Method for adjusting a multi-channel detection unit of a multi-beam particle microscope during an inspection of rechargeable samples, associated computer program product and multi-beam particle microscope

Abstract

A method for calibrating a multichannel detection unit of a multibeam particle microscope during the inspection of rechargeable samples is disclosed. Time-consuming and space-consuming serial imaging with the multibeam particle microscope for calibrating a multichannel detection unit during the inspection of rechargeable samples is avoided and replaced by a cleverly chosen ensemble image or ensemble measurement. The multibeam capability of the multibeam particle microscope is specifically used for detection calibration purposes.

Description

Field of invention

[0001] The invention relates generally to multi-beam particle microscopes that operate with a plurality of single-particle beams. Specifically, the invention relates to a method for adjusting a multi-channel detection unit of a multi-beam particle microscope during the inspection of rechargeable samples, an associated computer program, and an associated multi-beam particle microscope. State of the art

[0002] With the continuous development of increasingly smaller and more complex microstructures, such as semiconductor devices, there is a need for the further development and optimization of planar fabrication techniques and inspection systems for the production and inspection of these small microstructures. The development and fabrication of semiconductor devices, for example, requires verification of test wafer designs, and planar fabrication techniques necessitate process optimization for reliable, high-throughput manufacturing. Furthermore, the analysis of semiconductor wafers for reverse engineering and the customized configuration of semiconductor devices is increasingly required. Therefore, there is a need for inspection tools that can be used with high throughput to examine microstructures on wafers with high accuracy.

[0003] Typical silicon wafers used in the production of semiconductor devices have diameters of up to 300 mm. Each wafer is divided into 30 to 60 repeating sections ("dies") with a size of up to 800 mm. 2A semiconductor device comprises multiple semiconductor structures fabricated in layers on a wafer surface using planar integration techniques. Due to the manufacturing processes, semiconductor wafers typically have a flat surface. The feature size of the integrated semiconductor structures ranges from a few micrometers to critical dimensions (CDs) of 5 nm, with even smaller feature sizes expected in the near future; future feature sizes or critical dimensions (CDs) below 3 nm, for example 2 nm, or even below 1 nm, are anticipated. At these small feature sizes, defects of critical dimension size must be identified quickly across a very large area.For several applications, the specification requirement for the accuracy of a measurement provided by an inspection instrument is even higher, for example by a factor of two or an order of magnitude. For example, the width of a semiconductor feature must be measured with sub-1 nm accuracy, such as 0.3 nm or even less, and the relative position of semiconductor structures must be determined with a sub-1 nm superposition accuracy, such as 0.3 nm or even less.

[0004] A more recent development in the field of charged particle microscopes (CPM) is the MSEM, a multi-beam scanning electron microscope. A multi-beam scanning electron microscope is disclosed, for example, in US 7,244,949 B2 and US 2019 / 0355544 A1. In a multi-beam electron microscope, or MSEM, a sample is simultaneously irradiated with a multitude of single-electron beams arranged in a field or grid. For example, 4 to 10,000 single-electron beams can be provided as primary radiation, with each single-electron beam separated from an adjacent single-electron beam by a distance of 1 to 200 micrometers. For example, an MSEM has approximately 100 separate single-electron beams (beamlets) arranged, for example, in a hexagonal grid, with the single-electron beams separated by a distance of approximately 10 µm.A multitude of charged single-particle beams (primary beams) are individually focused onto the surface of the sample under investigation by a common large-field optics system, including a common objective lens. The sample can be, for example, a semiconductor wafer mounted on a wafer holder, which is itself mounted on a movable stage. During illumination of the wafer surface with the charged primary single-particle beams, interaction products, such as secondary electrons or backscattered electrons, are emitted from the wafer surface. Their respective starting points correspond to the locations on the sample onto which the multitude of primary single-particle beams are focused. The quantity and energy of the interaction products depend, among other things, on the material composition and the topography of the wafer surface.The interaction products form several secondary single-particle beams (secondary beams) that are collected by the common objective lens and directed by a projection imaging system of the multi-beam inspection system onto a detector located in a detection plane. The detector comprises several detection areas, each of which can contain multiple detection pixels, and the detector records an intensity distribution for each of the secondary single-particle beams. This results in an image field of, for example, 100 µm × 100 µm.

[0005] The prior art multi-beam electron microscope comprises a series of electrostatic and magnetic elements. At least some of the electrostatic and magnetic elements are adjustable to adapt the focus position and stigmatization of the multiple charged single-particle beams. The prior art multi-beam charged particle system also includes at least one intersection plane of the primary or secondary charged single-particle beams. Furthermore, the prior art system includes detection systems to facilitate adjustment. The prior art multi-beam particle microscope includes at least one deflection scanner for collectively scanning an area of ​​the sample surface using the multiple primary single-particle beams to obtain an image field of the sample surface.

[0006] As mentioned above, each secondary beam strikes a separate or assigned detector element. Even with detector elements that are essentially identical in construction, these elements are not usually 100% identical. Instead, their detector characteristics, in particular, can differ, which can lead to different brightness and / or contrast values ​​for individual images, even if the exact same structure was scanned for each image. These differences in brightness and / or contrast can prove problematic, for example, when different images are to be stitched together. Furthermore, these differences are relevant when measuring the dimensions of surface structures that extend across multiple images.Contour recognition can also be made more difficult by differences in brightness and / or contrast values, because the edges of the individual images could be incorrectly interpreted as contours.

[0007] To solve the aforementioned problem, it is known from WO 2020 / 057678 A1 to calibrate a detection unit with multiple detector elements or detection channels based on images of a reference sample. The various detector elements or detection channels are calibrated with each other so that the individual images generated by scanning the reference sample in a raster pattern exhibit the same brightness within a predetermined limit and / or the same contrast value within another predetermined limit. The disclosure of WO 2020 / 057678 A1 is incorporated in its entirety by reference into the present patent application.

[0008] Although detector calibration is generally known, another problem exists: A setting for the detection units of a multichannel detection unit, generated once using a reference sample, cannot be universally applied to other samples. An optimal detection setting depends on several parameters. The physical nature of a sample, or the sample material, is one of these parameters. However, other factors also influence the optimal detection setting for image acquisition. These include, for example, the beam current, the landing energy, the pixel size, and the dwell time per pixel, although this list is not exhaustive. Therefore, a new detector calibration or adjustment is always performed at the beginning of each inspection process on the sample actually being inspected.

[0009] When scanning a non-rechargeable sample, the secondary electron yield is constant and, in principle, does not depend on the illumination history. Accordingly, with non-rechargeable samples, it is possible to scan the same inspection point multiple times without generating different images or measurement results with each subsequent measurement. Therefore, it is also possible to perform detector calibration or adjustments at the same inspection point on a non-rechargeable sample, even if this process requires a sequence of multiple image acquisitions. Thus, it is possible to manually adjust the detector settings while observing live images. Furthermore, detector adjustment on a non-rechargeable sample can also be automated.

[0010] With a rechargeable sample, however, the situation is different: When scanning rechargeable samples, the generated secondary electron yield depends fundamentally on the illumination history. Therefore, it is not possible to perform multiple scans / image acquisitions at the same inspection site for a single detector setting. This would distort the image result. Instead, a new inspection site, previously unsampled or unilluminated, is accessed for each scan. This process is time-consuming and also results in the consumption / destruction of many inspection sites, rendering them unavailable for the actual inspection task.

[0011] The known methods for setting up a detection unit or multi-channel detection unit during the inspection of rechargeable samples are therefore time-consuming and require a relatively large amount of space on the sample, which is no longer available for the actual inspection.

[0012] US 2020 / 0211811 A1 describes a multi-beam inspection device. It addresses the problem that non-uniformity between scanning points increases with the number of partial beams. US 2020 / 0211811 A1 proposes providing an array of multi-deflectors with different multipole electrodes, where the number of electrodes in a multipole electrode increases with its distance from the particle-optical axis.

[0013] US 6,563,114 B1 deals with the measurement of physical and / or chemical properties of a substrate or wafer as a function of a focus setting. The publication describes how a brightness histogram can be used to determine whether a focusing condition in the secondary path is met.

[0014] DE 10 2013 016 113 A1 discloses a special detector design, namely a combination of a particle and a light detector. To compensate for the decreasing efficiency / quality of particle detection over longer operating times with a scintillator plate, it is proposed to shift the scintillator plate laterally after a certain period. This allows areas of the scintillator plate that were previously not used for scintillation to be used for particle detection. Description of the invention

[0015] The object of the present invention is therefore to provide an improved method for setting up a multi-channel detection unit for the inspection of rechargeable samples. The method should be fast, precise, and space-saving. In particular, it should also be applicable to wafer inspection.

[0016] The problem is solved by the subject matter of the independent patent claims. Advantageous embodiments of the invention are described in the dependent patent claims.

[0017] A key principle of the invention is to avoid time-consuming and space-consuming serial imaging with the multi-beam particle microscope for calibrating a multi-channel detection unit during the inspection of rechargeable samples, and to replace this with a cleverly chosen ensemble image or ensemble measurement. Thus, instead of having to expend increasing effort to calibrate a corresponding number of detection channels for an increasing number of individual particle beams, the multi-beam capability of the multi-beam particle microscope is specifically exploited for detection calibration purposes.

[0018] According to a first aspect of the invention, this relates to a method for adjusting a multi-channel detection unit of a multi-beam particle microscope during an inspection of chargeable samples, wherein the multi-beam particle microscope operates with a plurality of charged single-particle beams, wherein each detection channel of the multi-channel detection unit is assigned one of the charged single-particle beams for image generation, and wherein the method comprises the following steps: Providing the multi-channel detection unit with a multitude of individual reference settings Ri, wherein each of the reference settings Ri describes the reference setting Ri of an associated detection channel, wherein with this multitude of reference settings Ri a multitude of individual images can be generated when scanning a reference sample, which satisfy a predefined detector calibration criterion; Detuning the reference settings Ri of the multi-channel detection unit, whereby the multitude of individual reference settings Ri per detection channel is each changed by an individual detuning Δi, i.e. Ri_detune = Ri + Δi; Approaching a first inspection point of a chargeable sample; scanning the first inspection point with the multitude of charged single-particle beams, whereby a multitude of particle-optical single images are generated with the detuned settings Ri_detune of the multi-channel detection unit; Evaluating the large number of individual images with regard to a predefined detection quality criterion; Determining the best single image from the multitude of single images based on the detection quality criterion and determining the detuning Δi_best that belongs to the best single image; Determining a channel-independent global detuning Δ_global based on the detuning Δi_best; and Applying the global detuning Δ_global to the reference settings Ri of all detection channels i, thereby setting optimal settings Ri_opt of the multi-channel detection unit.

[0019] The charged single-particle beams can consist of electrons, positrons, muons, ions, or other charged particles. The individual field regions of the object assigned to each single-particle beam are scanned rasterically, for example, row by row or column by column. The single-particle beams can be arranged in a raster configuration, such as a rectangular or hexagonal grid. In the hexagonal case, it is advantageous if the number of charged single-particle beams is 3n (n-1)+1, where n is any natural number.

[0020] The interaction products can be backscattered electrons or secondary electrons. For analytical purposes, it is preferable to use the lower-energy secondary electrons for image generation.

[0021] The multi-channel detection unit or multi-detector according to the invention can comprise one or more detectors of the same or different types. The detectors themselves can be pixelated, but need not be. The multi-channel detection unit can, for example, have or consist of one or more particle detectors. A particle detector can, in turn, be designed as a single unit or in multiple parts. It is also possible to combine one or more particle detectors and light detectors with each other or to connect them in series within the detection unit. In the context of the present invention, the term "detection channel" is always understood to mean that each detection channel is assigned exactly one charged single-particle beam. The number of detection channels thus corresponds to the number of charged single-particle beams used for imaging.

[0022] According to the invention, the multichannel detection unit is first provided with a plurality of individual reference settings Ri, each of which describes the reference setting Ri of an associated detection channel. With this plurality of reference settings Ri, a plurality of individual images can be generated during a raster scanning of a reference sample, which fulfill a predefined detector calibration criterion. The reference sample is not identical to the chargeable sample to be inspected. Providing the multichannel detection unit with a plurality of individual reference settings Ri does not necessarily mean that the corresponding detector calibration or the determination of the reference settings Ri is actually performed during the inventive process, although this may be the case.It is preferable, however, that the corresponding reference settings Ri of the multichannel detection unit for the respective multibeam particle microscope used for the method according to the invention are determined prior to the execution of the method. The individual reference settings Ri can, for example, be stored in a memory of the multibeam particle microscope and read from the memory into a controller of the multibeam particle microscope for the method according to the invention. It is also possible that several sets of reference settings Ri of the multibeam particle microscope are stored in a memory, for example, depending on a sample material and / or depending on different operating points of the multibeam particle microscope. The reference sample can be a sample similar to the rechargeable sample to be inspected, but this need not be the case.The reference sample can also be a non-rechargeable sample. Nevertheless, it is advantageous to start the inventive method for adjusting a multi-channel detection unit of a multi-beam particle microscope during the inspection of chargeable samples with the best possible individual reference setting Ri.

[0023] In a further process step, the reference settings Ri of the multi-channel detection unit are detuned. Each individual reference setting Ri for a given detection channel is modified by an individual detuning value Δi, i.e., Ri_detune = Ri + Δi. Thus, the setting of each detection channel is specifically changed in this process step, allowing for a targeted ensemble image acquisition with different detection settings during subsequent image acquisition. Each detection channel is individually detuned by Δi. Preferably, the individual detuning values ​​Δi of the detection channels are pairwise distinct. This allows for subsequent image acquisition with the largest possible ensemble of different detection channel settings. The individual detuning values ​​Δi are relatively small.The change in the setting of the detection channels thus tends to be a fine-tuning adjustment rather than a complete readjustment. Preferably, the deviation Δi should be a maximum of + / -10%, preferably a maximum of + / -8% or a maximum of + / -5% of the value of the reference setting Ri.

[0024] According to the invention, a first inspection point of a rechargeable sample, for example a wafer, is then approached. However, the sample can also be another type of rechargeable sample.

[0025] The first inspection point is then scanned using a raster pattern with the multitude of charged single-particle beams, whereby a multitude of particle-optical images are generated using the detuned Ri_detune settings of the multi-channel detection unit. This process step corresponds to the ensemble measurement already mentioned.

[0026] The numerous individual images are then evaluated according to a predefined detection quality criterion. This detection quality criterion can be defined in various ways. It can consist of several sub-criteria, or a single criterion may suffice. Specific examples of detection quality criteria are described in more detail below.

[0027] Next, the best individual image is determined from the multitude of individual images based on the detection quality criterion, and the detuning Δi_best associated with the best individual image is determined. Thus, the best individual image is selected or determined from the ensemble of individual images. Since it is now known in principle which detuning Δi_best led to the acquisition of the best individual image for a specific detection channel, this knowledge is used to adjust the other detection channels accordingly. According to the invention, a channel-independent global detuning Δ_global is determined based on the detuning Δi_best. The term "global" indicates that the determined detuning Δ_global is identical for all detection channels.

[0028] The global detuning Δ_global is then applied to the reference settings Ri of all detection channels i, thereby establishing optimal settings Ri_opt for the multichannel detection unit. These optimal settings Ri_opt are specific to the rechargeable sample being inspected. The settings Ri_opt are optimal in the sense that the calibration or fine-tuning of the multichannel detection unit has now been performed so effectively that, in a further optional process step, the actual sample inspection can proceed directly using the determined optimal settings Ri_opt.

[0029] When applying the global detuning Δ_global to the reference settings Ri of all detection channels i, a transformation of the detuning Δi_best from channel i_best to the other channels can be performed. These transformations can be based on the characteristic curve or a family of characteristic curves that describes a detection unit of a respective detection channel.

[0030] According to a preferred embodiment of the invention, a second inspection site of the chargeable sample is approached, wherein the second inspection site differs from the first inspection site. The second inspection site is then scanned using a multiple of charged single-particle beams, generating a multiple of particle-optical images with the optimal Ri_opt settings of the multi-channel detection unit. Similarly, a third, fourth, fifth, and many further inspection sites of the chargeable sample can first be approached and then scanned with the multiple of charged single-particle beams, again using the optimal Ri_opt settings of the multi-channel detection unit for image acquisition.According to the invention, only a single inspection point needs to be used for setting up the multi-channel detection unit. All other inspection points are not used for setting up the multi-channel detection unit and are instead available for the actual inspection process.

[0031] According to a preferred embodiment of the invention, the global detuning Δ_global is identical to the detuning Δi_best that corresponds to the best individual frame. In this embodiment, the accuracy with which a best detuning Δi_best can be determined is thus predetermined by the proximity of the various detunings Δi to one another. Many strongly differing detunings Δi lead to a less accurate result, while a large number of only slightly differing detunings Δi allow for a more precise determination of the global detuning Δ_global. The latter embodiment is preferred. For this, however, it may be important to start the inventive method with relatively well-known individual reference settings Ri.

[0032] According to an alternative embodiment of the invention, the global detuning Δ_global is determined by interpolation. This involves determining not only the best frame and its corresponding detuning Δi_best, but also the second-best frame and its corresponding detuning Δi_best2. Normally, these two detunings differ only slightly from each other or are adjacent to each other in the configuration space, so that an even more precise value for the global detuning Δ_global can be determined by means of interpolation.

[0033] Each detection channel is assigned a reference setting Ri at the beginning of the method according to the invention. This reference setting Ri can be defined by one parameter or by several parameters. When different detector types are connected sequentially within a detection channel, for example, a combination of a particle detector and a light detector in each detection channel, the number of parameters that define the reference setting Ri for channel i can be particularly large. In general, a reference setting Ri can be defined by a plurality of parameters j = 1..L, with L∈ℕ.

[0034] According to a preferred embodiment of the invention, each reference setting Ri of a detection channel of the multi-channel detection unit is defined by exactly one parameter. Then j = 1. The individual detuning Δi for each detection channel i then comprises a detuning of this exactly one parameter.

[0035] According to an alternative embodiment of the invention, each reference setting Ri of a detection channel of the multi-channel detection unit is defined by at least two parameters, i.e., j ≥ 2. The individual detuning Δi for each detection channel i comprises a detuning of at least two parameters. Generally, the definition of the reference setting Ri and the individual detuning Δi of this reference setting Ri take place in the same parameter space or configuration space.

[0036] According to a preferred embodiment of the invention, each reference setting Ri of a detection channel i and / or each individual detuning Δi is determined by values ​​and / or changes in values ​​of at least one of the following parameters: gain, offset. The two parameters gain and offset are the two most important setting parameters in practice when setting up a detection unit or a detection channel with such a detection unit.

[0037] The gain essentially indicates how much output is generated from a given amount of input. Specifically, the gain represents the ratio of the number of particles of the second particle type (out) in the output to the number of particles of the first particle type (in) in the input. In the case of an avalanche photodiode, the input consists of photons and the output of electrons. Analogous definitions apply to all other detectors, such as direct electron detectors (DEDs), photomultiplier tubes (PMTs), and so on. Furthermore, many detectors that convert the signal into an electrical signal (current or voltage drop across an output resistor) incorporate a post-amplifier whose gain is also adjustable.

[0038] The offset, in turn, indicates the level of the output signal when no primary particles are present. This is usually solved in the electronic downstream circuitry by a voltage adder. Ultimately, these systems compensate for leakage currents, etc. Offset and gain are generally not independent of each other.

[0039] According to a preferred embodiment of the invention, the individual detunings Δi of the detection channels i are each pairwise distinct. This allows for the largest possible ensemble measurement for adjusting the multi-channel detection unit. According to a preferred embodiment of the invention, the individual detunings Δi with respect to at least one parameter are generated by incremental changes of Δi by a constant value k. For example, it is thus possible to detune the offset by different values ​​for each detection channel, with adjacent values ​​differing by the constant value k. In principle, the gain can also be changed incrementally for the different detection channels in an analogous manner.Furthermore, it is possible to first fix the value of one parameter and selectively change another parameter in different detection channels, then selectively change the first parameter once, and then again traverse the value range of the second parameter in the other detection channels, and so on. In this way, a matrix of individual tunings Δi (ik) can be generated. A solution space for the best optimal settings Ri_opt can be very well covered in this way. However, there are also many cases in which the individual tunings Δi only need to be performed with respect to one parameter in order to optimally cover the solution space.

[0040] According to a preferred embodiment of the invention, evaluating the plurality of individual images includes generating brightness histograms of the individual images. It is not necessary to output or explicitly display these brightness histograms as graphics. The focus is solely on computationally acquiring the brightness histograms so that they can be evaluated in a suitable form based on a detection quality criterion. An individual adjustment Δi in the form of a change in offset typically only shifts the curve or values ​​within the histogram. An individual adjustment Δi in the form of a change in contrast typically has more complex consequences, altering the width of the distribution as well as the brightness value or mean brightness value of the distribution.

[0041] According to a preferred embodiment of the invention, the detection quality criterion is determined based on at least one of the following criteria: - Frequencies of brightness levels in a first interval of the brightness histogram, which represents very low brightness levels. This interval thus describes a boundary region of the brightness histogram, whereby it is possible, but not mandatory, that the brightness or the brightness value of zero is also included in this first interval or boundary interval. - Frequencies of brightness levels in a second interval of the brightness histogram, which represents very high brightness levels. This second interval thus denotes a second boundary region with very high or even extremely high brightness levels, whereby it is possible, but not necessary, for the highest or maximum brightness to lie within the second interval. - Total width of the frequency distribution with non-zero frequencies. In this way, the entire brightness range or range of grayscale values ​​(for example, 256 bits) can be made available for image analysis. - Frequency of the maximum detectable brightness. If this frequency is very high, this indicates that the detection channel is not yet optimally adjusted, but rather that the information is not optimally resolved at high brightness levels. - Frequency of the minimum detectable brightness. Similarly, a detection channel setting is not optimal if a high frequency of the lowest brightness is detected.

[0042] This list of criteria is not exhaustive, but should only be understood as an example.

[0043] According to a preferred embodiment of the invention, a single image is rated higher according to the detection quality criterion the less frequently brightness values ​​are counted in the first interval. Additionally or alternatively, a single image is rated higher according to the detection quality criterion the less frequently brightness values ​​are counted in the second interval. The underlying principle of these two ratings is to prevent too much information from being contained or unresolved in the peripheral regions of a brightness histogram and thus being lost. Conversely, it is preferred that the overall width of the frequency distribution be large enough to allow for good information resolution. This is not the case if a frequency distribution has only a single sharp peak. In that case, many brightness values ​​or grayscale values ​​remain unused, and the particle-optical image contains less detailed information.

[0044] According to a preferred embodiment of the invention, a detuned setting Ri_detune of a detection channel fulfills a minimum quality requirement if a maximum of 0.20%, preferably a maximum of 0.15%, or most preferably 0.10% of all brightness values ​​lie within the first interval. Additionally or alternatively, a detuned setting Ri_detune of a detection channel fulfills a minimum quality requirement if a maximum of 0.20%, preferably a maximum of 0.15%, or most preferably 0.10% of all brightness values ​​lie within the second interval. Thus, only a small amount of information or frequency is present in the edge regions of the brightness histogram; however, this information is preferably present to ensure that all gray values ​​or bits of the brightness histogram are used effectively. The size of the edge regions is therefore not a determining factor.For example, intervals I1 and I2 in a brightness histogram with grayscale values ​​from 0 to 255 can be defined as follows: I1 = [10, 70] and I2 = [170, 245]. This corresponds to a definition according to ISO TS 24597. However, other interval sizes and limits are also possible.

[0045] According to a preferred embodiment of the invention, a detuned setting Ri_detune of a detection channel meets a minimum quality requirement if a maximum of 0.20%, preferably a maximum of 0.15%, or most preferably a maximum of 0.10% of all brightness values ​​lie within the first and second intervals. Here, the brightness values ​​of the first and second intervals are considered together to check whether the minimum quality requirement is met. Of course, it is possible to combine this common criterion for the first and second intervals with the separate criteria for each interval, as described below.

[0046] According to a preferred embodiment of the invention, the multi-channel detection unit comprises a combination of a particle detector and a light detection unit. The particle detector can, for example, be a scintillator plate. When second or secondary single-particle beams strike the scintillator plate, a light signal is emitted, which can then be evaluated separately by assigning it to the point of impact on the particle detector or the scintillator plate.

[0047] According to a preferred embodiment of the invention, the light detection unit comprises a plurality of avalanche photodiodes. Each detection channel is assigned one avalanche photodiode. When the reference settings Ri of the multichannel detection unit are detuned, the gain and / or offset of an avalanche photodiode is changed. Avalanche photodiodes are highly sensitive and very fast photodiodes and are suitable for detecting even low radiation powers. They are therefore ideally suited for use in multi-channel particle microscopes. Adjusting the gain and / or offset of an avalanche photodiode can, for example, be achieved by changing the current or voltage.

[0048] According to a second aspect of the invention, it relates to a computer program product comprising program code for executing the method as described above in several embodiments. In principle, the program code can be written in any programming language. In particular, the program code can be modular. For example, it is advantageous to provide code for controlling the multi-beam particle microscope in a separate module, while another module contains the routines for the actual setup of the multi-channel detection unit.

[0049] According to a third aspect of the invention, it relates to a multi-beam particle microscope configured to perform the method according to one of the embodiments of the first aspect of the invention described above.

[0050] The embodiments and aspects of the invention described above can be combined in whole or in part, provided that no technical contradictions result from this combination.

[0051] The invention will be better understood with reference to the accompanying figures. These show: Fig. 1: shows a multi-beam particle microscope in schematic representation (MSEM); Fig. 2: shows a schematic representation of the structure of a multi-channel detection unit; Fig. 3: schematically shows a multi-image with a large number of individual images; Fig. 4: shows various brightness histograms as examples and illustrates their relationship with a detector characteristic curve; Fig. 5: schematically shows the effects of a changed detector offset on a brightness histogram; Fig. 6: schematically shows the effects of a changed gain on a brightness histogram; Fig. 7: schematically illustrates criteria for a detection quality criterion; Fig. Figure 8: schematically illustrates an adjustment setting of a detection channel to meet a detection quality criterion for each individual image or for each individual particle beam; Fig. 9: schematically illustrates the method according to the invention; Fig. Figure 10: schematically shows brightness histograms of a large number of individual images, which were recorded simultaneously with the large number of single-particle beams with individually detuned detector settings, in a combined graphic; Fig. Figure 11: schematically shows the determination of the best individual image from the multitude of individual images based on a detection quality criterion; and Fig. Figure 12: schematically shows brightness histograms after adjusting the multi-channel detection unit according to the invention.

[0052] Fig. Figure 1 schematically shows a multi-beam particle beam system 1 in the form of a multi-beam particle microscope 1. The multi-beam particle microscope 1 has a beam generation device 300 with a particle source, for example, an electron source. Charged particles or electrons are generated by the beam generation device 300, for example, by thermal field emission. The emitted charged particles form a diverging particle beam 309, which is collimated by a sequence of condenser lenses 303.1 and 303.2 and strikes a multi-beam particle generator 305 with a multi-aperture arrangement. The multi-beam particle generator 305 comprises several multi-aperture plates 304, 306 and a field lens 307. A multitude of single-particle beams 3 or 3 are emitted by the multi-beam particle generator 305.Single-electron beams 3 are generated, arranged in a field which is mapped onto another field formed by beam spots 5 in the object plane 101. The distance between the centers of apertures of a multi-aperture plate 306 can be, for example, 5 µm, 100 µm, and 200 µm. The diameters D of the apertures are smaller than the distance between the centers of the apertures; examples of the diameters are 0.2 times, 0.4 times, and 0.8 times the distances between the centers of the apertures.

[0053] The multi-aperture arrangement 305 and the field lens 308 are configured to generate a multitude of focal points 323 of primary beams 3 in a grid arrangement on a surface 321. The surface 321 need not be a flat surface, but can be a spherically curved surface to accommodate field curvature of the subsequent particle optical system.

[0054] The multi-beam particle microscope 1 further comprises a system of electromagnetic lenses 103 and an objective lens 102, which reduce the size of the beam foci 323 from the intermediate image plane 321 onto the object plane 101. The first individual particle beams 3 pass through the beam splitter 400 and a collective beam deflection system 500, which deflects the multitude of the first individual particle beams 3 during operation and scans the image field. The first individual particle beams 3 incident on the object plane 101 form, for example, a substantially regular field, with distances between adjacent point locations 5 being, for example, 1 µm, 10 µm, or 40 µm. The field formed by the point locations 5 can, for example, have a rectangular or hexagonal symmetry.

[0055] The object 7 to be examined can be of any type, for example a semiconductor wafer or a biological sample, and it can comprise an array of miniaturized elements or the like. The surface 15 of the object 7 is located in the object plane 101 of the objective lens 102. The objective lens 102 can comprise one or more electron-optical lenses. It can be, for example, a magnetic objective lens and / or an electrostatic objective lens.

[0056] The primary particles 3 striking object 7 generate interaction products such as secondary electrons, backscattered electrons, or primary particles that have undergone a reversal of motion for other reasons. These products originate from the surface of object 7 or from the first plane 101 or object plane 101. The interaction products emanating from the surface 15 of object 7 are shaped into secondary particle beams 9 by the objective lens 102. After passing through the objective lens 102, the secondary beams 9 pass through the beam splitter 400 and are directed to a projection system 200. The projection system 200 has an imaging system 205 with projection lenses 206, 208 and 210, a contrast aperture 214 and a multi-particle detector 207. The impact points 25 of the second single-particle beams 9 on the detection areas of the multi-particle detector 207 are located in a third field at a regular distance from each other.Examples of values ​​are 10 µm, 100 µm and 200 µm.

[0057] The multi-beam particle microscope 1 further comprises a computer system or a control unit or controller 10, which in turn may be designed as a single unit or as a multi-part unit, and which is designed both for controlling the individual particle-optical components of the multi-beam particle microscope 1 and for evaluating and analyzing the signals obtained with the multi-detector 207 or the detection unit.

[0058] The computer system 10 can be composed of several individual computers or corresponding components. According to a preferred embodiment, the computer system 10 comprises a control computer system 270 and one or more image generation computers 280. The provision of several image generation computers 280 is advantageous due to the large amounts of data generated and allows for the parallel evaluation of detector signals. The method according to the invention can also be carried out using the described computer system 10; that is, the computer system 10 is configured by a program for carrying out, among other things, the method according to the invention.

[0059] Further information on such multi-beam particle beam systems or multi-beam particle microscopes 1 and components used therein, such as particle sources, multi-aperture plates and lenses, can be obtained from the international patent applications WO 2005 / 024881 A2, WO 2007 / 028595 A2, WO 2007 / 028596 A1, WO 2011 / 124352 A1 and WO 2007 / 060017 A2 and the German patent applications DE 10 2013 016 113 A1 and DE 10 2013 014 976 A1, the disclosures of which are fully incorporated into the present application by reference.

[0060] Fig. Figure 2 is a schematic representation to illustrate several details of detector 209. Detector 209, as a particle detector, comprises the scintillator plate 207, onto which the interaction products, for example secondary electron beams, are directed by an electron optic. This electron optic, when it enters the multi-beam particle microscope 1 of the Fig. The electron-optical components of the particle optics, which shape the electron beams 9, are integrated into the system. These include, for example, the objective lens 102, which directs the electron beams 9 towards the detector 209, such as the beam splitter 400, and the lens 2010, which focuses the electron beams 9 on the surface of the scintillator plate 207. The electron beams 9 strike the scintillator plate 207 at impact locations 25. Even when the electron beams 9 are focused on the surface of the scintillator plate 207, beam spots are formed on the surface, the diameter of which is not arbitrarily small. The centers of the beam spots can be considered the impact locations 25, which are arranged at a distance P2 from each other.

[0061] The scintillator plate 207 contains a scintillator material which is excited by the incident electrons of the electron beams 9 to emit photons. Each of the impact sites 25 thus forms a source of photons. In Fig. Figure 2 shows only a single corresponding beam path 221, which originates from the point of impact 25 of the middle of the five depicted electron beams 9. The beam path 221 passes through a light optic 223, which in the example shown comprises a first lens 225, a mirror 227, a second lens 229, and a third lens 231, and then strikes a light-receiving surface 235 of a light detection system 237. The light-receiving surface 235 is formed by an end face of an optical fiber 239, into which at least some of the photons are coupled and guided to a light detector 241. The light detector 241 can, for example, comprise a photomultiplier, an avalanche photodiode, a photodiode, or other types of suitable light detectors. The light optics 223 is configured to optically image the surface 208 of the scintillator plate 207 into an area 243 in which the light receiving surface 235 is located.Due to this optical imaging, optical images of the impact locations 25 are generated in the area 243. For each of the impact locations 25, a separate light-receiving surface 235 of the light detection system 237 is provided in the area 243. Each of the further light-receiving surfaces 235 is formed by an end face of an optical fiber 239, which directs the light coupled into the end face to a light detector 241. Due to the optical imaging, a light-receiving surface 235 is assigned to each of the impact locations 213, whereby the light entering a respective light-receiving surface 235 is detected by a separate light detector 241. The light detectors 241 output electrical signals via signal lines 245. These electrical signals represent intensities of the particle beams 9.Thus, the locations on the surface of the scintillator plate 207, which are imaged onto the light-receiving surfaces of light detectors 241, define different detection points or detection areas. Due to the electron optics described above, interaction products, for example, electrons emitted from two different single-field regions of an object, are also projected onto different detection areas of the scintillator plate 207. In the embodiment described here, the light detectors 241 are arranged remotely from the light-receiving surfaces 235, onto which the light optics 223 image the scintillator plate 207, and the received light is directed to the light detectors 241 via optical fibers 239. However, it is also possible for the light detectors 241 to be arranged directly where the light optics generate the image of the scintillator plate, and the light-sensitive surfaces of the light detectors thus form the light-receiving surfaces.

[0062] Fig. Figure 2, as mentioned, only schematically illustrates some details of the detector 209. It should be noted here that the scanning motion of the primary particle beams 3 across an object 7 or sample 7 irradiates or scans many points of the sample 7. Each primary particle beam 3 sweeps out a single field region of the object 7, either completely or partially. Each primary particle beam 3 is assigned its own single field region of the object 7. Interaction products, such as secondary electrons, then emerge from these single field regions of the object 7. These interaction products are then projected onto the detection areas of the particle detector or onto the scintillator plate 207 in such a way that the interaction products emerging from two different single field regions are projected onto different detection areas of the scintillator plate 207.When interaction products, e.g., secondary electrons, strike each detection area of ​​the scintillator plate 207, light signals are emitted. These light signals are then directed to a light detector 241 assigned to that area. In other words, each primary particle beam 3 has its own detection area on the scintillator 207 and its own light detector 241. Therefore, in the multi-beam particle microscope 1 described, all characteristic curve deviations of the detection areas or detectors are visible in the individual images produced.This means that the brightness and / or contrast in each individual image can vary due to the described deviations, and that the composite full image based on all primary particle beams will not appear satisfactory until the different detection areas or detectors are aligned with each other with respect to brightness and / or contrast, i.e., until a predefined detector alignment criterion is met. Such a criterion could be, for example, that all individual images of a multi-image have the same brightness value within a defined first limit and / or the same contrast value within a defined second limit.

[0063] In carrying out the method according to the invention, the light detectors 241, e.g. avalanche photodiodes, can be adjusted. Other types besides those described in [reference missing] can also be used. Fig. The detection architectures shown in section 2 are suitable for carrying out the methods according to the invention. Reference is made, for example, to the DED (direct electron detection) method described above, which does not require light detectors and in which secondary electrons are directly converted into a current signal.

[0064] Fig. Figure 3 schematically shows a multi-image with a large number of individual images, created using the method described in the Fig. 1 and Fig. The multi-beam particle microscope shown in Figure 2 can be generated. In the example shown, the multi-image is composed of 91 individual images. The individual images are arranged in a hexagonal overall configuration. Each individual image is generated by scanning a sample surface using an assigned single-particle beam 3 and is thus assigned to a detection channel of the multi-detector 209 or a multi-channel detection unit 209. Fig. The individual images are numbered. Each image is built up line by line by a corresponding raster or scan movement of the primary particle beam. The number of pixels in the individual image can be very high, ranging from approximately 1000 x 1200 to approximately 8000 x 9300 pixels. The in Fig. Three schematically represented individual images exhibit different brightness and contrast values, which can be corrected by means of a detector calibration known per se. Thus, for example, all individual images within a defined first limit can have the same brightness value and / or within a defined second limit the same contrast value. Further information on this can be found in WO 2020 / 057678 A1, the disclosure of which is incorporated in full by reference into the present patent application.

[0065] Fig. Figure 4 schematically illustrates the relationship between a brightness histogram and a characteristic curve of a detection area or detection channel. For each individual image generated with a specific detector setting, a single-image analysis can be performed and a brightness histogram determined. In this brightness histogram, the brightness, i.e., the gray level of the output signal, is plotted on the x-axis, and the number of pixels in the individual image that possess the respective brightness is indicated on the y-axis. For a first individual image, a first curve K1 is obtained, which is shown in Fig. 4a is represented by the solid line. It includes a clearly defined peak. For a second single image, a curve K2 results, which is shown in Fig. Figure 4a is represented by the dashed curve K2. It comprises two peaks, which partially merge into one another. The two curves K1 and K2 are merely examples of brightness histograms. In any case, it is possible to evaluate curves K1 and K2 and thus determine the brightness and contrast of a single image. There are various ways to define brightness and contrast. A brightness value can be defined, for example, as the measured mean of the distribution or as the mean of a Gaussian distribution fitted to a curve. The latter would be possible for histograms that more closely resemble curve K1. A contrast C1 can then be defined, for example, as a multiple of the standard deviation of the Gaussian distribution, such as two sigma of the Gaussian distribution. This is represented by the double arrow in curve K1.Even for histograms that are more similar to the type of curve K2, a contrast can be defined, for example as the distance between the two peaks, which in turn is in . Fig. 4a is represented as double arrow C2. There are many possible meaningful definitions. However, a definition actually used is then, of course, equally binding for all individual images.

[0066] Fig. Figure 4b illustrates an example detector characteristic curve. The graph plots the output signal of a detector against the beam intensity. For illustrative purposes, the axis scales are normalized to one. It is generally evident that the output signal increases as the beam intensity increases. There is at least a section-by-section linear relationship between the beam intensity and the output signal. Section-by-section, a straight line can be fitted to the shown characteristic curve, i.e., a tangent can be drawn, with the slope β of the tangent describing the gain. The characteristic curve intersects the y-axis at a height d. Therefore, even without an incoming beam or without photons incident on the detector (depending on the detector), a weak output signal is obtained. Values ​​for the output signal must be adjusted for this so-called offset d.The gain β and the offset d change when the overall characteristic curve changes. Therefore, both the gain β and the offset d can be adjusted by changing the characteristic curve.

[0067] Different detector settings result in different brightness histograms. Fig. Figure 4a shows that a pure detector offset only affects the brightness value. In other words, curves K1 and K2 in the diagram shift left or right when the offset of a detector in a detection channel, for example, the offset of an avalanche photodiode, is changed. However, if the gain of the detector characteristic is changed, two values ​​change in combination: both the brightness value and the width of the distribution, specifically, for example, the standard deviation (two sigma width) of a Gaussian distribution (see curve K1).

[0068] Fig. Figure 5 schematically shows the effects of a changed detector offset, such as an offset of an avalanche photodiode, on a brightness histogram. Fig. Figure 5a shows a first brightness histogram at a first offset D1. Curve K1 is located relatively far to the right in the brightness histogram. In the example shown, the brightness value of the brightness histogram is determined by position P. av defined, which lies between the positions of the two peaks P1 and P2. In the example shown from Fig. 5a is P av approximately 217.

[0069] In Fig. In 5b, only the offset of the detection unit, for example an avalanche photodiode, was adjusted to the value D2. This does not change the shape of curve K1 in the brightness histogram itself, but it does change its position: In the example shown from Fig. In 5b, the curve K1 is shifted entirely to the left. This also changes the brightness value P. av to about 130.

[0070] In Fig. 5c is set to the offset on a third value D3. This in turn has shifted the position of curve K1 even further to the left: The brightness value P av It is now around 91. A change in the offset d therefore shifts the average brightness value P. av in the brightness histogram.

[0071] Fig. Figure 6, in contrast, schematically shows the effects of a changed gain β on a brightness histogram: In Fig. 6a is again the initial situation according to Fig. Figure 5a shows the gain set at the detection unit or detection channel. The gain is β1. Fig. Figure 6b schematically illustrates the effect a change in gain can have on the value β2. The other parameters of the detection unit, especially the offset, have been left unchanged. This can be seen in Fig. 6b, that the change in gain has changed the contrast value of the brightness histogram: The contrast C2 is lower than the contrast C1 according to Fig. 6a. In addition to the change in contrast, the brightness value of the brightness histogram has also changed: The value P av has decreased. Fig. Figure 6c shows an example of a gain β3, where the other setting parameters of the detection unit have again been left unchanged. In particular, the offset has not been changed. According to Fig. In 6c, the curve K1c is significantly wider than in the examples according to Fig. 6a and Fig. 6b. The contrast has increased again and is now C3. The average brightness value P av has changed again and is now located approximately in the middle of the brightness histogram.

[0072] It should be noted that Fig. Figure 6 is only a schematic representation; the effect shown for changing the gain β is also exaggerated to illustrate the effects of the gain change on a detection unit.

[0073] By changing an offset and / or gain, it is possible to selectively modify a brightness histogram. A brightness histogram can therefore be adjusted to meet specific requirements. Fig. Figure 7 schematically shows criteria for a detection quality criterion: The following is shown in Fig. 7a is a first interval I1 of the brightness histogram, representing very low brightness levels. In the example shown, the value 0 is the lower limit of interval I1. Furthermore, in Fig. Figure 7a shows a second interval, I2, of the brightness histogram, which represents very high brightness levels. In the example shown, the maximum brightness value of 255 is part of interval I2. In this example, values ​​of curve K1 lie in both intervals I1 and I2; however, in intervals I1 and I2, these values ​​drop sharply towards the outer edges, approaching zero. Due to this steep drop in intervals I1 and I2 towards the edges of the brightness histogram, it can be assumed that all the necessary information is represented within the histogram; no significant information about a single frame is lost at the edges. Therefore, it is possible to define a minimum quality requirement such that a maximum of 0.20%, and in particular a maximum of 0.15% or 0.10%, of all brightness values ​​lie within the first interval, I1.Additionally or alternatively, it can be required that a maximum of 0.20%, preferably a maximum of 0.15%, or most preferably a maximum of 0.10% of all brightness values ​​may lie within the second interval I2. In addition to or alternatively to the two aforementioned criteria for intervals I1 and I2 separately, a combined minimum quality requirement for a detection channel can be considered fulfilled if a maximum of 0.20%, preferably a maximum of 0.15%, or most preferably a maximum of 0.10% of all brightness values ​​lie within the first and second intervals I1 and I2 combined. This requirement is even stricter.

[0074] The position of intervals I1 and I2 can also be defined somewhat differently, as exemplified in Fig. 7b is shown: In the case of the Fig. In the example shown in Figure 7b, the gray value 0, representing the lowest brightness, does not belong to interval I1, and the value 255, representing the maximum brightness, does not belong to interval I2. With such a definition of intervals I1 and I2, it is possible to define additional, separate minimum quality requirements. For example, the frequency of the maximum detectable brightness (here: the value at 255) as well as the value of the minimum detectable brightness (here: the value at 0) can serve as criteria. It can be required, for instance, that the value 0 is not recorded or that it may only occur in 0.001% of all brightness values. The same applies to the frequency of the highest detectable brightness, which has the value 255. This ensures that no important information is lost.

[0075] Another criterion for defining a detection quality criterion is the total width of the frequency distribution with frequencies ≠ 0. For example, it can be required that the curve K1 only drops to zero within the boundary regions or intervals I1 and I2. A contrast value C is determined according to the criteria in Fig. The 6 definitions explained as examples are therefore relatively large.

[0076] Fig. Figure 8 schematically illustrates an adjustment setting of a detection channel to fulfill a detection quality criterion for each individual image or for each individual particle beam. In the example shown, only the offset d is changed for setting the detection quality criterion in the detection channel, for example, an avalanche photodiode: While according to Fig. 8a if the peak P2 of curve K1 still lies within the second interval I2 near the edge of the brightness histogram, this is the case with the changed setting according to Fig. 8b is no longer the case. At offset d2, significantly fewer pixels are in interval I2. However, there are in interval I1. Fig. 8b shows no pixels at all. Therefore, curve K1 is still too narrow for optimal settings. This can be changed by adjusting the gain β of the detection unit; a corresponding result with a gain β2 is shown in Fig. Figure 8c shows that pixels of curve K1 lie in both interval I1 and interval I2; however, the total number of pixels in intervals I1 and I2 is limited or lies below a predefined limit.

[0077] Fig. Figure 9 schematically illustrates a method according to the invention for setting up a multi-channel detection unit 209 of a multi-beam particle microscope 1 during an inspection of chargeable samples 7. In an initial process step S0, the multi-beam particle microscope 1, which operates with a plurality of charged single-particle beams 3, is provided. Each detection channel of the multi-channel detection unit 209 is assigned one of the charged single-particle beams 3 for image generation. The multi-beam particle microscope can, for example, be the one described in Fig. Figure 1 shows a multi-beam particle microscope. The multi-channel detection unit 209 can, for example, be configured as shown in Fig. 2. Each channel of the multichannel detection unit 209 can, for example, include an avalanche photodiode.

[0078] In process step S1, a multitude of individual reference settings Ri are provided, each of which describes the reference setting Ri of an associated detection channel i. With this multitude of reference settings Ri, a multitude of individual images can be generated during a raster scanning of a reference sample, all of which fulfill a predefined detector criterion. For example, according to the detector calibration criterion, it is possible that all individual images within a defined first limit have the same brightness value and / or within a defined second limit have the same contrast value. This process step S1 focuses on providing the reference settings Ri; it is not necessary for them to be determined during the process.Instead, it is advantageous to have the individual reference settings Ri stored or available in a memory of the multibeam particle microscope 1.

[0079] In process step S2, the reference settings Ri of the multi-channel detection unit 209 are detuned. The multitude of individual reference settings Ri for each detection channel is changed by an individual detuning value Δi. Thus, Ri_detune = Ri + Δi. This individual detuning value Δi is generated individually for each detection channel i and is preferably chosen such that the individual detuning values ​​Δi of the detection channels are each pairwise distinct. Furthermore, the individual detuning value Δi for each channel is to be understood as meaning that at least one parameter of the detection channel i is changed for this individual detuning value Δi. However, multiple parameters of a detection channel can also be changed or detuned.These parameters include, for example, an offset and / or a gain of a detection channel or a detection unit arranged therein, such as an avalanche photodiode.

[0080] In process step S3, a first inspection point of a rechargeable sample, for example a wafer, is approached.

[0081] In process step S4, this first inspection point is scanned intermittently with the multitude of charged single-particle beams 3. Using the detuned settings Ri_detune of the multi-channel detection unit 209, a multitude of particle-optical individual images are generated. This process step thus involves an ensemble measurement with a multitude of detection units or detection channels i that are detuned relative to each other and set differently.

[0082] In process step S5, the large number of individual images are evaluated with regard to a predefined detection quality criterion. Evaluating the large number of individual images can include generating a brightness histogram for each individual image. An example of this is shown in Fig. 10 shown: The brightness histograms in Fig. The 10 values ​​are different for each detection channel or for each particle-optical image recorded with it.

[0083] In process step S6, the best individual image is determined from the multitude of individual images based on the detection quality criterion. Furthermore, the detuning Δi_best associated with the best individual image is determined. This is exemplified in Fig. 11 shown: The curve Ki that represents this based on Fig. The curve K3 best fulfills the detection quality criterion shown in section 7b. Firstly, curve K3 has the required minimum extent or histogram width (and thus a high contrast). Secondly, within intervals I1 and I2, the number of pixels with the respective brightness levels drops sharply in the outer regions of the brightness histogram, so that overall only a small number of pixels lie in intervals I1 and I2. This criterion is not fulfilled for the other curves K1, K2, K4, and K5. By determining the best single image or the best histogram curve K bestIt is now also known which adjustment Δi led to the generation of the best single image or the best brightness histogram. For example, it is now known which change in an offset and / or gain, such as in an avalanche photodiode, led to the best single image or the best brightness histogram of that single image.

[0084] In process step S7, a channel-independent global detuning Δ_global is determined based on the detuning Δi_best. It is possible that the global detuning Δ_global is identical to the detuning Δi_best associated with the best frame. However, it is also possible that the global detuning Δ_global is determined by interpolation. With this latter approach, one will typically not only determine the best detuning Δi_best and the best frame, but also the second-best frame with the second-best detuning Δi_best2.

[0085] In process step S8, the global detuning Δ_global is applied to the reference settings Ri of all detection channels i, thereby setting optimal Ri_opt settings for the multichannel detection unit 209. In this process step, for example, an offset and / or gain determined as the best detuning Δi_best is transferred to all detection channels i. For instance, the reference settings Ri are modified according to the best offset change and / or the best gain change. The known characteristic curves of the detection channels or detection units can also be used, since the offset and gain of a detection channel are linked by the characteristic curve. A fundamental assumption is that the sample to be inspected is essentially homogeneous.

[0086] In an optional process step S9, a second inspection point of the chargeable sample is approached, whereby the second inspection point differs from the first inspection point. In process step S10, the second inspection point is scanned using the multiple charged single-particle beams 3, whereby a multitude of particle-optical images are generated using the optimal Ri_opt settings of the multi-channel detection unit.

[0087] Procedure steps S9 and S10 can be repeated for a third inspection point or a further inspection point of the chargeable sample with the same optional Ri_opt settings, where the third or further inspection point is again different from those already visited. It is important to remember that a chargeable sample 7 is being inspected, and the inspection result depends on the illumination history of the inspection point. For this reason, each inspection point is visited or illuminated only once.

[0088] Fig. Figure 12 schematically shows brightness histograms of individual images generated with the optimal Ri_opt settings of the multi-channel detection unit 209: The brightness histogram of each curve K1 to K5 now fulfills the detection quality criterion, for example, the detection quality criterion determined by Fig.7 has been described in more detail. All curves K1 to K5 drop sharply in the boundary intervals I1 and I2, or even to zero, and only a few pixels lie in the intervals I1 and I2. Therefore, the second inspection point of the rechargeable sample 7 can be scanned with optimal detection settings for all channels i of the multichannel detection unit 209. The same applies to all subsequent inspection points on the rechargeable sample 7.

[0089] The description of the figures is not to be understood as limiting the present invention. Instead, the figures merely facilitate a better understanding of the invention.

[0090] A method for calibrating a multichannel detection unit of a multi-beam particle microscope during the inspection of rechargeable samples is disclosed. Time-consuming and space-consuming serial imaging with the multi-beam particle microscope for calibrating a multichannel detection unit during the inspection of rechargeable samples is avoided and replaced by a cleverly chosen ensemble image or ensemble measurement. The multi-beam capability of the multi-beam particle microscope is specifically used for detection calibration purposes. Reference symbol list 1. Multi-beam particle system, multi-beam particle microscope 3 primary particle beams, first single-particle beams 5 beam spots, points of impact 7. Object, sample, wafer 9 secondary particle beams, second single-particle beams 10 Computer system, control 15 Sample surface, wafer surface 25 pixels of a second single-particle beam 101 Object level 102 lens 103 Field lens 105 axle 108 Beam crossing, Cross-over 200 detector system 205 Projection lens system 206 Projection lens 207 Multi-particle detector 208 Projection lens 210 Projection lens 212 Beam crossing, Cross-over 214 aperture filter, contrast diaphragm 221 Beam path 222 Collective Anti-Deflection System 223 Light optics 225 lens 227 mirrors 229 lens 231 lens 235 Light receiving area 237 Light detection system 239 fiber optics 241 Light detector 243 area 245 Signal line 270 Control computer system 280 image generation computers 300 beam generating device 301 Particle source 303 Collimation lens system 304 multi-aperture array, filter plate 305 Micro-optics, multi-aperture arrangement, multi-beam particle generator 306 multi-aperture plate, multi-aperture array 307 Field lens, aperture plate 308 Field lens 309 Particle beam 321 Intermediate image plane 323 beam foci 400 beam switch, magnetic arrangement 500 scan deflectors 600 Moving table or positioning device x direction y direction z direction K curve of a brightness histogram d Offset β Amplification P1 Peak P2 Peak P av Brightness value C Contrast I1 Interval I2 interval

Claims

[1] Method for adjusting a multi-channel detection unit (207) of a multi-beam particle microscope (1) during an inspection of chargeable samples (7), wherein the multi-beam particle microscope (1) operates with a plurality of charged single-particle beams (3, 9), wherein each detection channel of the multi-channel detection unit (207) is assigned one of the charged single-particle beams (3, 9) for image generation, and wherein the method comprises the following steps: Providing (S1) the multi-channel detection unit (207) with a plurality of individual reference settings Ri, wherein each of the reference settings Ri describes the reference setting Ri of an associated detection channel, wherein with this plurality of reference settings Ri a plurality of individual images can be generated when scanning a reference sample which satisfy a predefined detector matching criterion; Detuning (S2) of the reference settings Ri of the multi-channel detection unit (207), wherein the multitude of individual reference settings Ri per detection channel is each changed by an individual detuning Δi, i.e. Ri_detune = Ri + Δi; Approach (S3) a first inspection point of a rechargeable sample (7); Raster scanning (S4) of the first inspection point with the multitude of charged single-particle beams (3), whereby a multitude of particle-optical single images are generated with the detuned settings Ri_detune of the multi-channel detection unit (207); Evaluating (S5) the large number of individual images with regard to a predefined detection quality criterion; Determining (S6) a best single image from the multitude of single images based on the detection quality criterion and determining the detuning Δi_best that is associated with the best single image; Determine (S7) a channel-independent global detuning Δ_global based on the detuning Δi_best; and Applying (S8) the global detuning Δ_global to the reference settings Ri of all detection channels i, thereby setting optimal settings Ri_opt of the multi-channel detection unit (207). [2] The method according to claim 1, further comprising the following steps: Approach (S9) a second inspection point of the chargeable sample (7), wherein the second inspection point is different from the first inspection point; and Raster scanning (S10) of the second inspection point with the multitude of charged single-particle beams (3), whereby with the optimal settings Ri_opt of the multi-channel detection unit (207) a multitude of particle-optical single images is generated. [3] Method according to any of the preceding claims, wherein the global detuning Δ_global is identical to the detuning Δi_best that is associated with the best frame. [4] Method according to one of claims 1 to 2, wherein the global detuning Δ_global is determined by interpolation or by extrapolation. [5] Method according to one of the preceding claims, wherein according to the detector calibration criterion all individual images within a defined first limit have the same brightness value (P av ) and / or exhibit the same contrast value (C) within a defined second boundary. [6] Method according to any one of the preceding claims, wherein each reference setting Ri of a detection channel of the multichannel detection unit (207) is determined by exactly one parameter; and where the individual detuning Δi for each detection channel comprises a detuning of exactly one parameter. [7] Method according to any one of claims 1 to 5, wherein each reference setting Ri of a detection channel of the multi-channel detection unit (207) is determined by at least two parameters, and where the individual detuning Δi for each detection channel comprises a detuning of at least two parameters. [8] Method according to any of the preceding claims, wherein each reference setting Ri of a detection channel and / or each individual detuning Δi is determined by values ​​and / or changes in values ​​of at least one of the following parameters: Gain (β), Offset (d). [9] Method according to one of the preceding claims, wherein the individual tunings Δi of the detection channels are each pair different from each other. [10] Method according to one of the preceding claims, wherein the individual tunings Δi with respect to at least one parameter are each generated by incremental changes of Δi by a constant value k. [11] Method according to one of the preceding claims, wherein the evaluation of the plurality of individual images comprises generating brightness histograms of the individual images. [12] Method according to the preceding claim, wherein the detection quality criterion is determined based on at least one of the following criteria: Frequencies of brightness levels in a first interval (11) of the brightness histogram, which represents very low brightness levels; Frequencies of brightness levels in a second interval (I2) of the brightness histogram, which represents very high brightness levels; Total width of the frequency distribution with frequencies other than zero; Frequency of maximum detectable brightness; Frequency of the minimum detectable brightness. [13] Method according to the preceding claim, where, according to the detection quality criterion, a single image is rated better the less frequently brightness levels are counted in the first interval (11); and / or where, according to the detection quality criterion, a single image is rated better the less frequently brightness levels are counted in the second interval (I2). [14] Method according to any one of claims 12 to 13, where a detuned setting Ri_detune of a detection channel meets a minimum quality requirement if a maximum of 0.20%, in particular a maximum of 0.15% or a maximum of 0.10%, of all brightness values ​​lie within the first interval (11); and / or where a detuned setting Ri_detune of a detection channel meets a minimum quality requirement if a maximum of 0.20%, in particular a maximum of 0.15% or a maximum of 0.10%, of all brightness values ​​lie within the second interval (I2). [15] Method according to one of claims 12 to 14, wherein a detuned setting Ri_detune of a detection channel meets a minimum quality requirement if a maximum of 0.20%, in particular a maximum of 0.15% or a maximum of 0.10%, of all brightness values ​​lie within the first and second intervals (I1, I2). [16] Method according to one of the preceding claims, wherein the multi-channel detection unit (207) comprises a combination of a particle detector (200) and a light detection unit (237). [17] Method according to the preceding claim, wherein the light detection unit (237) comprises a plurality of avalanche photodiodes and wherein each detection channel is assigned one avalanche photodiode, and wherein when the reference settings Ri of the multichannel detection unit (207) are detuned, a gain (β) and / or an offset (d) of an avalanche photodiode is changed. [18] Computer program product comprising program code for performing the method according to any of the preceding claims. [19] Multi-beam particle microscope (1) configured to perform the method according to any one of claims 1 to 17.

Images