Method for determining the position of a manipulator tip

Abstract

The subject of the invention is a method for finding the position of the tip of a manipulator, consisting in obtaining at least three different images of a sample, where either the table or the manipulator is moved between obtaining the individual images and at least one image is in the field of view with the tip of the manipulator, or consisting in obtaining at least two different images of a sample, where either the table or the manipulator is moved between obtaining the individual images or the manipulator is moved between the acquisition of individual images, and at least one image has the tip of the manipulator in the field of view, and at the same time, difference images are obtained by subtracting each pair of acquired sample images.

Description

[0001] Method for determining the position of a manipulator tip

[0002] Field of Art

[0003] The invention relates to a method for automatically finding (determining) the position of a manipulator tip, in particular in charged particle beam devices.

[0004] Background Art

[0005] To examine the internal structure of samples, for example, using a transmission electron microscope, it is necessary to have a very thin sample, a so-called lamella, which is a part of the sample containing the area of interest. Such lamellas are usually prepared using a focused ion beam (FIB) device, usually in combination with a scanning electron microscope (SEM) for observing the process of preparing the lamella. The device may also contain a device generating a photon beam. The process usually takes place by inserting a sample containing the area of interest into the device. The said area of interest is found using the SEM and then a lamella is prepared from the sample using the FIB or using the photon beam. The thus created lamella is transferred using a manipulator to a lamella holder (so-called grid), where it can be further processed, for example, using a charged particle beam.

[0006] A manipulator is a movable device located in the chamber of an electron microscope, which is generally terminated by a tip. The tip can be a needle forming a separate part detachably connected to the rest of the manipulator, or it can be the manipulator itself, which has a tip-shaped end. A lamella can be attached to the manipulator by various methods, e.g. by means of material deposition. In this method, various types of gases containing the desired material are delivered to the lamella, which is still connected to the sample, using one or more nozzles, and the desired material is deposited by treatment using a charged particle beam. The deposit of the material (e.g. a pure element or a combination thereof) then forms a solid connection between the lamella and the manipulator tip. Subsequently, the lamella is separated from the sample, e.g. by FIB, and transferred by the manipulator to a grid, to which it is again attached, e.g. by FIB and material deposition. The manipulator tip is then separated from the lamella by means of FIB. This method is described e.g. in US9040908B2.

[0007] The process of preparing the lamella is time-consuming and technically demanding and requires the constant presence of a qualified professional who must monitor and control the process. Efforts have been made to develop a method for preparing the lamella that could be performed automatically, reliably and at the same time efficiently by the device, without the need for a permanent presence of the qualified professional.

[0008] One of the necessary steps towards such automated procedure is to automate the finding (determining of the position) of the manipulatortip in an image, so that the manipulatortip can be accurately navigated to the lamella. In the prior art, there are methods using two images, as described, for example, in US10825651 B2. This document describes acquiring a first image of the manipulator tip, then moving the manipulator to another position and acquiring a second image of the manipulator tip. These images are compared and based on the differences between them, the position of the manipulator tip is found. However, this solution has the disadvantage of being unreliable when the background in the image, for example the surface of the sample, has a pattern that is similar to the shape of the manipulator. Such a pattern can be, for example, straight lines formed by, for example, layers of a semiconductor chip.

[0009] Another option for automating the finding (determining the position) of the manipulator tip is to use a mark created on the manipulator tip, as also described in the aforementioned patent US10825651 B2. However, this solution is complicated, since it requires creating the aforementioned mark. Another option described in the same patent is to use a machine-recognizable shape of the manipulator. However, this solution is also unsuitable, as it does not solve the problem in the case of a sample with a surface similar to the shape of the manipulator, and at the same time, the shape of the manipulator may also change due to the deposition of, for example, sputtered material or residues after cutting the lamella from the tip of the manipulator.

[0010] It would therefore be desirable to provide a robust and reliable method for automatic determination of the position of the manipulator even if, for example, its shape is similar to the surface of the sample, without the need to adapt the manipulator in any way.

[0011] Summary of the Invention

[0012] The above problems are solved by a first aspect of the invention which is a method for determining the position of a manipulator tip using a charged particle beam device comprising at least one source of charged particles, at least one detector of signal particles, a movable manipulator, a movable stage adapted for receiving the sample, a sample placed on the movable stage and an evaluation unit for performing the method. The method for finding the manipulator tip comprises the steps of acquiring at least three different images of the sample by irradiating the sample with a charged particle beam generated by a source of charged particles and detecting signal particles with the detector of signal particles, wherein between acquiring individual images ofthe sample, eitherthe stage orthe manipulator is moved, and wherein at least one image of the sample shows the manipulator tip in the field of view, wherein the acquired images of the sample are subsequently evaluated and the position of the manipulator tip is found in the acquired images of the sample which shows the manipulator tip in the field of view. The method achieves the above objective by using at least three different images of the sample for determining the position of the manipulator tip. The use of three or more images increases the probability of correctly finding the position of the manipulator tip. In a preferred embodiment, the evaluation of the acquired images of the sample is performed using a neural network, the input of which is the at least three images of the sample and the output is information enabling the position of the manipulator tip to be determined, such as coordinates of the manipulator tip on the image of the sample which shows the manipulator tip in the field of view, or a binary image allowing to determine the said coordinates. The advantage of using a neural network for evaluating the images of the sample lies in the simplicity and robustness of such an evaluation method compared to methods based on classical mathematical methods, which are not suitable for evaluating more than two images.

[0013] The above described problems are also solved by a second aspect of the present invention which is a method for determining the position of the manipulator tip using a charged particle beam device comprising at least one charged particle source, at least one detector of signal particles, a movable manipulator, a movable stage adapted for receiving the sample, a sample placed on the movable stage and an evaluation unit for performing the method. The method for determining the position of the manipulator tip involves acquiring at least two different images of the sample by irradiating the sample with charged particle beam produced by a source of charged particles and detecting signal particles using the detector of signal particles, wherein between acquiring individual images of the sample, either the stage orthe manipulator is moved and at the same time at least one image of the sample shows the manipulator tip in the field of view, wherein the acquired images of the sample in each pair of images are subsequently subtracted from each other, wherein each pair of images has either the stage or the manipulator at the same position in both images of the sample and these are subtracted from each other, wherein difference images are acquired, wherein the acquired images of the sample and the difference images are subsequently evaluated and the position of the manipulator tip is determined in the acquired image of the sample which shows the manipulator tip in the field of view. The method achieves the above-mentioned objective by using at least two different images of the sample and at least one difference image to determine the position of the manipulator tip. Using three or more images increases the probability of correctly finding the position of the manipulator tip.

[0014] In a preferred embodiment, the evaluation of the acquired images of the sample and the difference images is performed using a neural network, the input of which is the at least two image of a samples and the at least one difference image and the output is information enabling the position of the manipulator tip to be determined, such as coordinates of the manipulator tip on the image of the sample which shows the manipulator tip in the field of view, or a binary image allowing to determine the said coordinates. The advantage of using a neural network for evaluating the images of the sample lies in the simplicity and robustness of such an evaluation method compared to methods based on classical mathematical methods, which are not suitable for evaluating more than two images.

[0015] Signal particles are in particular secondary or back-reflected particles, or other particles emitted by the sample due to its interaction with the incident charged particle beam or particles reflected by the sample. Brief description of the drawings

[0016] The invention is further explained by way of examples, which are described with reference to the attached drawings. For better clarity, only those parts of the device that are important with respect to the principle of the present invention are shown in the drawings.

[0017] Fig. 1 is a schematic illustration of a charged particle beam device;

[0018] Fig. 2 is a first acquired image of a sample according to a first exemplary embodiment;

[0019] Fig. 3 is a second acquired image of a sample according to a first exemplary embodiment;

[0020] Fig. 4 is a third acquired image of a sample according to a first exemplary embodiment;

[0021] Fig. 5 is a first acquired image of a sample according to a second exemplary embodiment;

[0022] Fig. 6 is a second acquired image of a sample according to a second exemplary embodiment;

[0023] Fig. 7 is a third acquired image of a sample according to a second exemplary embodiment;

[0024] Fig. 8 is a binary image corresponding to the first acquired image of a sample according to a second exemplary embodiment;

[0025] Fig. 9 is a binary image corresponding to a second acquired image of a sample according to a second exemplary embodiment;

[0026] FIG. 10 is a binary image corresponding to a third acquired image of a sample according to a second exemplary embodiment;

[0027] FIG. 11 is an output binary image according to a second exemplary embodiment;

[0028] FIG. 12 is a first acquired image of a sample according to a third exemplary embodiment;

[0029] FIG. 13 is a second acquired image of a sample according to a third exemplary embodiment;

[0030] FIG. 14 is a difference image according to a third exemplary embodiment;

[0031] FIG. 15 is a first acquired image of a sample according to a fourth exemplary embodiment;

[0032] FIG. 16 is a second acquired image of a sample according to a fourth exemplary embodiment;

[0033] FIG. 17 is a third acquired image of a sample according to a fourth exemplary embodiment;

[0034] FIG. 18 is a first difference image according to a fourth exemplary embodiment;

[0035] FIG. 19 is a second difference image according to a fourth exemplary embodiment;

[0036] FIG. 20 is a binary image corresponding to a third acquired image of a sample according to a fourth exemplary embodiment;

[0037] FIG. 21 is a first acquired image of a sample according to a fifth exemplary embodiment;

[0038] FIG. 22 is a second acquired image of a sample according to a fifth exemplary embodiment;

[0039] FIG. 23 is a third acquired image of a sample according to a fifth exemplary embodiment;

[0040] FIG. 24 is a fourth acquired image of a sample according to a fifth exemplary embodiment;

[0041] FIG. 25 is a first difference image according to a fifth exemplary embodiment;

[0042] FIG. 26 is a second difference image according to a fifth exemplary embodiment;

[0043] FIG. 27 is a third difference image according to a fifth exemplary embodiment;

[0044] FIG. 28 is a binary image corresponding to a second acquired image of a sample according to a fifth exemplary embodiment;

[0045] FIG. 29 is a binary image corresponding to a third acquired image of a sample according to a fifth exemplary embodiment; FIG. 30 is a binary image corresponding to a fourth acquired image of a sample according to a fifth exemplary embodiment;

[0046] FIG. 31 is an output binary image according to a fifth exemplary embodiment;

[0047] Detailed description of the Invention

[0048] The below embodiments illustrate exemplary variants of the invention, which, however, should not be construed as limiting the scope of protection.

[0049] The method of determining the position of the tip of the manipulator 6 is performed using a charged particle beam device shown in Fig. 1 . The particles are generated using at least one source 1_ of charged particles. The source of charged particles is an electron source or an ion source.

[0050] The device further comprises at least one column 2 and a chamber 3. The column 2 is connected to the chamber s. The source 1. of charged particles is located in the column 2. A set 7_of elements for shaping and directing particles is further located in the column 2. The elements for shaping and directing particles are, for example, electromagnetic lenses, stigmators, electrodes on potential, apertures and other commonly used components for particle optics. In the case where the device comprises two columns connected to the chamber 3, these are connected so that their optical axes form an angle greater than 0° and less than 180°. An example of such a device is a scanning electron microscope with a second column producing a focused ion beam. In an exemplary embodiment of a device with a charged particle beam, a device generating a photon beam is further connected to the chamber 3.

[0051] The source 1. of charged particles produces a beam which irradiates its field of view using a set 7 of elements for forming and directing particles.

[0052] The particle beam device further comprises at least one detector 8 of signal particles. The signal particles are in particular secondary or back-reflected particles, or other particles emitted by the sample 5 due to its interaction with the incident charged particle beam or particles reflected by the sample 5. The detector 8 of signal particles can be located in the chamber 3_or in the column 2.

[0053] The chamber 3 further comprises a movable stage 4 for receiving at least one sample. The stage 4 is adapted for movement along at least two mutually perpendicular axes. In an exemplary embodiment of the stage 4, the stage 4 is adapted for movement along three mutually perpendicular axes. In another exemplary embodiment of the stage 4, the stage 4 is further adapted for rotation around at least one axis and for tilting around at least one axis. The exemplary embodiments of the stage 4 can be freely combined. The stage 4 is adapted for receiving at least one sample 5.

[0054] The chamber s further comprises a movable manipulator s. The end ofthe manipulator 6_has the shape of a tip. In an alternative embodiment of the manipulator s, the end of the manipulator s is formed by a needle detachably connected to the manipulator 6 and forming the tip of the manipulator 6. In an exemplary embodiment of the possibilities of movement of the manipulator 6, the manipulator 6 is movable along three axes, which are mutually perpendicular in the exemplary embodiment. In another exemplary embodiment of the possibilities of movement of the manipulator, the manipulator 6 further allows rotation around at least one axis.

[0055] The device further comprises an evaluation unit containing software for performing the method.

[0056] The method for finding the tip of the manipulator s includes the step of acquiring different images of the sample 5 by irradiating the sample 5 with a charged particle beam generated by a source 1 of charged particles and detecting signal particles. Between the acquisition of individual images of the sample 5, either the stage 4 or the manipulator s is always moved. Thus, a pair of acquired images always has either the stage 4 or the manipulator 6 in the same position. In other words, for each acquired image of the sample 5, there must be another image of the sample 5, which will have either the stage 4 or the manipulator 6 in the same position. At the same time, at least one of the acquired images of the sample 5 has the tip of the manipulator 6 in the field of view.

[0057] In an exemplary embodiment of the step of acquiring images of the sample 5, at least three different images of the sample 5 are acquired.

[0058] In another exemplary embodiment of the step of acquiring images of the sample 5, at least two different images of the sample 5 are acquired.

[0059] In the next step of the method for finding the tip of the manipulator s, the acquired images of the sample 5_are evaluated using an evaluation unit, and the position of the tip of the manipulator 6_in one of the acquired images of the sample 5 with the tip of the manipulator 6 in the field of view is found, preferebly in the last acquired images of the sample 5 with the tip of the manipulator s in the field of view. Part of the evaluation process is the use of a neural network. The neural network can be any neural network that is adapted for image segmentation. In an exemplary embodiment of the neural network, the neural network is a two-dimensional convolutional neural network with a U-Net architecture. The neural network is trained using annotated simulated or real data on which various samples and manipulators are visible. The neural network is trained to recognize the shape of a needle in the image.

[0060] In an exemplary embodiment of the neural network inputs, the neural network input is the at least three acquired images of the sample 5.

[0061] In another exemplary embodiment of the neural network inputs, the neural network input is the at least two acquired images of the sample 5 as well as at least one difference image. The difference image is created by subtracting two images of the sample 5 having either the stage 4 or the manipulator 6 at the same position in both images. When subtracting two images of the sample 5 having a certain object at the same location, this object should be removed from the resulting difference image, however, due to, for example, the temperature drift of the sample 5, the change in the electric field acting on the particle beam caused by the change in the position of the manipulator 6, the noise of the signal particles or generally the noise of the detector 8 of the signal particles, the resulting difference image may also contain artifacts of the object that should have been removed from the image because it was at the same location in both images subject to the subtracting.

[0062] The output ofthe neural network is information enabling the position of the tip of the manipulator 6 to be determined.

[0063] In an exemplary embodiment, the information enabling the position of the tip of the manipulator s to be determined is at least one binary image corresponding to the image of the sample 5 at the input of the neural network, wherein the manipulator 6 is distinguished from the background in the binary image. The binary image is not created for the image of the sample 5 which does not contai the manipulator s.

[0064] In another exemplary embodiment, the information enabling the position of the tip of the manipulator s to be determined is the coordinates of the tip of the manipulator s according to the last of the acquired images of the sample 5 having the manipulator tip in the field of view.

[0065] In the case where the information enabling the position of the tip of the manipulator 6 to be found is a binary image, the image is further processed using standard image processing techniques so that the coordinates of the tip of the manipulator 6_are found according to the last of the acquired images of the sample 5 having the tip of the manipulator s in the field of view. In one of the exemplary embodiments of image evaluation, in the case where the output of the neural network is an odd number of binary images and there are at least three binary images, binary image processing is further performed using the "majority voting" technique, where the images are compared at the pixel level and the result is again a binary image whose individual pixel values correspond to the majority values. In other words, if a pixel on the two input binary images has the value 0 and on one input binary image has the value 1 , then the value of this pixel in the output binary image will be 0. In the case where the binary image or images correspond to the acquired images of the sample 5 with the manipulator 6 in different places, it is necessary to shift this image or images so that the manipulator 6 is in the same place as in the case of the last acquired image of the sample 5, before processing using "majority voting" is performed, e.g. using image correlation or by shifting by the value of the displacement of manipulator s relative to the last acquired image of the sample 5.

[0066] In a first exemplary embodiment of the method for finding (determining the position) the tip of the manipulator 6, a charged particle beam device is used. First, images of the sample 5 are acquired by irradiating the sample 5 with an electron beam and detecting signal particles using the signal particle detector 8. A first acquired image of the sample 5 shown in Fig. 2 is an image of the sample 5, where the stage 4 is located in a first position of the stage 4 and the manipulator 6 is located in a first position of the manipulator 6 with the tip of the manipulator s located outside the field of view. Subsequently, the manipulator s is moved to a second position of the manipulator s with the tip of the manipulator s in the field of view and the stage 4 does not move (i.e., is kept in its first position) and a second image of the sample 5 is acquired, shown in Fig. 3. Subsequently, the stage 4_is moved in an axis perpendicular to the surface of the stage 4 on which the sample 5 is placed and the manipulator 6 does not move and a third image of the sample 5 is acquired, shown in Fig. 4. Subsequently, these acquired images of the sample 5 are used as inputs for the neural network, and the output from the neural network are the coordinates of the tip of the manipulator 6 according to the last, i.e. the third acquired image of the sample 5.

[0067] In a second exemplary embodiment of the method of finding the tip of the manipulator 6, a charged particle beam device is used. First, images of the sample 5 are acquired by irradiating the sample 5 with an electron beam and detecting signal particles using the signal particle detector 8. A first acquired image of the sample 5_shown in Fig. 5 is an image of the sample 5, where the stage 4 is located in a first position of the stage 4 and the manipulator 6_is located in a first position of the manipulator s with the tip of the manipulator s located in the field of view. Subsequently, the manipulator s is moved to a second position of the manipulator s located also in the field of view and the stage 4 is kept in its first position move and a second image of the sample 5 shown in Fig. 6 is acquired. 6. Subsequently, the stage 4 is moved in a plane parallel to the surface of the stage 4 on which the sample 5 is placed, and the manipulator 6 does not move, and a third image of the sample 5 is acquired, shown in Fig. 7. Subsequently, these acquired images of the sample 5 are used as inputs for the neural network, and the output from the neural network are three binary images shown in Fig. 8, 9 and 10 corresponding to the acquired images of the sample 5. In the first acquired image of the sample 5, the manipulator 6 is located in the first position of the manipulator s, in the remaining images, the manipulator 6_is located in the second position of the manipulator 6. Subsequently, the binary image corresponding to the first image of the sample 5 is shifted so that the position of the manipulator s corresponds to the second position of the manipulator s. Subsequently, the binary images are compared using the "majority voting" technique, the result of which is the output binary image shown in Fig. 1 1 , in which the tip of the manipulator 6 is subsequently found and its coordinates are determined.

[0068] In a third exemplary embodiment of the method forfinding the tip of the manipulators, a charged particle beam device is used. First, images of the sample 5 are acquired by irradiating the sample 5 with an electron beam and detecting signal particles using the signal particle detector 8. A first acquired image of the sample 5 shown in Fig. 12 is an image of the sample 5 where the stage 4_is located in a first position of the stage 4 and the manipulator s is located in a first position of the manipulator 6 with the tip of the manipulator s located in the field of view. Subsequently, the stage 4 with the sample 5 is moved in an axis perpendicular to the surface of the stage 4 on which the sample 5 is located, the manipulator 5 does not move and a second image of the sample 5 as shown in Fig. 13 is acquired. Subsequently, the first image of the sample 5 and the second image of the sample 5 are subtracted from each other to give a difference image shown in Fig. 14. Subsequently, the first image of the sample 5, the second image of the sample 5_and the difference image are used as inputs for the neural network and the output from the neural network is the coordinates of the tip of the manipulator 6_according to the last, i.e. the second acquired image of the sample 5.

[0069] In a fourth exemplary embodiment of the method of finding the tip of the manipulators, a charged particle beam device is used. First, images of the sample 5 are acquired by irradiating the sample 5 with an electron beam and detecting signal particles using the signal particle detector 8. A first acquired image of the sample 5 shown in Fig. 15 is an image of the sample 5 where the stage 4 is located in a first position of the stage 4 and the manipulator 6_is located in a first position of the manipulator 6 with the tip of the manipulator s located outside the field of view. Subsequently, the manipulator s is moved to a second position of the manipulator s located in the field of view and the stage 4 is kept in its first position and a second image of the sample 5 as shown in Fig. 16 is acquired. Subsequently, the stage 4 is moved in a plane parallel to the surface of the stage 4 on which the sample 5 is placed and the manipulator s is kept in its second position and a third image of the sample 5 is acquired, shown in Fig. 17. Subsequently, a first difference image shown in Fig. 18 is created by subtracting the first image of the sample 5 and the second image of the sample 5, and a second difference image shown in Fig. 19 is created by subtracting the second image of the sample 5 and the third image of the sample 5. Subsequently, the first image of the sample 5, the second image of the sample 5, the third image of the sample 5, the first difference image and the second difference image are used as inputs for the neural network and the output from the neural network is a binary image shown in Fig. 20 corresponding to the third image of the sample 5 in which the tip of the manipulator 6 is found according to the last, i.e. the third acquired image of the sample 5 and its coordinates are determined.

[0070] In a fifth exemplary embodiment of the method of finding the tip of the manipulator s, a charged particle beam device is used. First, images of the sample 5 are acquired by irradiating the sample 5 with an electron beam and detecting signal particles with the signal particle detector 8. a first acquired image of the sample 5 shown in Fig. 21 is an image of the sample 5 where the stage 4_is located in a first position of the stage 4 and the manipulator 6 is located in a first position of the manipulator 6 with the tip of manipulator 6_located outside the field of view. Subsequently, the manipulator 6_is moved to a second position of the manipulator s located in the field of view and the stage 4 is kept in its first position and a second image of the sample 5 is acquired, shown in Fig. 22. Subsequently, the manipulator s is moved to a third position of the manipulator 6 located in the field of view wherein the third position is different from the second position of the manipulators and the stage 4 is kept in its first position and a third image of sample 5 is acquired, shown in Fig. 23. Subsequently, the manipulator s is moved to a fourth position of the manipulator s located in the field of view wherein the fourth position is different from the second and third positions of the manipulator s and the stage 4 is kept in its first position and a fourth image of the sample 5 is acquired, shown in Fig. 24. Subsequently, a first difference image is created, as shown in Fig. 25, by subtracting the first image of the sample 5 and the second image of the sample 5; a second difference image shown in Fig. 26 is created by subtracting the first image of the sample 5 and the third image of the sample 5; and a third difference image shown in Fig. 27 is created by subtracting the first image of the sample 5 and the fourth image of the sample 5. Subsequently, the first image of the sample 5, the second image of the sample 5, the third image of the sample 5, the fourth image of the sample 5, the first difference image, the second difference image and the third difference image are used as inputs for the neural network and the output from the neural network are the binary images shown in Figs. 28, 29 and 30 corresponding to the second image of the sample 5, the third image of the sample 5 and the fourth image of the sample 5. Given that on each image of the sample 5 the manipulators is in a different position, and therefore on the binary images corresponding to the images of sample 5 the manipulator 6 is in different positions, it is necessary to shift the binary images corresponding to the second and third images of the sample 5 so that they are in the same position as on the binary image corresponding to the fourth image of sample 5. The shift is performed using image correlation. Subsequently, the binary images are compared using the "majority voting" technique and the result is the output binary image shown in Fig. 31 , in which the tip of the manipulator 6 is found and its coordinates are determined.

[0071] Industrial applicability

[0072] The above-described method of finding (i.e., determining the position of) the tip of the manipulator can also be used in an optical microscope instead of a device with a charged particle beam.

[0073] List of reference symbols

[0074] 1 - Source of charged particles

[0075] 2 - Column

[0076] 3 - Chamber

[0077] 4 - Stage

[0078] 5 - Sample

[0079] 6 - Manipulator

[0080] 7 - Set of elements for shaping and directing particles

[0081] 8 - Detector of signal particles

Claims

CLAIMS1 . A method for determining the position of the tip of a manipulator using a charged particle beam device comprising at least one source of charged particles, at least one detector of signal particles, a movable manipulator, a movable stage adapted for receiving a sample, a sample placed on the movable stage and an evaluation unit for performing the method, characterized in that the said method comprises the steps of obtaining at least three different images of the sample, wherein at least two of these images are acquired by irradiating the sample with a charged particle beam produced by a source of charged particles and detecting signal particles with the detector of signal particles, wherein between acquiring individual images of the sample, either the stage or the manipulator is moved, and at least one acquired image of the sample contains the tip of the manipulator in the field of view, and wherein one or more images may be a difference image of the sample obtained by subtracting two images of the sample acquired by irradiating the sample from each other,- wherein the at least three obtained images of the sample are subsequently evaluated and the position of the tip of the manipulator in one of the acquired images of the sample having the tip of the manipulator in the field of view is determined.

2. The method according to claim 1 , wherein in the step of obtaining at least three different images of the sample, at least three different images of the sample are acquired by irradiating the sample with a charged particle beam produced by a source of charged particles and detecting signal particles with the detector of signal particles, wherein between acquiring individual images of the sample, eitherthe stage or the manipulator is moved so that in at least two images the manipulator is in the same position, and in at least two images the stage is in the same position, and at least one image of the sample is with the tip of the manipulator in the field of view.

3. The method according to claim 1 , wherein in the step of obtaining at least three different images of the sample includes acquiring at least two different images by irradiating the sample with a charged a charged particle beam and detecting signal particles with the detector of signal particles, wherein between obtaining at least one pair of these sample images, the stage is moved, and at the same time at least one image of the sample is in the field of view with the tip of the manipulator, wherein pairs of obtained sample images are subsequently taken, each pair always having either the stage or the manipulator at the same position in both sample images, these images are subtracted from each other, whereby difference images are obtained, including at least one third image.

4. The method according to any one of claims 1 to 3, wherein the evaluation of the obtained images of the sample is performed using a neural network, the input of which is the acquired images of the sample and optionally difference images, and the output is information enabling the position of the manipulator tip to be found.

5. The method according to claim 4, wherein the information enabling the position of the manipulator tip is selected from coordinates of the manipulator tip on the image of the sample which shows the manipulator tip in the field of view, and a binary image corresponding to the said image, said binary image allowing to determine the said coordinates.

Images