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

Abstract

The invention relates to a method for setting a multi-channel detection unit of a multi-beam particle microscope during an inspection of charging samples. Time-consuming and space- consuming serial recordings by the multi-beam particle microscope that serve the purpose of setting a multi-channel detection unit during an inspection of charging samples are avoided and replaced with a skilfully chosen ensemble recording or ensemble measurement. The multi- beam property of the multi-beam particle microscope is used in a targeted manner for detection setting purposes.

Description

[0001] Method for setting a multi-channel detection unit of a multi-beam particle microscope during an inspection of charging samples, associated computer program product and multi-beam particle microscope

[0002] Field of the invention

[0003] In general, the invention relates to multi-beam particle microscopes which operate using a multiplicity of individual particle beams. Specifically, the invention relates to a method for setting a multi-channel detection unit of a multi-beam particle microscope during an inspection of charging samples, an associated computer program product and an associated multi-beam particle microscope.

[0004] Prior art

[0005] With the ongoing development of ever smaller and ever more complex microstructures such as semiconductor components, there is a need to further develop and optimize planar production techniques and inspection systems for producing and inspecting small dimensions of the microstructures. For instance, the development and production of the semiconductor components require monitoring of the design of test wafers, and the planar production techniques require process optimization for reliable production with high throughput. Moreover, there have been recent demands for an analysis of semiconductor wafers for reverse engineering and for a customized, individual configuration of semiconductor components. Therefore, there is a need for inspection means which can be used with high throughput to examine the microstructures on wafers with high accuracy.

[0006] Typical silicon wafers used in the production of semiconductor components have diameters of up to 300 mm. Each wafer is divided into 30 to 60 repeating regions ("dies") with a size of up to 800 mm2. A semiconductor device comprises multiple semiconductor structures, which are produced in layers on a surface of the wafer by planar integration techniques. Semiconductor wafers typically have a plane surface on account of the production processes. The structure size of the integrated semiconductor structures in this case extends from a few pm to the critical dimensions (CDs) of 5 nm, and the structure sizes will become even smaller in the near future; in future, structure sizes or critical dimensions (CDs) are expected to be less than 3 nm, for example 2 nm, or even less than 1 nm. In the case of the aforementioned small structure sizes, defects of the order of the critical dimensions must be identified quickly over a very large area. For multiple applications, the specification requirement regarding the accuracy of a measurement provided by an inspection device is even higher, for example by a factor of two or one order of magnitude. For example, a width of a semiconductor feature must be measured with an accuracy better than 1 nm, for example 0.3 nm or even less, and a relative position of semiconductor structures must be determined with an overlay accuracy better than 1 nm, for example 0.3 nm or even less.

[0007] The MSEM, a multi-beam scanning electron microscope, is a relatively new development in the field of charged particle systems (“charged particle microscopes”, CPMs). For example, a multi-beam scanning electron microscope is disclosed in US 7 244 949 B2 and in US 2019 / 0355544 A1. In the case of a multi-beam electron microscope or MSEM, a sample is irradiated simultaneously by a multiplicity of individual electron beams arranged in a field or raster. For instance, 4 to 10 000 individual electron beams may be provided as primary radiation, with each individual electron beam being separated from an adjacent individual electron beam by a pitch of 1 to 200 micrometres. For example, an MSEM has approximately 100 separate individual electron beams ("beamlets"), which are arranged for example in a hexagonal grid, with the individual electron beams being separated by a pitch of approximately 10 pm. The multiplicity of individual charged particle beams (primary beams) are focused, in each case on an individual basis, on the surface of a sample to be examined by way of common large-field optics unit including, inter alia, a common objective lens. For example, the sample can be a semiconductor wafer that is secured to a wafer holder mounted on a movable stage. When the wafer surface is illuminated by the primary individual charged particle beams, interaction products, for example secondary electrons or backscattered electrons, emanate from the surface of the wafer. Their respective start points correspond to those locations on the sample on which the plurality of primary individual particle beams are focused in each case. The amount and the energy of the interaction products depend inter alia on the material composition and the topography of the wafer surface. The interaction products form multiple secondary individual particle beams (secondary beams) that are collected by the common objective lens and, following a passage through a projection imaging system of the multi-beam inspection system, are incident on a detector arranged in a detection plane. The detector comprises multiple detection regions, each of which may comprise multiple detection pixels, and the detector acquires an intensity distribution for each of the secondary individual particle beams. An image field of 100 pm x 100 pm, for example, is obtained in the process.

[0008] The multi-beam electron microscope of the prior art comprises a sequence of electrostatic and magnetic elements. At least some of the electrostatic and magnetic elements can be set in order to adapt the focus position and the stigmation of the multiplicity of individual charged particle beams. The state-of-the-art multi-beam system with charged particles moreover comprises at least one crossover plane of the primary or the secondary individual charged particle beams. Moreover, the state-of-the-art system comprises detection systems to facilitate setting. The state-of-the-art multi-beam particle microscope comprises at least one beam deflector (deflection scanner) for collective scanning of a region of the sample surface by means of the multiplicity of primary individual particle beams in order to obtain an image field of the sample surface.

[0009] As already mentioned above, each secondary beam is incident on a separate detector element, or detector element assigned thereto. However, even in the case of detector elements that are structurally identical in principle, these detector elements are normally not 100% identical. Rather, their detector characteristic curves, in particular, may deviate from one another, which may result in different brightness and / or contrast values of individual images, even if exactly the same structure was scanned for the individual images. These differences in brightness and / or in contrast may prove to be problematic e.g. if different individual images are intended to be juxtaposed (so-called stitching). Moreover, the differences discussed play a part if dimensions of surface structures extending over a plurality of individual images are intended to be measured. Contour identification can also be made more difficult as a result of the differences in brightness and / or contrast values because here the edges of the individual images could be interpreted incorrectly as a contour.

[0010] To solve the aforementioned problem, WO 2020 / 057678 A1 has disclosed the practice of setting a detection unit with multiple detector elements or detection channels on the basis of recordings of a reference sample. In this case, the various detector elements or detection channels are matched to each other such that the individual images generated during a rastertype scan of the reference sample have the same brightness level within a predetermined limit and / or the same contrast value within a further predetermined limit. The disclosure of WO 2020 / 057678 A1 is incorporated in this patent application in full by reference.

[0011] Although a detector calibration is in principle known per se, there is another problem: A setting of detection units of a multi-channel detection unit, once generated on the basis of a reference sample, in general cannot be transferred to other samples. This is because an optimal detection setting depends on multiple parameters. The physical nature of a sample or sample material is one of these parameters. However, other influencing factors also influence an optimal detection setting for an image recording. For example, these include the beam current, the landing energy, the pixel size and the dwell time per pixel, with this list not being exhaustive. Therefore, at the beginning of each inspection process, a new detector calibration or detector setting is implemented as a matter of principle on the basis of the sample that is actually to be inspected.

[0012] In the case of a scanning process of a non-charging sample, a secondary electron yield is constant or does not depend on the illumination history as a matter of principle. Accordingly, in the case of non-charging samples, it is possible to scan the same inspection site multiple times without this generating different images / measurement results during the repeated measurement. Therefore, it is also possible to implement a detector calibration or any other detector setting at the same inspection site on a non-charging sample, even if this procedure requires a sequence of multiple image recordings. It is therefore possible to manually set the detector while observing live images. Moreover, a setting of a detector can also be automated for a non-charging sample.

[0013] Then again, in the case of a charging sample, the situation is different: When scanning charging samples, the generated secondary electron yield depends on the illumination history as a matter of principle. It is therefore not possible to perform multiple scans / image recordings for a detector setting at the same inspection site. This would falsify the recording result. Instead, there is a move to a new inspection site, which has not yet been scanned or illuminated, for each scanning procedure. Firstly, this move takes time and, secondly, this uses up / destroys many inspection sites, and these are no longer available for the actual inspection task.

[0014] The known methods for setting a detection unit or multi-channel detection unit during an inspection of charging 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.

[0015] US 2020 / 0211811 A1 discloses a multi-beam inspection apparatus. It addresses the problem that a non-uniformity of corresponding probe spots increases with the number of beamlets. US 2020 / 0211811 A1 suggests to provide an array of multi-deflectors with different multi-pole electrodes, wherein the number of electrodes of a multi-pole electrode increases with the distance to the particle optical axis.

[0016] US 6,563,114 B1 deals with the measurement of physical and / or chemical characteristics of a substrate or wafer as a function of a focus setting. It describes how a brightness histogram can be used to determine whether a focusing condition in the secondary path is met. DE 10 2013 016 113 A1 discloses a special design of the detector, namely a combination of a particle detector and a light detector. To compensate for the decrease in efficiency / quality of particle detection over longer periods of operation using a scintillator plate, it is proposed to move the scintillator plate sideways after a certain period of time. This allows areas of the scintillator plate that were previously not used for scintillation to be used for particle detection.

[0017] Description of the invention

[0018] Therefore, the problem addressed by the present invention is that of providing an improved method for setting a multi-channel detection unit during an inspection of charging samples. The method should work quickly, precisely and in a space-saving manner. In particular, it should also be suitable for wafer inspection.

[0019] The problem is solved by the subject matter of the independent claims. Advantageous embodiments of the invention are evident from the dependent patent claims.

[0020] The present patent application claims the priority of the German patent application No. 102024 128 162.6 filed on 27 September 2024, the disclosure of which in the full scope thereof is incorporated in the present patent application by reference.

[0021] A basic concept of the invention is that of avoiding time-consuming and space-consuming serial recordings by the multi-beam particle microscope that serve the purpose of setting a multi-channel detection unit during an inspection of charging samples and of replacing these with a skilfully chosen ensemble recording or ensemble measurement. Therefore, instead of having to invest evermore effort for setting a corresponding number of detection channels in the event of an increasing number of individual particle beams, the multi-beam property of the multi-beam particle microscope is used specifically for detection adjustment purposes.

[0022] According to a first aspect of the invention, the latter relates to a method for setting a multichannel detection unit of a multi-beam particle microscope during an inspection of charging samples or samples prone to charging, wherein the multi-beam particle microscope operates with a multiplicity of individual charged particle beams, wherein each detection channel of the multi-channel detection unit is assigned one of the individual charged particle beams for image generation purposes and wherein the method comprises the following steps: providing the multi-channel detection unit with a multiplicity of individual reference settings Ri, wherein each of the reference settings Ri describes the reference setting Ri for an associated detection channel, wherein a multiplicity of individual images that meet a predefined detector calibration criterion can be generated using this multiplicity of reference settings Ri during a raster-type scan of a reference sample; detuning the reference settings Ri for the multi-channel detection unit, wherein the multiplicity of individual reference settings Ri per detection channel are each modified by an individual detuning Ai , i.e. Ri_detune = Ri + Ai; moving to a first inspection site of a charging sample; raster-scanning the first inspection site using the multiplicity of individual charged particle beams, wherein a multiplicity of particle-optical individual images are generated with the detuned settings Ri_detune of the multi-channel detection unit; evaluating the multiplicity of individual images with respect to a predefined detection quality criterion; determining a best individual image from the multiplicity of individual images on the basis of the detection quality criterion and determining the detuning Ai_best associated with the best individual image; ascertaining a channel-independent global detuning A_global on the basis of the detuning Ai_best; and applying global detuning A_global to the reference settings Ri for all detection channels i, whereby optimal settings Ri_opt of the multi-channel detection unit are set.

[0023] For example, the individual charged particle beams can be electrons, positrons, muons or ions or other charged particles. The individual field regions of the object that are assigned to each individual particle beam are raster-scanned, for example line by line or column by column. Overall, the individual particle beams may be arranged in a raster-type configuration, for example in a rectangular grid or in a hexagonal grid. In the hexagonal case, it is advantageous if the number of individual charged particle beams is 3n(n - 1) + 1 , where n is any natural number. The interaction products can be backscattered electrons or else secondary electrons. In this case, it is thus preferred for analysis purposes if the low-energy secondary electrons are used to generate the image.

[0024] The multi-channel detection unit according to the invention or the multi-detector according to the invention can comprise one detector or multiple detectors of the same type or of different types. The detectors themselves may be pixelated but this need not be the case. The multichannel detection unit may e.g. comprise or consist of one or more particle detectors. A particle detector in turn may be embodied in one piece or in many pieces. However, it is also possible to combine one or more particle detectors and light detectors with one another, or to connect them in series, in 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 individual charged particle beam. The number of detection channels thus corresponds to the number of individual charged particle beams used for imaging purposes.

[0025] According to the invention, the multi-channel detection unit is initially provided with a multiplicity of individual reference settings Ri, wherein each of the reference settings Ri describes the reference setting Ri for an associated detection channel. When raster-scanning a reference sample, a multiplicity of individual images that meet a predefined detector calibration criterion can be generated with this multiplicity of reference settings Ri. In this case, the reference sample is not identical to the charging sample that is to be inspected. Provision of the multichannel detection unit with a multiplicity of individual reference settings Ri does not mean that the corresponding detector calibration or the ascertainment of the reference settings Ri is also performed over the course of the method according to the invention, although this may be the case. Instead, it is preferably the case that for the respective multi-beam particle microscope that is used for the method according to the invention, the corresponding ascertainment of reference settings Ri for the multi-channel detection unit has already been implemented before the method is performed. For example, the individual reference settings Ri may be stored in a memory of the multi-beam particle microscope and read from the memory into a controller of the multi-beam particle microscope for the method according to the invention. It is also possible that multiple sets of reference settings Ri for the multi-beam particle microscope are stored in a memory, for example dependent on a sample material and / or dependent on different operating points of the multi-beam particle microscope. The reference sample may be a sample that is similar to the charging sample that is to be inspected, although this need not be the case. The reference sample may also be a non-charging sample. Nevertheless, it is meaningful to have the best possible individual reference setting Ri at the start of the method according to the invention for setting a multi-channel detection unit of a multi-beam particle microscope during an inspection of charging samples.

[0026] The reference settings Ri for the multi-channel detection unit is detuned in a further method step, wherein the multiplicity of individual reference settings Ri per detection channel are each modified by an individual detuning Ai , i.e. Ri_detune = Ri + Ai. The setting for each detection channel is thus modified in a targeted manner in this method step, and so an ensemble recording with different detection settings can be made in a very targeted manner during a subsequent image recording. Each detection channel is detuned by Ai on an individual basis. In this case, it is preferably the case that the individual detunings Ai of the detection channels in each case differ from each other in pairwise fashion. Then, it is possible to subsequently record an image with the largest possible ensemble of different detection channel settings. The individual detunings Ai are comparatively small in this case. The modification of the setting for the detection channels therefore tends to be within the scope of a fine adjustment and not within the scope of a complete readjustment. It is preferably the case that a detuning Ai is a maximum of + / - 10%, preferably a maximum of + / -8% or a maximum of + / -5%, of the value of the reference setting Ri.

[0027] Then, according to the invention, there is a move to a first inspection site of a charging sample, for example a wafer. However, the sample may also be any other charging sample.

[0028] Then there is raster-scanning of the first inspection site using the multiplicity of individual charged particle beams, wherein a multiplicity of particle-optical individual images are generated with the detuned settings Ri_detune of the multi-channel detection unit. This method step corresponds to the aforementioned ensemble measurement.

[0029] Then there is the evaluation of the multiplicity of individual images with respect to a predefined detection quality criterion. This detection quality criterion may be defined in various ways. The detection quality criterion may in turn consist of several sub-criteria, but it is also possible that a single criterion is sufficient as the detection quality criterion. Specific examples of the definition of a detection quality criterion are described in more detail below.

[0030] Then there is the determination of the best individual image from the multiplicity of individual images on the basis of the detection quality criterion and a determination of the detuning Ai_best associated with the best individual image. Thus, the best individual image is selected or ascertained from the ensemble of individual images. Since the detuning Ai_best that has led to the recording of the best individual image in a certain detection channel is now known in principle, this knowledge is used to also set the other detection channels accordingly:

[0031] According to the invention, a channel-independent global detuning A_global is ascertained on the basis of the detuning Ai_best. In this case, the term "global" indicates that the ascertained detuning A_global is identical for all detection channels.

[0032] Thereupon, the global detuning A_global is applied to the reference settings Ri for all detection channels i, whereby optimal settings Ri_opt of the multi-channel detection unit are set. These optimal settings Ri_opt are specific to the charging sample that is to be inspected. The settings Ri_opt are optimal in the sense that the setting or fine adjustment of the multi-channel detection unit has now been carried out so well that, in a further optional method step, the actual sample inspection may be continued directly with the ascertained optimal settings Ri_opt. When the global detuning A_global is applied to the reference settings Ri for all detection channels i, the detuning Ai_best may be transformed from the channel i_best to the other channels. These transformations may be based on the characteristic curve or a family of characteristic curves that describes or describe a detection unit of a respective detection channel.

[0033] According to a preferred embodiment of the invention, there is a move to a second inspection site of the charging sample, wherein the second inspection site differs from the first inspection site. Then there is raster-scanning of the second inspection site using the multiplicity of individual charged particle beams, wherein a multiplicity of particle-optical individual images are generated with the optimal settings Ri_opt of the multi-channel detection unit. Analogously, there may be a move to a third inspection site, a fourth inspection site, a fifth inspection site and many further inspection sites of the charging sample, and these may be raster-scanned using the multiplicity of individual charged particle beams, wherein the optimal settings Ri_opt of the multi-channel detection unit are used again for the image recording. Thus, according to the invention, only a single inspection site needs to be used for setting the multi-channel detection unit. All other inspection sites are not used to set the multi-channel detection unit but are instead available for the actual inspection process.

[0034] According to a preferred embodiment of the invention, the global detuning A_global is identical to the detuning Ai_best associated with the best individual image. In this embodiment of the invention, the accuracy with which a best detuning Ai_best can be ascertained is thus predetermined by the proximity of the different detunings Ai to one another. Many very different detunings Ai lead to a less accurate result; a multiplicity of detunings Ai that differ only slightly from one other allow for a more accurate definition of the global detuning A_global. The latter embodiment variant is preferred. For this, however, it may be important to start the method according to the invention with already relatively well-known individual reference settings Ri.

[0035] According to an alternative embodiment of the invention, the global detuning A_global is ascertained by interpolation. In this case, it is not only the best individual image and the associated detuning Ai_best that is ascertained, but also the second-best individual image with the associated detuning Ai_best2. Normally, these two detunings differ only slightly from each other or are adjacent to each other in the configuration space, and so an even more accurate value for the global detuning A_global can be ascertained by means of interpolation.

[0036] A reference setting Ri is assigned to each detection channel at the beginning of the method according to the invention. In this context, this reference setting Ri may be defined by one parameter or by multiple parameters. In a series connection of different detector types within a detection channel, i.e. 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 the channel i may even be particularly large. What applies in general is that a reference setting Ri may be defined by a plurality of parameters j = 1 , ... , L, with L e N.

[0037] According to a preferred embodiment of the invention, each reference setting Ri for a detection channel of the multi-channel detection unit is defined by exactly one parameter. In this case, j = 1 applies. Then the individual detuning Ai for each detection channel i comprises detuning of this exactly one parameter.

[0038] According to an alternative embodiment of the invention, each reference setting Ri for a detection channel of the multi-channel detection unit is defined by at least two parameters, i.e. j > 2. In this case, the individual detuning Ai for each detection channel i comprises detuning of the at least two parameters. What applies in general is that the reference setting Ri and the individual detuning Ai of this reference setting Ri are defined in the same parameter space or configuration space.

[0039] According to a preferred embodiment of the invention, each reference setting Ri for a detection channel i and / or each individual detuning Ai are / is defined by values and / or changes in value of at least one of the following parameters: gain, offset. In practice, the two parameters, gain and offset, are the two most important setting parameters when setting a detection unit or a detection channel using such a detection unit.

[0040] In principle, the gain indicates how much output is generated by how much input. Specifically, the gain indicates the number of particles of the second particle type (out) contained in the output in relation to the number of particles of the first particle type (in) contained in the input. In the case of an avalanche photodiode, the input is formed by photons and the output is electrons. The same may be defined analogously for all other detectors, DED (“direct electron detection”), PMT (“photomultiplier tube”), etc. Furthermore, many detectors that carry out conversion into an electrical signal (current or voltage drop across an output resistor) have a post amplifier built in, the gain of which is likewise settable.

[0041] The offset in turn indicates how high the level of the output signal is if no primary particles arrive. That is generally solved in the downstream electronic circuitry by means of a voltage adder. Ultimately, leakage currents, etc. are thus compensated for by means of these systems. Offset and gain are generally not independent of one another. According to a preferred embodiment of the invention, the individual detunings Ai of the detection channels i in each case differ from each other in pairwise fashion. This allows the largest possible ensemble measurement to be undertaken for the purpose of setting the multichannel detection unit. According to a preferred embodiment of the invention, the individual detunings Ai with respect to at least one parameter are each generated by incremental changes of Ai by a constant value k. For example, it is thus possible to undertake a detuning of the offset by different values for each detection channel, wherein adjacent values differ by the constant value k. In principle, a gain may also be modified incrementally in analogous fashion for the various detection channels. Moreover, it is possible to initially keep the value of one parameter constant and modify another parameter in other detection channels in a targeted manner, and thereupon modify the first parameter once in a targeted manner and then run through the value range of the second parameter in the other detection channels again, etc. In this way, as it were, a matrix of individual detunings Ai(ik) can be generated. A solution space for the best optimal settings Ri_opt can be covered very well in this way. However, there are also many cases in which the individual detunings Ai only need to be performed in relation to one parameter in order to optimally cover the solution space.

[0042] According to a preferred embodiment of the invention, the evaluation of the multiplicity of individual images comprises a generation of brightness histograms of the individual images. There is no need here to output or explicitly display these brightness histograms in the form of graphics. All that matters is the recording of brightness histograms in a computational manner such that they can be evaluated in a suitable form on the basis of a detection quality criterion. The usual effect on a brightness histogram of an individual detuning Ai in the form of a change in the offset is that the curve in the histogram or the values in the histogram are merely displaced. An individual detuning Ai in the form of a change in contrast usually has more complex consequences and modifies both the width of the distribution and the brightness value or mean brightness value of the distribution.

[0043] According to a preferred embodiment of the invention, the detection quality criterion is defined on the basis of at least one of the following criteria: frequencies of brightness levels in a first interval of the brightness histogram, which represents very low levels of brightness. This interval thus describes an edge region of the brightness histogram, wherein it is possible but not mandatory that also the brightness or the brightness value of zero is encompassed by this first interval or edge interval. frequencies of brightness levels in a second interval of the brightness histogram, which represents very high levels of brightness. This second interval therefore denotes a second edge region with very high or even the highest levels of brightness, wherein it is again possible but not mandatory that the greatest or maximum brightness is located within the second interval. overall width of the frequency distribution with non-zero frequencies. In this way, the entire brightness range or range of greyscale values (for example 256 bits) can be made available for an image evaluation. frequency of maximum detectable brightness. If this frequency is very high, this is an indication that the detection channel is not yet optimally set and that instead the information is not optimally resolved at high brightness. frequency of minimum detectable brightness. What applies analogously in this case is that a detection channel setting is not optimal if a high frequency is detected with lowest brightness.

[0044] This list of criteria should not be construed as exhaustive but only be understood to be exemplary.

[0045] According to the detection quality criterion, an individual image is assessed as being better according to a preferred embodiment of the invention the less frequently brightness levels are counted in the first interval. In addition to that or in an alternative, according to the detection quality criterion, an individual image is assessed as being better the less frequently brightness levels are counted in the second interval. The idea behind these two assessments is that the edge regions of a brightness histogram do not contain too much information or that the information is not resolved in these regions and shall be lost. Then again, it is preferable for the overall width of the frequency distribution to be large enough to be able to resolve information well. This is not the case should a frequency distribution only have a single sharp peak. In that case, many brightness or greyscale values remain unused, and the particle- optical image contains less detailed information.

[0046] According to a preferred embodiment of the invention, a detuned setting Ri_detune of a detection channel meets a minimum quality requirement should a maximum of 0.20%, preferably a maximum of 0.15% or most preferably 0.10% of all brightness values be located within the first interval. In addition to that or in an alternative, a detuned setting Ri_detune of a detection channel meets a minimum quality requirement should a maximum of 0.20%, preferably a maximum of 0.15% or most preferably 0.10% of all brightness values be located within the second interval. Hence the edge regions of the brightness histogram contain only little information or low frequencies; nevertheless, there is preferably information to ensure that all greyscale values or bits of the brightness histogram are also used meaningfully. For example, for a brightness histogram with greyscale values from 0 to 255, a size of the edge regions or intervals 11 , I2 may be set as follows: 11 = [10, 70] and I2 = [170, 245], This corresponds to a definition in accordance with ISO TS 24597. However, other interval sizes and boundaries are also possible.

[0047] According to a preferred embodiment of the invention, a detuned setting Ri_detune of a detection channel meets a minimum quality requirement should a maximum of 0.20%, preferably a maximum of 0.15% or most preferably a maximum of 0.10%, of all brightness values be located within the first and the second interval. Thus, the brightness values of the first and second intervals are considered together in this case in order to check whether the minimum quality requirement is met. Self-evidently, it is possible to combine this common criterion for the first and second intervals with the separate criteria for each interval, as described above.

[0048] 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. For example, the particle detector may be a scintillator plate. A light signal is emitted every time second or secondary individual particle beams are incident on the scintillator plate, and said light signal may then be evaluated separately in the event of a fixed assignment to the incidence location on the particle detector or the scintillator plate.

[0049] According to a preferred embodiment of the invention, the light-detection unit comprises a multiplicity of avalanche photodiodes. In this case a respective avalanche photodiode is assigned to each detection channel. A gain and / or an offset of an avalanche photodiode are / is modified, respectively, when the reference settings Ri for the multi-channel detection unit are detuned. 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-beam particle microscopes. For example, a gain and / or an offset for an avalanche photodiode can be set in each case by way of a change in current or voltage.

[0050] According to a second aspect of the invention, the latter relates to a computer program product having a program code for carrying out the method as described above in multiple embodiment variants. In this case, the program code may in principle be written in any desired programming language. In particular, the program code may have a modular structure. For example, it is useful to provide a code for controlling the multi-beam particle microscope separately in one module, while another module contains the routines for actually setting the multi-channel detection unit.

[0051] According to a third aspect of the invention, the latter relates to a multi-beam particle microscope configured to carry out the method according to one of the embodiment variants of the first aspect of the invention described above.

[0052] The above-described exemplary embodiments and aspects of the invention can be combined with one another in full or in part, provided that no technical contradictions arise as a result of this combination.

[0053] The invention will be understood even better with reference to the accompanying figures, in which:

[0054] Fig. 1: shows a schematic illustration of a multi-beam particle microscope (MSEM);

[0055] Fig. 2: shows a schematic illustration of the structure of a multi-channel detection unit;

[0056] Fig. 3: schematically shows a multi-image with a multiplicity of individual images;

[0057] Fig. 4: shows different brightness histograms by way of example and illustrates their relationship with a detector characteristic curve;

[0058] Fig. 5: schematically shows effects of a modified detector offset on a brightness histogram;

[0059] Fig. 6: schematically shows effects of a modified gain on a brightness histogram;

[0060] Fig. 7: schematically illustrates criteria for a detection quality criterion;

[0061] Fig. 8: schematically illustrates an adjustment setting of a detection channel for fulfilling a detection quality criterion for each individual image or for each individual particle beam;

[0062] Fig. 9: schematically illustrates the method according to the invention;

[0063] Fig. 10 schematically shows a combined graphic of brightness histograms of a multiplicity of individual images, which have been recorded simultaneously with the multiplicity of individual particle beams that each have individually detuned detector settings;

[0064] Fig. 11 : schematically shows the determination of a best individual image from the multiplicity of individual images on the basis of a detection quality criterion; and

[0065] Fig. 12: schematically shows brightness histograms following the setting, according to the invention, of the multi-channel detection unit. Fig. 1 schematically shows a multiple particle beam system 1 in the form of a multi-beam particle microscope 1. The multi-beam particle microscope 1 comprises a beam-generating apparatus 300 having a particle source, for example an electron source. By means of the beam-generating apparatus 300, charged particles or electrons are generated for example by means of thermal field emission. The emitted charged particles form a divergent particle beam 309, and the latter is collimated by a sequence of condenser lenses 303.1 and 303.2 and incident on a multi-beam particle generator 305 having a multi-aperture arrangement. The multi-beam particle generator 305 comprises multiple multi-aperture plates 304, 306 and a field lens 307. The multi-beam particle generator 305 generates a multiplicity of individual particle beams 3 or individual electron beams 3, which are arranged in a field, which is imaged onto a further field formed by beam spots 5 in the object plane 101. The pitch between centre points of apertures in a multi-aperture plate 306 can be for example 5 pm, 100 pm and 200 pm. The diameters D of the apertures are smaller than the pitch between the centre points of the apertures; examples of the diameters are 0.2 times, 0.4 times and 0.8 times the pitches between the centre points of the apertures.

[0066] The multi-aperture arrangement 305 and the field lens 308 are configured to generate a multiplicity of focal points 323 of primary beams 3 in a grid arrangement on a surface 321 . The surface 321 need not be a plane surface but rather can be a spherically curved surface in order to account for an image field curvature of the subsequent particle-optical system.

[0067] The multi-beam particle microscope 1 furthermore comprises a system of electromagnetic lenses 103 and an objective lens 102, which image the beam foci 323 from the intermediate image surface 321 into the object plane 101 with reduced size. In between, the first individual particle beams 3 pass through the beam splitter 400 and a collective beam deflection system 500, by means of which the multiplicity of first individual particle beams 3 are deflected during operation and the image field is scanned. The first individual particle beams 3 incident in the object plane 101 form for example a substantially regular field, wherein pitches between adjacent incidence locations 5 can be for example 1 pm, 10 pm or 40 pm. The field formed by the incidence locations 5 can have a rectangular or hexagonal symmetry, for example.

[0068] The object 7 to be examined may be of any desired type, for example a semiconductor wafer or a biological sample, and may comprise an arrangement of miniaturized elements or the like. The surface 15 of the object 7 is arranged in the object plane 101 of the objective lens 102. The objective lens 102 may comprise one or more electron-optical lenses. For example, this may be a magnetic objective lens and / or an electrostatic objective lens. The primary particles 3 incident on the object 7 generate interaction products, for example secondary electrons, backscattered electrons or primary particles, which have experienced a reversal of movement for other reasons, and these interaction products emanate from the surface of the object 7 or from the first plane 101 or object plane 101 . The interaction products emanating from the surface 15 of the object 7 are shaped by the objective lens 102 to form secondary particle beams 9. In the process, the secondary beams 9 pass through the beam splitter 400 downstream of the objective lens 102 and are supplied to a projection system 200. The projection system 200 comprises an imaging system 205 with projection lenses 206, 208 and 210, a contrast stop 214 and a multi-particle detector 207. Incidence locations 25 of the second individual particle beams 9 on detection regions of the multi-particle detector 207 are located with a regular pitch in a third field. Exemplary values are 10 pm, 100 pm and 200 pm.

[0069] The multi-beam particle microscope 1 furthermore comprises a computer system or control unit or controller 10, which in turn can be embodied integrally or in multipartite fashion and which is designed both to control the individual particle-optical components of the multi-beam particle microscope 1 and to evaluate and analyse the signals obtained by the multi-detector 207 or detection unit.

[0070] In this case, the computer system 10 may be constructed from a plurality of individual computers or corresponding components. According to a preferred embodiment variant, the computer system 10 comprises a control computer system 270 and one or more image generating computers 280. The provision of multiple image generating computers 280 is advantageous on account of the high volumes of data that arise and allows detector signals to be evaluated in parallel. The method according to the invention may also be performed with the aid of the described computer system 10, i.e. the computer system 10 is designed inter alia to perform the method according to the invention by means of a program.

[0071] Further information relating to such multi-beam particle beam systems or multi-beam particle microscopes 1 and component parts used therein, such as, for instance, particle sources, multi-aperture plate 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 102013016 113 A1 and DE 102013 014 976 A1 , the disclosure of each of which is incorporated in this patent application in full by reference. Fig. 2 is a schematic illustration for elucidating a number of details of the detector 209. As particle detector, the detector 209 in this case comprises the scintillator plate 207, onto which the interaction products, for example secondary electron beams, are directed by an electron optics unit. When integrated into the multi-beam particle microscope 1 from Fig. 1 , this electron optics unit comprises the electron-optical components of the particle optics unit that shape the electron beams 9, i.e. for example the objective lens 102, that direct the electron beams 9 towards the detector 209, such as e.g. the beam splitter 400, and that focus the electron beams 9 on the surface of the scintillator plate 207, such as e.g. the lens 2010. The electron beams 9 are incident on the scintillator plate 207 at incidence locations 25. Even if the electron beams 9 are focused on the surface of the scintillator plate 207, beam spots having diameters that are not arbitrarily small are formed on the surface. The midpoints of the beam spots can be regarded as the incidence locations 25, which are arranged at the pitch P2 from one another.

[0072] The scintillator plate 207 contains a scintillator material that is excited to emit photons by the incident electrons of the electron beams 9. Each of the incidence locations 25 thus forms a source of photons. Fig. 2 illustrates just a single corresponding beam path 221 that emanates from the incidence location 25 of the central electron beam of the five electron beams 9 illustrated. The beam path 221 passes through a light-optics unit 223, which comprises a first lens 225, a mirror 227, a second lens 229 and a third lens 231 in the example shown, and is subsequently incident on 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 fibre 239, into which at least some of the photons are coupled and guided to a light detector 241. The light detector 241 may comprise e.g. a photomultiplier, an avalanche photodiode, a photodiode or other types of suitable light detectors. The light-optics unit 223 is configured such that it optically images the surface 208 of the scintillator plate 207 into a region 243 in which the light-receiving surface 235 is arranged. On account of this optical imaging, optical image representations of the incidence locations 25 are generated in the region 243. In the region 243, a separate lightreceiving surface 235 of the light-detection system 237 is provided for each of the incidence locations 25. Each of the further light-receiving surfaces 235 is formed by an end face of an optical waveguide 239, which guides the light input coupled into the end face to a light detector 241 . On account of the optical imaging, a light-receiving surface 235 is assigned to each of the incidence locations 213, wherein 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. Consequently, 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 regions. On account of the electron optics unit described above, interaction products, for example electrons, which emanate from two different individual field regions of an object are also projected onto different detection regions of the scintillator plate 207. In the exemplary embodiment explained here, the light detectors 241 are arranged at a distance from the lightreceiving surfaces 235, onto which the light-optics unit 223 images the scintillator plate 207, and the received light is guided to the light detectors 241 through optical fibres 239. However, it is also possible for the light detectors 241 to be arranged directly where the light-optics unit generates the image of the scintillator plate, and the light-sensitive surfaces of the light detectors thus form the light-receiving surfaces.

[0073] In this case - as stated - Fig. 2 merely schematically elucidates some details of the detector 209. It should still be pointed out at this juncture that by virtue of the raster scanning / scanning movement of the primary particle beams 3 over an object 7 or a sample 7, many points of the sample 7 are irradiated or scanned. In this case, each primary particle beam 3 sweeps wholly or partly over an individual field region of the object 7. In this case, each primary particle beam 3 is allocated a dedicated individual field region of the object 7. From these individual field regions of the object, interaction products, for example secondary electrons, then in turn emanate from the object 7. The interaction products are then projected onto the detection regions of the particle detector or onto the scintillator plate 207 in such a way that the interaction products emanating from two different individual field regions are projected onto different detection regions of the scintillator plate 207. Light signals are emitted by each detection region of the scintillator plate 207 upon incidence of the interaction products, e.g. secondary electrons, on said detection region, wherein the light signals emitted by each detection region are fed to a light detector 241 assigned to the respective detection region. In other words, the situation is that each primary particle beam 3 comprises its own detection region on the scintillator 207 and also its own light detector 241. Therefore, in the case of the multi-beam particle microscope 1 described, all characteristic curve deviations of the detection regions or detectors are visible in the individual images generated in each case. That means that the brightness and / or the contrast in each individual image may vary owing to the deviations described and that the combined full image on the basis of all of the primary particle beams has an unsatisfactory appearance until the different detection regions or detectors are equalized with one another with regard to brightness and / or contrast, i.e. a predefined detector calibration criterion is satisfied. For example, such a criterion may be 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.

[0074] For example, the light detectors 241 , e.g. avalanche photodiodes, can be set when the method according to the invention is carried out. Detection architectures other than the one illustrated in Fig. 2 are also suitable for carrying out the methods according to the invention. For example, reference is made to the method of DED (“direct electron detection”), already described further above, which manages without light detectors and in which secondary electrons are directly converted into a current signal.

[0075] Fig. 3 schematically shows a multi-image which has a multiplicity of individual images and can be generated by means of the multi-beam particle microscope shown in Figures 1 and 2. In the example shown, the multi-image is composed of 91 individual images. The individual images are arranged as a tessellation in a hexagonal overall arrangement. In this case, each individual image is generated by means of an assigned individual particle beam 3 by rasterscanning over a sample surface and is thus assigned to a detection channel of the multi detector 209 or a multi-channel detection unit 209. In Fig. 3, the individual images are numbered consecutively. In this case, each individual image has been constructed line by line by means of a corresponding raster-scanning or scanning movement of the primary particle beam. The number of pixels in the individual image may be very high and is of the order of e.g. approx. 1000 x 1200 to approx. 8000 x 9300 pixels. The individual images schematically depicted in Fig. 3 have different brightness values and contrast values, which may be corrected by means of a detector calibration known per se. Then, for example, all individual images may have the same brightness value within a defined first limit and / or the same contrast value within a defined second limit. Detailed information in this respect can be gathered from WO 2020 / 057678 A1 , the disclosure of which is incorporated in this patent application in full by reference.

[0076] Fig. 4 schematically illustrates the relationship between a brightness histogram and a characteristic curve of a detection region or of a detection channel. For each individual image generated with a specific detector setting, an individual image analysis can be performed, and a brightness histogram can be ascertained. In this brightness histogram, the brightness, i.e. the greyscale value of the output signal, is plotted on the X-axis, and the number of pixels in the individual image that have the respective brightness is indicated on the Y-axis. A first curve K1 , which is depicted in Fig. 4a by means of the solid line, emerges for a first individual image. It comprises a fairly well delimited peak. A curve K2, which is depicted in Fig. 4a by means of the dashed curve K2, emerges for a second individual image. It comprises two peaks, with these peaks partly merging. The two curves K1 and K2 should be understood to be merely examples of brightness histograms. In any case, it is possible to evaluate the curves K1 , K2 or to ascertain a brightness of the individual image and a contrast of the individual image. In turn, there are different options for the definitions of brightness and contrast. For example, a brightness value may be defined as the measured mean value of the distribution or the mean value of a Gaussian distribution fitted to a curve. The latter would be possible for histograms that more closely resemble the type of curve K1. A contrast C1 can then be defined as e.g. a multiple of the standard deviation of the Gaussian distribution, for example two sigma of the Gaussian distribution. This is indicated in curve K1 by the double-headed arrow. A contrast may also be defined for histograms that rather resemble the type of curve K2, for example as distance between the two peaks, which in turn is shown in Fig. 4a as double-headed arrow 02. There are numerous concrete options for meaningful definitions. It is self-evident that any definition actually used is then equally binding to all individual images.

[0077] Fig. 4b illustrates a detector characteristic curve by way of example. The obtained output signal of a detector is plotted against the beam intensity. For illustrative purposes, the axis scalings are normalized to one in each case. It is evident, in principle, that the output signal becomes stronger when the beam intensity increases. In this case, there is at least in sections a linear relationship between the beam intensity and the output signal. In sections, therefore, a straight line can be fitted, i.e. a tangent can be applied, to the characteristic curve shown, wherein the slope p of the tangent describes the gain. The characteristic curve in turn intersects the y axis at a level d. Even without an incoming beam or without photons incident on the detector (depending on the detector), an, albeit weak, output signal is thus obtained. This so-called offset d must therefore be eliminated, where necessary, from values for the output signal. The gain and the offset d thus vary when the characteristic curve overall varies. Both the gain p and the offset d can thus be set by way of a variation of the characteristic curve.

[0078] Different detector settings, in turn, result in different brightness histograms. Fig. 4a reveals that a pure detector offset only affects the brightness value. In other words, the curves K1 , K2 in the diagram are displaced towards the left or right when the offset of a detector of a detection channel, for example the offset of an avalanche photodiode, is modified. By contrast, if the gain of the detector characteristic curve is modified, then two values change in combination with each other, namely both the brightness value and the width of the distribution, specifically for example the standard deviation (two sigma width) of a Gaussian distribution (cf. curve K1).

[0079] Figure 5 schematically shows effects of a modified detector offset, for instance an offset of an avalanche photodiode, on a brightness histogram. Figure 5a depicts a first brightness histogram at a first offset D1. The curve K1 is relatively far to the right in the brightness histogram. In the example shown, the brightness value of the brightness histogram is defined by way of the position Pav, which is located between the positions of the two peaks P1 and P2. In the example shown in Figure 5a, Pavis approximately 217. In Figure 5b, the only adjustment to the detection unit, for example an avalanche photodiode, was the adjustment of the offset to the value D2. As a result, the shape of curve K1 per se does not change in the brightness histogram, but there is a change in the position of said curve: In the example shown in Figure 5b, the curve K1 is displaced to the left as a whole. This also changes the brightness value Pavto approximately 130.

[0080] In Figure 5c, the offset is set to a third value D3. As a result, the position of curve K1 has been displaced even further to the left: The brightness value Pavis now approximately 91. A change in the offset d thus displaces the mean brightness value Pavin the brightness histogram.

[0081] By contrast, Figure 6 schematically shows effects of a modified gain p on a brightness histogram: Figure 6a once again shows the initial situation according to Figure 5a. The gain set on the detection unit or the detection channel is pi . Figure 6b schematically shows the effect that a modification of the gain to the value 2 might entail. The other parameters of the detection unit, in particular the offset, have been left unchanged. It is evident from Figure 6b that the contrast value of the brightness histogram has been modified as a result of the change in the gain: The contrast C2 is lower than the contrast C1 according to Figure 6a. In addition to the change in contrast, the brightness value of the brightness histogram has also been modified: The value Pavhas decreased. Figure 6c shows an example of a gain p3, wherein the other setting parameters of the detection unit have once again been left unchanged. In particular, the offset was not modified. According to Figure 6c, the curve K1c is significantly wider than in the examples in Figures 6a and 6b. The contrast has increased again and is C3. The mean brightness value Pavhas been modified again and now is approximately in the centre of the brightness histogram.

[0082] It should be noted that Figure 6 is only a schematic illustration, and moreover the depicted effect for a modified gain has been exaggerated in order to explain the effects of the change in gain at a detection unit.

[0083] Thus, a brightness histogram can be modified in a targeted manner by changing an offset and / or gain. A brightness histogram may also be set in such a way that it meets specific requirements. Figure 7 schematically shows criteria for a detection quality criterion: Figure 7a depicts a first interval 11 of the brightness histogram, which represents very low brightness. In the example shown, the value 0 is the lower limit of the interval 11. Furthermore, Figure 7a shows a second interval I2 of the brightness histogram, which represents very high brightness. In the example shown, the maximum brightness value of 255 is part of the interval I2. What holds true in the example shown is that values of curve K1 are located in both intervals 11 and 12, but these values fall sharply to the value of zero towards the outer edges in intervals 11 and I2. On account of this sharp decrease towards the edges of the brightness histogram in the intervals 11 and I2, the assumption can be made that information to be obtained is completely represented in the brightness histogram, but no essential information about an individual image is lost at the edges of the brightness histogram. Therefore, a minimum quality requirement to be fulfilled can be defined in such a way that a maximum of 0.20%, in particular a maximum of 0.15% or a maximum of 0.10% of all brightness values are located within the first interval 11. In addition to that or in an alternative, it may 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 be located within the second interval I2. In addition to considering the two aforementioned criteria for the intervals 11 and I2 separately or in an alternative, a common requirement for the minimum quality of a detection channel may be considered to be met if a maximum of 0.20%, preferably a maximum of 0.15% or most preferably a maximum of 0.10% of all brightness values taken together are located within the first and second interval 11 , I2. This requirement is slightly more stringent.

[0084] The position of the intervals 11 and I2 can also be defined somewhat differently, as is illustrated by way of example in Figure 7b: In the example illustrated in Figure 7b, the greyscale value 0 with lowest brightness does not belong to interval 11 , and the value of 255 with the maximum brightness does not belong to interval I2. In the case of such a definition of intervals 11 and I2, it is possible to additionally define separate minimum quality requirements. For example, the frequency of the maximum detectable brightness (here: value at 255) and the value of the minimum detectable brightness (here: value at 0) may serve as a criterion. For example, it may be required that the 0 value is not occupied or may only be for 0.001 % of all brightness values. The same applies to the frequency of the highest detectable brightness with the value 255. This in turn makes it possible to ensure that no information that is important is lost.

[0085] A further criterion for defining a detection quality criterion is the overall width of the frequency distribution with frequencies 0. For example, it may be required that curve K1 may only drop to zero within each of the edge regions or intervals 11 and I2. A contrast value C thus is relatively large according to the definitions explained in Figure 6 by way of example.

[0086] Figure 8 schematically illustrates an adjustment setting of a detection channel for fulfilling a detection quality criterion for each individual image or for each individual particle beam. In the example shown, only the offset d for setting the detection quality criterion is modified for the detection channel, for example an avalanche photodiode: While the peak P2 of curve K1 is still located within the second interval I2 near the edge of the brightness histogram according to Figure 8a, this is no longer the case with the modified setting according to Figure 8b. Significantly fewer pixels are in interval 12 with the offset d2. However, there are no pixels at all in the interval 11 in Figure 8b. Therefore, the curve K1 overall is still too narrow for an optimal setting. This can be changed by changing the gain B of the detection unit; a corresponding result with a gain B2 is depicted in Figure 8c. Both the interval 11 and the interval I2 contain pixels of curve K1 , but the number of all pixels in the intervals 11 and I2 is limited or below a predefined limit.

[0087] Figure 9 schematically illustrates a method according to the invention for setting a multichannel detection unit 209 of a multi-beam particle microscope 1 during an inspection of charging samples 7. In an initial method step SO, the multi-beam particle microscope 1 that operates with a multiplicity of individual charged particle beams 3 is provided. In this case, each detection channel of the multi-channel detection unit 209 is assigned one of the individual charged particle beams 3 for image generation. For example, the multi-beam particle microscope might be the multi-beam particle microscope 1 illustrated in Figure 1. For example, the multi-channel detection unit 209 may be constructed as illustrated in Figure 2. For example, each channel of the multi-channel detection unit 209 may comprise an avalanche photodiode.

[0088] Method step S1 implements the provision of a multiplicity of individual reference settings Ri, wherein each of the reference settings Ri describes the reference setting Ri for an associated detection channel i. When raster-scanning a reference sample, a multiplicity of individual images that meet a predefined detector criterion can be generated with this multiplicity of reference settings Ri. For example, it is possible that, according to the detector calibration criterion, all individual images have the same brightness value within a defined first limit and / or the same contrast value within a defined second limit. This method step S1 relates to the provision of the reference settings Ri; it is not necessary for these to be ascertained during the method. Instead, it is advantageous to have the individual reference settings Ri stored or available in a memory of the multi-beam particle microscope 1.

[0089] In method step S2, the reference settings Ri for the multi-channel detection unit 209 are detuned. In the process, each of the multiplicity of individual reference settings Ri per detection channel is modified by an individual detuning Ai. Thus Ri_detune = Ri + Ai applies. This individual detuning Ai is generated on an individual basis for each detection channel i and is preferably chosen such that the individual detuning Ai of the detection channels in each case differ from each other in pairwise fashion. Furthermore, the individual detuning Ai for each channel should be understood to mean that at least one parameter of the detection channel i is changed for this individual detuning Ai. However, multiple parameters of a detection channel might also be changed or detuned. These parameters for example are an offset and / or a gain of a detection channel or a detection unit arranged therein, for instance an avalanche photodiode.

[0090] In method step S3, there is a move to a first inspection site of a charging sample, for example a wafer.

[0091] In method step S4, there is a raster-type scan of this first inspection site using the multiplicity of individual charged particle beams 3. In the process, the detuned settings Ri_detune of the multi-channel detection unit 209 are used to generate a multiplicity of individual particle-optical images. Thus, an ensemble measurement is performed in this method step using a multiplicity of detection units or detection channels i that are detuned from one another and set differently.

[0092] In method step S5, there is the evaluation of the multiplicity of individual images with respect to a predefined detection quality criterion. Evaluating the multiplicity of individual images may comprise the generation of a brightness histogram for each individual image. An example in this respect is illustrated in Figure 10: The brightness histograms in Figure 10 are different for each detection channel or for each particle-optical image recorded therewith.

[0093] In method step S6, a best individual image is determined from the multiplicity of individual images on the basis of the detection quality criterion. Moreover, the detuning Ai_best, which is associated with the best individual image, is also determined. This is illustrated by way of example in Figure 11 : The curve Ki that best meets the detection quality criterion represented on the basis of Figure 7b is the curve K3. Firstly, the curve K3 has the required minimum extent or histogram width (and hence a high contrast), and, secondly, the number of pixels with the respective brightness drops sharply within the intervals 11 and I2 in the edge regions of the brightness histogram, and so, overall, only a small number of pixels are located in the intervals 11 and I2. None of the remaining curves K1 , K2, K4 and K5 meet this criterion to this extent. By ascertaining a best individual image or a best histogram curve Kbest, the detuning Ai that led to the generation of the best individual image or the generation of the best brightness histogram is now also known. For example, the change in an offset and / or a gain, for example in an avalanche photodiode, that has led to the best individual image or the best brightness histogram of the individual image is now known.

[0094] In a method step S7, a channel-independent global detuning A_global is ascertained on the basis of the detuning Ai_best. In the process, it is possible that the global detuning A_global is identical to the detuning Ai_best associated with the best individual image. However, it is also possible that the global detuning A_global is ascertained by interpolation. In this alternative, it is typically not only the best detuning Ai_best and the best individual image that has been ascertained, but also the second-best individual image with the second-best detuning Ai_best2.

[0095] In method step S8, the global detuning A_global is now applied to the reference settings Ri for all detection channels i, whereby optimal settings Ri_opt of the multi-channel detection unit 209 are set. What may hold true in this method step is that, for example, an offset and / or gain ascertained as the best detuning Ai_best are / is transferred to all detection channels i. For example, it may be the case that the reference settings Ri are modified in accordance with the best offset change and / or the best gain change. In this context, it is also possible to make use of the characteristic curves, known per se, of the detection channels or detection units, since the parameters of offset and gain of a detection channel are linked to each other by means of the characteristic curve. It is a basic assumption here that the sample to be inspected is essentially homogeneous.

[0096] In an optional method step S9, there now is a move to a second inspection site of the charging sample, wherein the second inspection site differs from the first inspection site. In method step S10, there is raster-scanning of the second inspection site using the multiplicity of individual charged particle beams 3, wherein a multiplicity of individual particle-optical images are generated with the optimal settings Ri_opt of the multi-channel detection unit.

[0097] Method steps S9 and S10 may be repeated with the same optional settings Ri_opt for a third inspection site or a further inspection site of the charging sample, with the third or further inspection site in turn being different inspection sites to those that have already been approached. Attention is again drawn to the fact that a charging sample 7 is inspected, with an inspection result depending on the illumination history of the inspection site. For this reason, each inspection site is only approached or illuminated once.

[0098] Figure 12 schematically shows brightness histograms of individual images that have been generated with the optimal settings Ri_opt of the multi-channel detection unit 209: The brightness histogram of each curve K1 to K5 now meets the detection quality criterion, for example the detection quality criterion described in more detail on the basis of Figure 7. All curves K1 to K5 fall sharply or fall down to the value zero in the edge intervals 11 and I2, and only a few pixels are located in the intervals 11 and I2. In this respect, the second inspection site of the charging sample 7 may be scanned with optimal detection settings for all channels i of the multi-channel detection unit 209. A corresponding statement applies to all subsequent inspection sites on the charging sample 7.

[0099] The description of the figures should not be construed as limiting the present invention. Instead, the figures merely facilitate a better understanding of the invention.

[0100] The invention relates to a method for setting a multi-channel detection unit of a multi-beam particle microscope during an inspection of charging samples. Time-consuming and spaceconsuming serial recordings by the multi-beam particle microscope that serve the purpose of setting a multi-channel detection unit during an inspection of charging samples are avoided and replaced with a skilfully chosen ensemble recording or ensemble measurement. The multibeam property of the multi-beam particle microscope is used in a targeted manner for detection setting purposes.

[0101] List of reference signs

[0102] 1 Multiple particle beam system, multi-beam particle microscope

[0103] 3 Primary particle beams, first individual particle beams

[0104] 5 Beam spots, incidence locations

[0105] 7 Object, sample, wafer

[0106] 9 Secondary particle beams, second individual particle beams

[0107] 10 Computer system, controller

[0108] 15 Sample surface, wafer surface

[0109] 25 Image point of a second individual particle beam

[0110] 101 Object plane

[0111] 102 Objective lens

[0112] 103 Field lens

[0113] 105 Axis

[0114] 108 Beam crossover

[0115] 200 Detector system

[0116] 205 Projection lens system

[0117] 206 Projection lens

[0118] 207 Multi-particle detector

[0119] 208 Projection lens

[0120] 210 Projection lens

[0121] 212 Beam crossover

[0122] 214 Aperture filter, contrast stop

[0123] 221 Beam path

[0124] 222 Collective anti-deflection system

[0125] 223 Light-optical unit

[0126] 225 Lens

[0127] 227 Mirror

[0128] 229 Lens

[0129] 231 Lens

[0130] 235 Light-receiving surface

[0131] 237 Light-detection system

[0132] 239 Optical fibre

[0133] 241 Light detector

[0134] 243 Region

[0135] 245 Signal line

[0136] 270 Control computer system 280 Image generation computers

[0137] 300 Beam-generating apparatus

[0138] 301 Particle source

[0139] 303 Collimation lens system

[0140] 304 Multi-aperture array, filter plate

[0141] 305 Micro-optical unit, multi-aperture arrangement, multi-beam particle generator

[0142] 306 Multi-aperture plate, multi-aperture array

[0143] 307 Field lens, aperture plate

[0144] 308 Field lens

[0145] 309 Particle beam

[0146] 321 Intermediate image plane

[0147] 323 Beam foci

[0148] 400 Beam splitter, magnet arrangement

[0149] 500 Scan deflector

[0150] 600 Displacement stage or positioning device x Direction y Direction z Direction

[0151] K Curve of a brightness histogram d Offset

[0152] P Gain

[0153] P1 Peak

[0154] P2 Peak

[0155] PavBrightness value

[0156] C Contrast

[0157] 11 Interval

[0158] 12 Interval

Claims

1. Claims1. Method for setting a multi-channel detection unit of a multi-beam particle microscope during an inspection of charging samples, wherein the multi-beam particle microscope operates with a multiplicity of individual charged particle beams, wherein each detection channel of the multi-channel detection unit is assigned one of the individual charged particle beams for image generation purposes and wherein the method comprises the following steps: providing the multi-channel detection unit with a multiplicity of individual reference settings Ri, wherein each of the reference settings Ri describes the reference setting Ri for an associated detection channel, wherein a multiplicity of individual images that meet a predefined detector calibration criterion can be generated using this multiplicity of reference settings Ri during a raster-type scan of a reference sample; detuning the reference settings Ri for the multi-channel detection unit, wherein the multiplicity of individual reference settings Ri per detection channel are each modified by an individual detuning Ai , i.e. Ri_detune = Ri + Ai; moving to a first inspection site of a charging sample; raster-scanning the first inspection site using the multiplicity of individual charged particle beams, wherein a multiplicity of individual particle-optical images are generated with the detuned settings Ri_detune of the multi-channel detection unit; evaluating the multiplicity of individual images with respect to a predefined detection quality criterion; determining a best individual image from the multiplicity of individual images on the basis of the detection quality criterion and determining the detuning Ai_best associated with the best individual image; ascertaining a channel-independent global detuning A_global on the basis of the detuning Ai_best; and applying the global detuning A_global to the reference settings Ri for all detection channels i, whereby optimal settings Ri_opt of the multi-channel detection unit are set.

2. Method according to Claim 1, furthermore including the following steps: moving to a second inspection site of the charging sample, wherein the second inspection site differs from the first inspection site; and raster-scanning the second inspection site using the multiplicity of individual charged particle beams, wherein a multiplicity of individual particle-optical images are generated with the optimal settings Ri_opt of the multi-channel detection unit.

3. Method according to either of the preceding claims,30 wherein the global detuning A_global is identical to the detuning Ai_best associated with the best individual image.

4. Method according to either of Claims 1 and 2, wherein the global detuning A_global is ascertained by interpolation or extrapolation.

5. Method according to any of the preceding claims, wherein, according to the detector calibration criterion, all individual images have the same brightness value within a defined first limit and / or the same contrast value within a defined second limit.

6. Method according to any of the preceding claims, wherein each reference setting Ri for a detection channel of the multi-channel detection unit is defined by exactly one parameter; and wherein the individual detuning Ai for each detection channel comprises detuning of the exactly one parameter.

7. Method according to any of Claims 1 to 5, wherein each reference setting Ri for a detection channel of the multi-channel detection unit is defined by at least two parameters and wherein the individual detuning Ai for each detection channel comprises detuning of the at least two parameters.

8. Method according to any of the preceding claims, wherein each reference setting Ri for a detection channel and / or each individual detuning Ai are / is defined by values and / or changes in value of at least one of the following parameters: gain, offset.

9. Method according to any of the preceding claims, wherein the individual detunings Ai of the detection channels in each case differ from each other in pairwise fashion.

10. Method according to any of the preceding claims, wherein the individual detunings Ai with respect to at least one parameter are each generated by incremental changes of Ai by a constant value k.

11. Method according to any of the preceding claims,wherein the evaluation of the multiplicity of individual images comprises a generation of brightness histograms of the individual images.

12. Method according to the preceding claim, wherein the detection quality criterion is defined on the basis of at least one of the following criteria: frequencies of brightness levels in a first interval of the brightness histogram, which represents very low levels of brightness; frequencies of brightness levels in a second interval of the brightness histogram, which represents very high levels of brightness; overall width of the frequency distribution with non-zero frequencies; frequency of maximum detectable brightness; frequency of minimum detectable brightness.

13. Method according to the preceding claim, wherein, according to the detection quality criterion, an individual image is assessed as being better the less frequently brightness levels are counted in the first interval; and / or wherein, according to the detection quality criterion, an individual image is assessed as being better the less frequently brightness levels are counted in the second interval.

14. Method according to either of Claims 12 and 13, wherein a detuned setting Ri_detune of a detection channel meets a minimum quality requirement should a maximum of 0.20%, in particular a maximum of 0.15% or a maximum of 0.10%, of all brightness values be located within the first interval; and / or wherein a detuned setting Ri_detune of a detection channel meets a minimum quality requirement should a maximum of 0.20%, in particular a maximum of 0.15% or a maximum of 0.10% of all brightness values be located within the second interval.

15. Method according to any of Claims 12 to 14, wherein a detuned setting Ri_detune of a detection channel meets a minimum quality requirement should a maximum of 0.20%, in particular a maximum of 0.15% or a maximum of 0.10% of all brightness values be located within the first and the second interval.

16. Method according to any of the preceding claims, wherein the multi-channel detection unit comprises a combination of a particle detector and a light-detection unit.

17. Method according to the preceding claim, wherein the light-detection unit comprises a multiplicity of avalanche photodiodes and wherein each detection channel is in each case assigned an avalanche photodiode, and wherein a gain and / or an offset of an avalanche photodiode are / is modified, respectively, when the reference settings Ri for the multi-channel detection unit are detuned.

18. Computer program product having a program code for carrying out the method according to any of the preceding claims.

19. Multi-beam particle microscope configured to carry out the method according to any of Claims 1 to 17.

Images