pattern).

[0150] The explanations in connection with the section of the Ewald sphere shown in Fig. 1 also already illustrate a 2D scattering analysis, as can be carried out, for example, for a symmetric sample with a line detector according to the concepts disclosed herein.

[0151] In general, EUV / X-ray sources with higher spatial coherence allow for the laterally investigation and reconstructing of larger areas of the sample.

[0152] Furthermore, additional areas of the sample can be examined and reconstructed if different areas of the sample are coherently illuminated, for example, using a scanning method, and these areas partially overlap coherently. This is known as ptychography.

[0153] Fig. 9 illustrates the previously mentioned approach of expanding the XCT measurement data (exemplarily in the form of scatter pattern stacks) by scanning multiple surface sections 120, 130, 140 of the sample. The surface sections 120, 130, 140 are shown in two lateral dimensions x and y, along with the schematically indicated sample depth in the z-direction (examined volume regions 122, 132, 142). Each examined surface section 120, 130, 140 is assigned a scatter pattern stack 124, 134, 144.

[0154] The scanning is carried out, for example, as follows: After the iterative generation of the scatter images has been carried out on a first surface section 120 with the reference point xy_l and a scatter image stack 124 has been generated for the corresponding volume area 122 of the sample, a displacement of the sample, for example, to the reference point xy_2 is carried out for an irradiation of the volume area 132. In Fig. 9 one can see a partial overlap of the surface sections 120 and 130; accordingly, the volume regions 122 and 132 and the scatter pattern stacks 124 and 134. As an example, in Fig. 9, the displacement occurs in both directions x and y. A further displacement enables the generation of the scatter pattern stack 144 for the reference point xy_3 and the volume region 142.

[0155] Based on the intensity values ​​of the scatter image stacks 124, 134, 144, a common XCT measurement data set can then be generated and evaluated for the entire scanned surface area. In other words, the EUV / X-ray measurement beam can be irradiated onto a plurality of spatially laterally extended, adjacent, or partially overlapping surface sections of the sample. The recorded intensity values ​​of the scatter image stacks assigned to the surface sections can be combined to form the three-dimensional XCT measurement data set £>(Kx, Ky, Kz).

[0156] As an alternative to the sequential generation of adjacent, contiguous or, in particular, partially overlapping, scatter image stacks, scanning can be carried out in one or two directions, each for a set (narrowband) structured frequency range, so that extended scatter images are sequentially acquired for the majority of (narrowband) structured frequency ranges and subsequently combined into a common scatter image stack.

[0157] It is explicitly emphasized that all features disclosed in the description and / or the claims are to be considered separate and independent of each other for the purpose of the original disclosure as well as for the purpose of limiting the claimed invention, regardless of the feature combinations in the embodiments and / or the claims. It is explicitly stated that all range specifications or specifications of groups of units disclose every possible intermediate value or subgroup of units for the purpose of the original disclosure as well as for the purpose of limiting the claimed invention, in particular also as a limit of a range specification.

Claims

Patent claims 1. Method for generating a multidimensional XCT measurement data set (£>(Kx, Ky, Kz)) of a sample (P) using spectral multistructured scattering imaging with the steps: - Providing EUV / X-ray radiation (6) with a broadband spectrum in the EUV to X-ray spectral range, - generating a plurality of scattering images (19), wherein each of the plurality of scattering images (19) is generated by a measuring process comprising the steps of: a) structuring the broadband spectrum of the EUV / X-ray radiation (6) as a structured frequency range to provide an EUV / X-ray measuring beam (12) with a measuring beam spectrum comprising at least one of a plurality of narrowband structured frequency ranges, wherein at least one axial frequency space coordinate (Kz) in the frequency space (Kx, Ky, Kz) is assigned to each of the narrowband structured frequency ranges, b) irradiating the EUV / X-ray measuring beam (12) onto a spatially laterally extended surface section of the sample (P), c) detecting scattered radiation which is backscattered by the sample (P) with a detector (17) which outputs intensity values ​​for the structured frequency range belonging to the measuring process, to which intensity values ​​are assigned in each case at least one lateral frequency space coordinate (Kx,Ky) in the frequency domain (Kx, Ky, Kz), the intensity values ​​forming the scatter pattern (19) for the associated measuring process; and, - Combining the intensity values ​​to form the multidimensional XCT measurement data set (£>(Kx, Ky, Kz)), whereby the positions of the intensity values ​​in the frequency space (Kx, Ky, Kz) are given by at least one lateral frequency space coordinate (Kx, Ky) and one axial frequency space coordinate (Kz).

2. Method according to claim 1, wherein the scattering patterns (19) are detected in a plurality of measuring processes with different structured frequency ranges, and / or wherein the plurality of scattering patterns (19) are correspondingly recorded for a plurality of narrowband structured frequency ranges, in particular for a plurality of central wavelengths.

3. Method according to claim 1 or 2, wherein the EUV / X-ray measuring beam (12) is spectrally structured differently for N measuring processes, in particular is set for several narrow-band structured frequency ranges in order to carry out N measuring processes with different spectral resolution.

4. Method according to one of claims 1 to 3, wherein the broadband spectrum is structured by a monochromator diffracting a narrowband frequency range from the EUV / X-ray radiation (6), which propagates as the EUV / X-ray measuring beam (12) along an optical beam path (7) to the sample (P).

5. Method according to one of claims 1 to 4, wherein the scattered radiation is detected by detecting the scattered radiation of an associated structured frequency range reflected by the sample in one lateral dimension with a line detector or in two lateral dimensions with a planar detector and outputting it as a one-dimensional or two-dimensional scattering image.

6. The method according to any one of claims 1 to 5, wherein the at least one lateral frequency space coordinate (Kx, Ky) is determined by parameters of the EUV / X-ray measuring beam and parameters of the measuring process and / or wherein the axial frequency space coordinate (Kz) is given by the photon energy of the radiation detected at a lateral frequency space coordinate (Kx, Ky).

7. The method according to any one of claims 1 to 6, wherein the intensity values ​​of a scatter pattern of a measurement process are projected onto an axial frequency space coordinate assigned to the measurement process, in particular onto a mean axial frequency space coordinate assigned to the respective measurement process, for example given by: Kz = (2 sin a) k; where k is a wave vector assigned to a narrow-band structured frequency range, and a is an angle of incidence of the EUV / X-ray measurement beam (12) to the surface section of the sample (P).

8. Method according to one of claims 1 to 6, wherein each of the intensity values ​​of a scatter pattern of a measuring process is projected onto an axial frequency space coordinate (Kz) of a plurality of axial frequency space coordinates, in particular depending on the at least one lateral frequency space coordinate of each of the intensity values, and / or wherein, based on projected axial frequency space coordinates (Kz), axially uniform scattering patterns are compiled, which comprise intensity values ​​from scattering patterns acquired in different measuring processes at the same axial frequency space coordinate (Kz).

9. Method according to one of claims 1 to 6, wherein for at least one measuring process, a plurality of axial frequency space coordinates are assigned to the associated structured frequency range, and in one measuring process, intensity values ​​for a plurality of scattering patterns are obtained by assigning the intensity values ​​to respectively associated axial frequency space coordinates of a narrow-band structured frequency range.

10. The method according to one of claims 1 to 9, wherein a first subgroup of measuring processes are carried out at a first relative position between the incident EUV / X-ray measuring beam (12) and the spatially laterally extended surface section of the sample (P) and a second subgroup of measuring processes are carried out at a second relative position between the incident EUV / X-ray measuring beam (12) and the spatially laterally extended surface section of the sample (P), and furthermore the multidimensional XCT measurement data set (£>(Kx, Ky, Kz)) is formed from the intensity values ​​of the first subgroup and the second subgroup in the frequency space (Kx, Ky, Kz), in particular taking into account the axial frequency space coordinates (Kz).

11. Method according to one of claims 1 to 10, wherein the EUV / X-ray measuring beam (12) is irradiated onto a plurality of spatially laterally extended, adjacent or partially overlapping surface sections of the sample (P) and the recorded intensity values ​​of the scattering images assigned to the surface sections are combined together to form the three-dimensional XCT measurement data set (£>(Kx, Ky, Kz)).

12. The method according to any one of claims 1 to 11, wherein the plurality of scatter images combined as a two-dimensional Fourier measurement data set or as a three-dimensional Fourier measurement data set form the multi-dimensional XCT measurement data set (£>(Kx, Ky, Kz)).

13. Computer-implemented method for reconstructing a multidimensional structure of a sample (P) from a multidimensional XCT measurement data set (£>(Kx, Ky, Kz)) comprising the steps: - Reading, into a processor (21B), intensity values ​​of a multidimensional XCT measurement data set (£>(Kx, Ky, Kz)), whereby positions of the intensity values ​​in the frequency domain (Kx, Ky, Kz) are given by at least one lateral frequency space coordinate (Kx, Ky) and one axial frequency space coordinate (Kz), and - with the processor (21B), executing a Fourier transform algorithm which converts the multidimensional XCT measurement data set (£>(Kx, Ky, Kz)) into a sample structure data set (O(x, y, z)), wherein the sample structure data set (O(x, y, z)) is in a position space (x, y, z) comprises a depth direction coordinate (z) of the sample (P) which is assigned to the axial frequency space coordinate (Kz), and at least one lateral coordinate (x, y) of the sample (P) which is assigned to the at least one lateral frequency space coordinate (Kx, Ky).

14. The computer-implemented method of claim 13, further comprising - a scaling step in which, for each of the plurality of axial frequency space coordinates (Kz) of the frequency space (Kx, Ky, Kz), intensity values ​​in the at least one lateral frequency coordinate (Kx, Ky) are scaled based on a common photon energy, so that in particular the increments in the at least one lateral frequency coordinate (Kx, Ky) underlying the multidimensional XCT measurement data set (£>(Kx, Ky, Kz)) are adapted to one another for all axial frequency space coordinates (Kz), and / or - a field calculation step in which the square root of each intensity value is taken.

15. Computer-implemented method according to claim 13 or 14, wherein the Fourier transform algorithm is configured to determine a phase of the detected scattered light by means of Fourier transforms, wherein in particular the phase is reconstructed iteratively by repeatedly executing a loop with at least one Fourier transform, at least one inverse Fourier transform and setting boundary conditions in the Fourier or position space for the Fourier transform and / or the inverse Fourier transform.

16. A computer-implemented method according to any one of claims 13 to 15, wherein the Fourier transform algorithm comprises - an XCT reconstruction step (97 A) for one-dimensional phase reconstruction in a depth direction coordinate (z) of the spatial space (x, y, z), which, based on the axial frequency space coordinate (Kz) for lateral positions (Kx, Ky), derives a partially reconstructed sample structure (O'(Kx', Ky', z)) which comprises a plurality of depth-resolved diffractograms (20(1), 20(2)), wherein the XCT reconstruction step comprises in particular an ID phase reconstruction XCT algorithm, and - a lateral reconstruction step (97B) for one- or two-dimensional phase reconstruction in at least one lateral coordinate (x, y), which derives the sample structure data set (O(x, y, z)) based on the at least one lateral frequency space coordinate (Kx, Ky) and at least one subgroup of the plurality of depth-resolved diffractograms (20(1), 20(2)).

17. The computer-implemented method of claim 16, further comprising - a signal evaluation step in which cumulative signal contributions of the depth-resolved diffractograms (20(1), 20(2)) are calculated in order to identify depth positions (z) contributing to the sample structure data set (O(x, y, z)), and wherein the subset of the plurality of depth-resolved diffractograms (20(1), 20(2)) for the lateral reconstruction step comprises the depth-resolved diffractograms (20(1), 20(2)) of the contributing depth positions (z).

18. The computer-implemented method according to any one of claims 13 to 15, wherein the Fourier transform algorithm comprises a 3D reconstruction step (117) for multidimensional phase reconstruction in a depth direction coordinate (z) of the sample structure data set (O(x, y, z)) and at least one lateral coordinate (x, y) of the sample structure data set (O(x, y, z)) based on the axial frequency space coordinate (Kz) and the at least one lateral frequency space coordinate (Kx, Ky).

19. The computer-implemented method of claim 18, wherein a region of direct reflection is excluded from the joint reconstruction.

20. Computer-implemented method according to one of claims 13 to 19, wherein the multi-dimensional XCT measurement data set (£>(Kx, Ky, Kz)) was generated according to a method according to one of claims 1 to 12, and / or wherein a plurality of scatter images (19) are supplied to the computer-implemented method as a two-dimensional Fourier measurement data set or as a three-dimensional Fourier measurement data set, the scatter images (19) being detected in a plurality of measurement processes with different structured frequency ranges.

21. Device for obtaining a multidimensional XCT measurement data set (£>(Kx, Ky, Kz)) of a sample (P) comprising: - an EUV / X-ray radiation source (3) for generating broadband EUV / X-ray radiation (6), - an optical device (11) for structuring the broadband spectrum of the EUV / X-ray radiation (6) as a structured frequency range and outputting an EUV / X-ray measuring beam (12) with a measuring beam spectrum comprising at least one of a plurality of narrowband structured frequency ranges for a measuring process, wherein at least one axial frequency space coordinate (Kz) in the frequency space (Kx, Ky, Kz) is assigned to the narrowband structured frequency ranges, - a sample holder (10) for supporting the sample (P) at an angle of incidence of the EUV / X-ray measuring beam (6) onto a surface portion of the sample (P); - a focusing element (15) configured to form a spatially lateral beam expansion parameter of the EUV / X-ray measuring beam (12) on the surface portion of the sample (P), wherein the beam expansion parameter is in a range from one micrometer to several hundred micrometers measured based on the FWHM intensity of the EUV / X-ray measuring beam (12), - a detector (17) arranged to record intensity values ​​in the far field to the sample (P), and - a control device (21) which - is connected to the optical device (11) for structuring the broadband spectrum for setting the structured frequency range for a plurality of measuring processes and to the detector (17) for reading out the intensity values ​​and - is designed to compile the multidimensional XCT measurement data set (£>(Kx, Ky, Kz)) from the intensity values.

22. Device according to claim 21, wherein the EUV / X-ray radiation source (3) is a synchrotron-based or laser-based radiation source, and / or wherein the sample holder (10) is formed - for supporting the sample (P) for adjusting the angle of incidence of the EUV / X-ray measuring beam (12) and / or - for storing the sample (P) in focus or in front of the focus of the focusing element (15) and / or - for scanning the sample by laterally moving the sample (P) with respect to the EUV / X-ray measuring beam (12).

23. Device according to claim 21 or 22, wherein the optical device (11) is designed to set different structured frequency ranges, in particular a plurality of narrow-band structured frequency ranges, in particular with a plurality of central wavelengths, for detecting the scattering images (19) in a plurality of measuring processes.

24. Device according to one of claims 21 to 23, wherein the device for obtaining a multidimensional XCT measurement data set is configured to structure the EUV / X-ray measurement beam (12) differently spectrally for N measurement processes, in particular to adjust it for a plurality of narrow-band structured frequency ranges in order to carry out N measurement processes with different spectral resolution.

25. Device according to one of claims 21 to 24, wherein the optical device (11) is designed as a mono- or multichromator, in particular a grating-based, grating-pair-based or multi-slit-based monochromator, and wherein in particular - a grating-based or grating-pair-based mono- or multichromator is rotatably mounted to adjust a central frequency of the structured frequency range and / or - a multi-slit-based mono- or multichromator for setting a central frequency of the structured frequency range with respect to a multi-slit configuration, such as slit width, and / or wherein the focusing element (15) is arranged upstream or downstream of the optical device (11) for structuring the broadband spectrum.

26. Device according to one of claims 21 to 25, wherein the focusing element (15) is designed to form a spatially lateral beam expansion parameter of the EUV / X-ray measuring beam (12) on a surface section of the sample (P), which defines a lateral dimension of the sample (P) and accordingly a lateral frequency component (Kx, Ky) and in particular lies in a range from 1 pm to 1 mm, in particular from 10 pm to 500 pm measured based on the FWHM intensity of the EUV / X-ray measuring beam (12).

27. Device according to one of claims 21 to 26, wherein the detector (17) - for recording one-dimensional scatter images (19) as a line detector for detecting a series of intensity values ​​or for recording two-dimensional scatter images (19) as a planar detector for detecting an array of intensity values, and / or - is rotatably mounted for aligning a detected lateral dimension, and / or wherein the intensity values ​​of a measuring process detected by the detector (17) each form a scatter pattern (19) or an intensity value detected by the detector (17) is assigned to one of a plurality of scatter patterns depending on an axial frequency space coordinate (Kz) assigned to the intensity value.

28. Device according to one of claims 21 to 27, wherein the control device (21) comprises a processor arranged to carry out a method according to one of claims 13 to 20.

Images