Charged particle beam device

Abstract

Provided is a charged particle beam device which achieves removal of a foreign matter adhering to the most distal end of a needle. The charged particle beam device comprises: a charged particle beam emission optical system that emits a charged particle beam; a sample piece relocation means provided with a sample stage that is moved while having the sample placed thereon, a needle that holds and transports a sample piece separated and extracted from the sample, and a needle drive mechanism that drives the needle; a holder fixing bed that holds a sample piece holder to which the sample piece is to be relocated; a machine learning model that has learned information including an image of the needle; and a computer that controls the charged particle beam emission optical system to process an object on the basis of the determination of the machine learning model.

Description

Technical Field

[0001] The present disclosure relates to a charged particle beam device, and is particularly an effective technology applicable to a charged particle beam device having a needle.

Background Art

[0002] Patent Document 1 proposes a charged particle beam device that extracts a sample piece produced by irradiating a sample with an ion beam and transfers it to a sample holder for transmission electron microscope observation.

[0003] In a charged particle beam device, a needle is used to extract a sample piece processed by irradiation with an ion beam. The tip of the needle may undergo shape changes due to adhesion and cutting with the sample piece or foreign substances may adhere. In such cases, a cleaning process is performed to shape the tip of the needle using an ion beam.

[0004] Patent Document 1 discloses a technique for creating a desired constant shape by using image processing technology to recognize the shape of the needle tip, setting a rectangular processing frame in the outer region from the upper and lower edges of the needle, and performing ion beam processing.

Prior Art Documents

Patent Documents

[0005]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0006] However, in the above-described cleaning process, deposits may remain on the very tip of the needle, or the shape of the needle may be deformed, making it impossible to use a needle with an optimal shape for transferring the sample piece in some cases. In Patent Document 2, there is no disclosure or suggestion regarding machine learning for deposits at the tip of the needle.

[0007] This disclosure provides a charged particle beam apparatus that enables the provision of a needle with an optimal shape for transferring sample pieces.

[0008] Other challenges and novel features will become apparent from the description and accompanying drawings in this specification. [Means for solving the problem]

[0009] A brief overview of some of the representative disclosures is as follows:

[0010] According to one embodiment, the charged particle beam apparatus is A charged particle beam irradiation optical system that irradiates with a charged particle beam, A sample stage on which the sample is placed and moved, A sample piece transfer means comprising a needle for holding and transporting sample pieces to be separated and extracted from the aforementioned sample, and a needle drive mechanism for driving the needle, A holder fixing base that holds the sample piece holder to which the sample piece is transferred, A machine learning model that has learned information including the image of the needle, The system includes a computer that controls the charged particle beam irradiation optical system to process an object based on the judgment of the machine learning model. [Effects of the Invention]

[0011] According to the charged particle beam apparatus of the above embodiment, it is possible to process the needle into a shape that is optimal for transferring the sample piece. [Brief explanation of the drawing]

[0012] [Figure 1] This figure shows an example of the configuration of a charged particle beam apparatus and an image processing computer according to the embodiment. [Figure 2] This figure shows an example of the configuration of a charged particle beam apparatus according to the embodiment. [Figure 3] This is a plan view showing a sample piece according to the embodiment. [Figure 4] It is a plan view of a sample piece holder according to an embodiment. [Figure 5] It is a side view of a sample piece holder according to an embodiment. [Figure 6] It is a diagram showing an example of the configuration of a computer for image processing according to an embodiment. [Figure 7] It is a diagram showing an example of an initial setting process according to an embodiment. [Figure 8] It is a plan view of a columnar part according to an embodiment. [Figure 9] It is a side view of a columnar part according to an embodiment. [Figure 10] It is a diagram showing an example of a learning image of a columnar part according to an embodiment. [Figure 11] It is a diagram showing an example of a columnar part in which the pillar according to the embodiment does not have a stepped-up structure. [Figure 12] It is a diagram showing an example of a learning image of a columnar part in which the pillar according to the embodiment does not have a stepped-up structure. [Figure 13] It is a diagram showing an example of a sample piece pickup process according to an embodiment. [Figure 14] It is a diagram showing an example of the movement process of a needle according to an embodiment. [Figure 15] It is a diagram showing an example of SEM image data including the tip of a needle according to an embodiment. [Figure 16] It is a diagram showing an example of SIM image data including the tip of a needle according to an embodiment. [Figure 17] It is a diagram showing an example of the tip of a needle according to an embodiment. [Figure 18] It is a diagram showing an example of a learning image of a needle according to an embodiment. [Figure 19] It is a diagram showing an example of SIM image data including a sample piece according to an embodiment. tr> [Figure 20] It is a diagram showing an example of a learning image of a sample piece according to an embodiment. [Figure 21] It is a diagram showing the cutting positions of the support parts of the sample and the sample piece in the SIM image data according to the embodiment. [Figure 22] This figure shows an example of the sample mounting process according to the embodiment. [Figure 23] This figure shows an example of a needle trimming process according to an embodiment. [Figure 24] This figure shows an example of a training image of the needle according to the embodiment. [Figure 25] This figure shows an example of a learning image of a needle whose tip thickness has changed due to cleaning according to the embodiment. [Figure 26] This figure shows an example of the original image of a needle with foreign matter attached according to the embodiment. [Figure 27] This figure shows an example of a training image that indicates the region of a foreign object in the original image of Figure 26 according to the embodiment. [Figure 28] This figure shows an example of the original image with the contrast, angle, magnification, and position of the needle according to the embodiment changed. [Figure 29] This figure shows an example of a set of training data consisting of an original image and a training image, where foreign matter is attached to the tip of a needle whose tip thickness has changed due to cleaning according to the embodiment. [Figure 30] This figure shows an example of a learning model for determining whether needle cleaning is necessary according to the embodiment. [Figure 31] This is a diagram illustrating an example of a processing frame related to a comparative example. [Figure 32] This diagram illustrates the problems with the processing frame related to the comparative example. [Figure 33] This is a diagram illustrating an example of a processing frame according to the embodiment. [Figure 34] This diagram illustrates an example of setting up a processing frame according to an embodiment. [Figure 35] This is a diagram illustrating the shaping of the needle according to the embodiment. [Modes for carrying out the invention]

[0013] The embodiments will be described below with reference to the drawings. However, in the following description, the same reference numerals will be used for the same components, and repeated explanations may be omitted. In addition, the drawings may be more schematic than the actual embodiments in order to make the explanation clearer, but they are merely examples and do not limit the interpretation of this disclosure. [Examples]

[0014] Embodiments of the present invention will be described in detail below with reference to the drawings. Figure 1 is a diagram showing an example of the configuration of a charged particle beam apparatus 10 and an image processing computer 30 according to this embodiment.

[0015] The control computer 22, provided in the charged particle beam apparatus 10, acquires image data obtained by irradiation with the charged particle beam. The control computer 22 transmits and receives data with the image processing computer 30. The image processing computer 30 determines the objects contained in the image data received from the control computer 22 based on a machine learning model M. Based on the determination results of the image processing computer 30, the control computer 22 performs actions such as controlling the position of the objects and removing foreign matter attached to the tip of the needle.

[0016] The control computer 22 is an example of a computer that controls the position of a second object based on a machine learning model that has learned first information including a first image of the first object, and second information including a second image acquired by irradiation with a charged particle beam. The image processing computer 30 may be provided in the charged particle beam apparatus 10.

[0017] (Overall configuration of the charged particle beam device) Next, the configuration of the charged particle beam apparatus 10 will be described with reference to Figure 2. Figure 2 is a diagram showing an example of the configuration of the charged particle beam apparatus 10 according to the embodiment.

[0018] The charged particle beam apparatus 10 comprises a sample chamber 11, a sample stage 12, a stage drive mechanism 13, a focused ion beam irradiation optical system (also called a charged particle beam irradiation optical system) 14, an electron beam irradiation optical system 15, a detector 16, a gas supply unit 17, a needle 18, a needle drive mechanism 19, an absorption current detector 20, a display device 21, a control computer 22, and an input device 23.

[0019] The sample chamber 11 maintains a vacuum inside. The sample stage 12 fixes the sample S and the sample piece holder P inside the sample chamber 11. The sample stage 12 includes a holder fixing base 12a for holding the sample piece holder P. This holder fixing base 12a may have a structure that can accommodate multiple sample piece holders P.

[0020] The stage drive mechanism 13 drives the sample stage 12. Here, the stage drive mechanism 13 is housed inside the sample chamber 11 while connected to the sample stage 12, and displaces the sample stage 12 around a predetermined axis in accordance with control signals output from the control computer 22. The stage drive mechanism 13 includes a movement mechanism 13a that moves the sample stage 12 parallel to at least the X and Y axes which are parallel to and orthogonal to the horizontal plane, and the Z axis which is perpendicular to the vertical direction and orthogonal to the X and Y axes. The stage drive mechanism 13 also includes a tilt mechanism 13b that tilts the sample stage 12 around the X or Y axis, and a rotation mechanism 13c that rotates the sample stage 12 around the Z axis.

[0021] The focused ion beam irradiation optical system 14 irradiates the target to be irradiated within a predetermined irradiation area (i.e., scanning range) inside the sample chamber 11 with a focused ion beam (FIB). Here, the focused ion beam irradiation optical system 14 irradiates the target to be irradiated, such as the sample S placed on the sample stage 12, the sample piece Q, and the needle 18 present within the irradiation area, with the focused ion beam directed from vertically upwards to downwards.

[0022] The focused ion beam irradiation optical system 14 comprises an ion source 14a for generating ions and an ion optical system 14b for focusing and deflecting the ions drawn from the ion source 14a. The ion source 14a and the ion optical system 14b are controlled according to control signals output from a control computer 22, and the irradiation position and irradiation conditions of the focused ion beam are controlled by the control computer 22.

[0023] The electron beam irradiation optical system 15 irradiates the target to be irradiated within a predetermined irradiation area inside the sample chamber 11 with an electron beam (EB). Here, the electron beam irradiation optical system 15 can irradiate the target to be irradiated, such as the sample S fixed to the sample stage 12, the sample piece Q, and the needle 18 present in the irradiation area, with the electron beam directed from above to below in an inclined direction tilted at a predetermined angle (e.g., 60°) with respect to the vertical.

[0024] The electron beam irradiation optical system 15 comprises an electron source 15a that generates electrons and an electron optical system 15b that focuses and deflects the electrons emitted from the electron source 15a. The electron source 15a and the electron optical system 15b are controlled according to control signals output from a control computer 22, and the irradiation position and irradiation conditions of the electron beam are controlled by the control computer 22.

[0025] Alternatively, the arrangement of the electron beam irradiation optical system 15 and the focused ion beam irradiation optical system 14 may be swapped, with the electron beam irradiation optical system 15 positioned vertically and the focused ion beam irradiation optical system 14 positioned vertically at a predetermined angle.

[0026] The detector 16 detects secondary charged particles (secondary electrons, secondary ions) R generated from the irradiated object by irradiation with a focused ion beam or electron beam. The gas supply unit 17 supplies gas G to the surface of the irradiated object. The needle 18 picks up a minute sample piece Q from the sample S fixed to the sample stage 12, holds the sample piece Q, and transfers it to the sample piece holder P. The needle drive mechanism 19 drives the needle 18 to transport the sample piece Q. Hereafter, the needle 18 and the needle drive mechanism 19 may collectively be referred to as the sample piece transfer means.

[0027] The absorption current detector 20 detects the inflow current (also called the absorption current) of the charged particle beam flowing into the needle 18, and outputs the detected result as an inflow current signal to the control computer 22.

[0028] The control computer 22 controls at least the stage drive mechanism 13, the focused ion beam irradiation optical system 14, the electron beam irradiation optical system 15, the gas supply unit 17, and the needle drive mechanism 19. The control computer 22 is located outside the sample chamber 11 and is connected to a display device 21 and an input device 23 such as a mouse or keyboard that outputs signals in response to the operator's input. The control computer 22 comprehensively controls the operation of the charged particle beam apparatus 10 using signals output from the input device 23 or signals generated by a preset automatic operation control process.

[0029] As described above, the control computer 22 controls the position of the object based on the determination result of the image processing computer 30. The control computer 22 is equipped with a communication interface for communicating with the image processing computer 30.

[0030] Furthermore, the control computer 22 images the inflow current signal output from the absorption current detector 20 as absorption current image data. Here, the control computer 22 converts the amount of secondary charged particles R detected by the detector 16 while scanning the irradiation position of the charged particle beam into a brightness signal corresponding to the irradiation position, and generates absorption current image data that shows the shape of the irradiation target by the two-dimensional position distribution of the amount of detected secondary charged particles R. In absorption current image mode, the control computer 22 detects the absorption current flowing through the needle 18 while scanning the irradiation position of the charged particle beam, and generates absorption current image data that shows the shape of the needle 18 by the two-dimensional position distribution of the absorption current (absorption current image). The control computer 22 displays on the display device 21 a screen for performing operations such as scaling, moving, and rotating each generated image data. The computer 22 also displays on the display device 21 a screen for performing various settings such as mode selection and processing settings in automatic sequence control.

[0031] The display device 21 displays image data and other information based on the secondary charged particles R detected by the detector 16.

[0032] Furthermore, the image processing computer 30 takes in the image data generated by the control computer 22, identifies the area of ​​foreign matter attached to the needle 18, and transmits the coordinate information of the area to the control computer 22. The control computer 22 sets up a processing frame that defines the operating area of ​​the charged particle beam according to the coordinates of the area of ​​foreign matter received from the image processing computer 30.

[0033] The charged particle beam apparatus 10 can perform imaging of the target surface, various processing methods such as sputtering (excavation, trimming, etc.), and deposition film formation by scanning the surface with a focused ion beam.

[0034] Figure 3 is a plan view showing a sample piece Q before it is extracted from the sample S, formed by irradiating the surface (shaded area) of the sample S with a focused ion beam in the charged particle beam apparatus 10 according to the embodiment. The symbol F indicates the processing frame by the focused ion beam, that is, the scanning range of the focused ion beam, and the area inside it (white area) indicates the processing region H that has been excavated by sputtering due to irradiation with the focused ion beam. The reference mark Ref is a reference point that indicates the position where the sample piece Q is formed (left unexcavated). A deposition film is used to know the approximate position of the sample piece Q, and micro-holes are used for precise positioning. In the sample S, the sample piece Q is etched so that the peripheral parts on the sides and bottom are removed, leaving a support part Qa connected to the sample S, and is cantilevered to the sample S by the support part Qa.

[0035] Next, the sample holder P will be described with reference to Figures 4 and 5.

[0036] Figure 4 is a plan view of the sample holder P, and Figure 5 is a side view. The sample holder P comprises a roughly semicircular plate-shaped base 42 having a notch 41, and a sample stage 43 fixed to the notch 41. The base 42 is formed of a circular plate-shaped metal as an example. The sample stage 43 has a comb-like shape and comprises multiple spaced-apart, protruding columnar parts (hereinafter also called pillars) 44 to which the sample piece Q is moved.

[0037] (Image processing computer) Next, the image processing computer 30 will be described with reference to Figure 6. Figure 6 is a diagram showing an example of the configuration of the image processing computer 30 according to this embodiment.

[0038] The image processing computer 30 comprises a control unit 300 and a storage unit 305.

[0039] The control unit 300 includes a learning data acquisition unit 301, a learning unit 302, a judgment image acquisition unit 303, and a judgment unit 304.

[0040] The training data acquisition unit 301 acquires training data. Training data is information used for training machine learning. Training data is a pair of a training image and information indicating the position of an object within the training image. Examples of objects in the training image include a sample piece, a needle, foreign matter attached to the tip of the needle, and a columnar part provided in the sample piece holder. Here, the type of object in the training image is the same as the type of object in the judgment image. For example, if the type of object in the training image is a sample piece, a needle, foreign matter attached to the tip of the needle, or a columnar part, then the types of objects in the judgment image will be the sample piece, a needle, foreign matter attached to the tip of the needle, or a columnar part, respectively.

[0041] In this embodiment, SIM images or SEM images obtained in advance by irradiating an object with a charged particle beam are used as training images. The charged particle beam is irradiated onto the object from a predetermined direction. In the charged particle beam apparatus 10, the direction of the microscope tube of the charged particle beam irradiation system is fixed, so the direction in which the charged particle beam is irradiated onto the object is predetermined.

[0042] Information indicating the position of an object within a training image is, for example, the coordinates that show the position of that object within the training image. These coordinates indicating the position within the training image may be, for example, two-dimensional Cartesian coordinates or polar coordinates.

[0043] The training images include both SIM and SEM images of the object. The training images include both a SIM image of the object viewed from a direction tilted at a predetermined angle relative to the vertical direction of the sample stage 12, and a SEM image of the object viewed from the vertical direction of the sample stage 12. In other words, the training images include an image of the object viewed from a first direction relative to the sample stage 12, and an image of the object viewed from a second direction. The second direction is different from the first direction relative to the sample stage 12.

[0044] The learning unit 302 performs machine learning based on the learning data acquired by the learning data acquisition unit 301. The learning unit 302 stores the learned results as a machine learning model M in the storage unit 305. The learning unit 302 performs machine learning for each type of object in the learning images included in the learning data. Therefore, a machine learning model M is generated for each type of object in the learning images included in the learning data. The machine learning model M is an example of a machine learning model in which first information, including the first image of the first object, has been learned. In the following explanation, objects captured or drawn in an image may be referred to as the objects in this image.

[0045] The machine learning performed by the learning unit 302 here is deep learning, such as a convolutional neural network (CNN). In this case, the machine learning model M is a multilayer neural network in which the weights between nodes are changed according to the correspondence between the training image and the position of the object in the training image. This multilayer neural network has an input layer with nodes corresponding to each pixel of the image and an output layer with nodes corresponding to each position in the image. When the brightness value of each pixel of the SIM image or SEM image is input to the input layer, a set of values ​​indicating the position in the image is output from the output layer.

[0046] The judgment image acquisition unit 303 acquires judgment images. Judgment images are SIM images and SEM images output from the control computer 22. Judgment images include images of the objects described above. Objects in the judgment images include objects related to the irradiation of the charged particle beam, such as the sample piece Q, the used needle 18, and foreign matter attached to the tip of the needle 18.

[0047] The judgment image consists of both a SIM image of the object viewed from an inclined direction tilted at a predetermined angle with respect to the vertical direction of the sample stage 12, and a SEM image of the object viewed from the vertical direction of the sample stage 12. In other words, the judgment image includes an image of the object viewed from a first direction and an image of the object viewed from a second direction. Here, the first direction is the direction relative to the sample stage 12, and the second direction is a direction different from the first direction relative to the sample stage 12.

[0048] The determination unit 304 determines the position of an object included in the determination image acquired by the determination image acquisition unit 303 based on the machine learning model M trained by the learning unit 302. Here, the position of an object included in the determination image includes, for example, the pickup position of a sample piece in the SIM image or SEM image, the position of the needle tip in the SIM image or SEM image, the position of a foreign object at the tip of the needle 18 in the SIM image or SEM image, and the position of a columnar part 44 in the SIM image or SEM image. As an example of the position of an object included in the determination image, the determination unit 304 determines the coordinates of an object in the determination image. The needle 18 here may have a sharpened cone shape, a sharpened cylindrical shape, or a sharpened prismatic shape at its tip. Depending on the shape of the needle 18 tip, a different learning model for the tip shape of the needle 18 can be used. This makes it possible to use needles 18 with various tip shapes in the charged particle beam apparatus 10. The learning model also includes a learning model for the shape of the needle 18 whose tip shape changes due to use, and it is possible to use different models depending on the usage of the needle 18. Here, "using different models" means changing the learning model according to the change in tip shape due to use. As a result, one needle 18 can be used for a long period of time in one charged particle beam apparatus 10. Therefore, the frequency of needle 18 replacement can be reduced, and the number of maintenance cycles for the charged particle beam apparatus 10 can be reduced. This allows the operating time of the charged particle beam apparatus 10 to be extended.

[0049] The image processing computer 30 may obtain a trained machine learning model from, for example, an external database. In that case, the control unit 300 does not need to include a training data acquisition unit 301 and a training unit 302.

[0050] The following describes the operation of automatic micro-sampling (MS) performed by the control computer 22, that is, the operation of automatically transferring the sample piece Q formed by processing the sample S with a charged particle beam (focused ion beam) to the sample piece holder P. This will be divided into four main steps: initial setup, sample piece pickup, sample piece mounting, and needle trimming.

[0051] (Initial setting process) Figure 7 shows an example of the initial setup process according to this embodiment.

[0052] Step S10: The control computer 22 sets the mode and processing conditions. Mode setting refers to setting whether or not to use the attitude control mode (described later) in response to the operator's input at the start of the automatic sequence. Processing condition setting refers to setting the processing position, dimensions, number of sample pieces Q, etc.

[0053] Step S20: The control computer 22 registers the position of the columnar portion 44. At this point, the control computer 22 transmits the SIM image or SEM image containing the columnar portion 44 as the object to the image processing computer 30.

[0054] In this embodiment, the absorbed current image data containing the object is a set of a SIM image of the object and a SEM image of the object. In other words, the SIM image and SEM image containing the object are a set of a SIM image of the object viewed from an inclined direction tilted at a predetermined angle with respect to the vertical direction of the sample stage 12 and a SEM image of the object viewed from the vertical direction of the sample stage 12.

[0055] The judgment image acquisition unit 303 acquires SIM images or SEM images as judgment images from the image processing computer 30. The judgment unit 304 determines the position of the columnar portion 44 included in the judgment image acquired by the judgment image acquisition unit 303 based on the machine learning model M. The judgment unit 304 outputs position information indicating the determined position of the columnar portion 44 to the control computer 22.

[0056] Here, the determination unit 304 determines the two-dimensional coordinates of the object's position on the sample stage 12 from the SIM image of the object viewed from an inclined direction tilted at a predetermined angle with respect to the vertical direction of the sample stage 12. On the other hand, the determination unit 304 determines the two-dimensional coordinates of the object's position in a plane perpendicular to the inclined direction from the SEM image of the object viewed from an inclined direction tilted at a predetermined angle with respect to the vertical direction of the sample stage 12. Based on the determined two-dimensional coordinates on the sample stage 12 and the two-dimensional coordinates in the plane perpendicular to the inclined direction, the determination unit 304 determines the object's position as a three-dimensional coordinate value.

[0057] The determination unit 304 uses directional information, which is information about the direction in which the electron beam irradiation optical system 15 and the focused ion beam irradiation optical system 14 are arranged and the angle between them, to calculate the value of the three-dimensional coordinates. The determination unit 304 either stores the directional information in the storage unit 305 in advance and reads it, or obtains it from the control computer 22. In step S20, the object is the columnar part 44. In the following steps, the process by which the determination unit 304 determines the position of the object is the same.

[0058] Here, referring to Figures 8 to 12, we will describe the columnar portion 44 and the training images of the columnar portion 44 used in generating the machine learning model M. Figures 8 and 9 show an example of the columnar portion 44 according to this embodiment. The columnar portion A0 shown in Figures 8 and 9 is an example of the design structure of the columnar portion 44. Here, Figure 8 is a top view of the columnar portion A0, and Figure 9 is a side view of the columnar portion A0. The columnar portion A0 has a structure in which a pillar A01 with a stepped structure is bonded to the base A02.

[0059] Figure 10 shows an example of a training image of the columnar portion 44 according to this embodiment. Training images X11, X12, and X13 are used to train the position of the columnar portion 44. In training images X11, X12, and X13, information indicating the position of the columnar portion is shown as a circle. In training images X11, X12, and X13, the shapes of pillars A11, A21, and 31 are different. On the other hand, the shapes of bases A12, A22, and A32 are the same in training images X11, X12, and X13.

[0060] Note that training images X11, X12, and X13 are training images used to determine the position of the columnar portion 44 included in SIM images and SEM images when the columnar portion 44 is viewed from the horizontal direction of the sample stage 12, as an example. In Figure 2, the focused ion beam irradiation optical system 14 and the electron beam irradiation optical system 15 do not face the sample stage 12 from the horizontal direction of the sample stage 12, but either the focused ion beam irradiation optical system 14 or the electron beam irradiation optical system 15 may face the sample stage 12 from the horizontal direction, and training images X11, X12, and X13 are training images used to determine the position of the columnar portion 44 in that case.

[0061] Figure 11 shows an example of a columnar section 44 in which the pillar does not have a stepped structure according to this embodiment. The columnar section A4 shown in Figure 11 is a side view of an example of the design structure of the columnar section 44 in which the pillar does not have a stepped structure.

[0062] Figure 12 shows an example of a training image of a columnar portion 44 in this embodiment where the pillar does not have a stepped structure. Training images X21, X22, and X23 are, as an example, training images for determining the position of the columnar portion 44 included in the SEM image when the columnar portion 44 is viewed from the vertical direction of the sample stage 12.

[0063] In training images X21, X22, and X23, the shapes of pillars A51, A61, and 71 are different. On the other hand, the shapes of bases A52, A62, and A72 are the same in training images X21, X22, and X23.

[0064] Since the machine learning model M is generated based on machine learning using training images that include the base of the columnar portion 44, the shape of the base is learned as a feature in the machine learning model M. Therefore, in the charged particle beam apparatus 10, the accuracy of determining the columnar portion is improved even when the shape of the pillar is different. It is preferable that the target objects in the training images include parts with the same shape among the target objects in multiple training images.

[0065] Returning to Figure 7, we continue the explanation of the initial setup process.

[0066] The control computer 22 registers the position of the columnar portion 44 based on position information indicating the position of the columnar portion 44 determined by the image processing computer 30.

[0067] Preferably, the training images of the columnar portion 44 include images of the columnar portions located at both ends of the sample stage 43. Based on the machine learning model M generated using this training data including the training images, the image processing computer 30 detects the columnar portions at both ends of the sample stage 43, distinguishing them from the other columnar portions. The control computer 22 may calculate the inclination of the sample holder P from the positions of the detected columnar portions at both ends. The control computer 22 may correct the coordinate values ​​of the object's position based on the calculated inclination.

[0068] Step S30: The control computer 22 controls the focused ion beam irradiation optical system 14 to process the sample S.

[0069] (Sample pickup process) Figure 13 shows an example of the sample piece pickup process according to this embodiment. Here, "pickup" refers to separating and extracting the sample piece Q from the sample S by processing with a focused ion beam or by using a needle.

[0070] Step S40: The control computer 22 adjusts the position of the sample. Here, the control computer 22 moves the sample stage 12 by the stage drive mechanism 13 to bring the target sample piece Q into the field of view of the charged particle beam. Here, the control computer 22 uses the relative positional relationship between the reference mark Ref and the sample piece Q. After moving the sample stage 12, the control computer 22 aligns the sample piece Q.

[0071] Step S50: The control computer 22 performs the movement of the needle 18.

[0072] Now, referring to Figure 14, the process for moving the needle 18 performed by the control computer 22 will be described. Figure 14 is a diagram showing an example of the needle 18 movement process according to this embodiment. Steps S510 to 540 in Figure 14 correspond to step S50 in Figure 13.

[0073] Step S510: The control computer 22 performs needle movement (coarse adjustment) by moving the needle 18 with the needle drive mechanism 19.

[0074] Step S520: The control computer 22 detects the tip of the needle 18. The control computer 22 then transmits absorbed current image data, which includes the needle 18 as the object, to the image processing computer 30.

[0075] The judgment image acquisition unit 303 acquires SIM images and SEM images as judgment images from the image processing computer 30. The judgment unit 304 determines the position of the needle 18 included in the judgment images acquired by the judgment image acquisition unit 303 as the position of the target object, based on the machine learning model M. The judgment unit 304 outputs position information indicating the determined position of the needle 18 to the control computer 22.

[0076] Next, the control computer 22 performs needle movement (fine adjustment) by moving the needle 18 using the needle drive mechanism 19, based on the position information indicating the position of the needle 18 determined by the image processing computer 30.

[0077] Here, we will refer to Figures 15 to 18 to explain the needle 18 and the training images of the needle 18 used to generate the machine learning model M. Figure 15 is a diagram showing an example of SEM image data including the tip of the needle 18 according to this embodiment. Figure 16 is a diagram showing an example of SIM image data including the tip of the needle 18 according to this embodiment.

[0078] Figure 17 shows an example of the tip of the needle 18 according to this embodiment. In Figure 17, as an example of the needle 18, needle B1 is shown as viewed from an inclined direction that is tilted at a predetermined angle with respect to the vertical direction of the sample stage 12.

[0079] Figure 18 shows an example of a training image of the needle 18 according to this embodiment. Training images Y31, Y32, and Y33 are used to train the position of the tip of the needle 18. In training images Y31, Y32, and Y33, the information indicating the position of the tip of the needle 18 is shown as a circle. In training images Y31, Y32, and Y33, the thickness of the needle tip is different in each case. On the other hand, in training images Y31, Y32, and Y33, the shape of the needle tip is the same.

[0080] The actual thickness of the needle tip 18 changes due to cleaning. Since the machine learning model M is generated based on machine learning using training images that include the tip of the needle 18, the shape of the needle tip is learned as a feature in the machine learning model M. Therefore, the charged particle beam apparatus 10 can improve the accuracy of needle tip detection even when the thickness of the needle tip varies.

[0081] Returning to Figure 14, we will continue the explanation of the movement process of needle 18.

[0082] Step S530: The control computer 22 detects the pickup position of the sample piece Q. At this point, the control computer 22 transmits the SIM image or SEM image containing the sample piece Q as the object to the image processing computer 30.

[0083] Here, referring to Figures 19 and 20, we will explain the sample piece Q and the training images of sample piece Q used to generate the machine learning model M.

[0084] Figure 19 shows an example of SIM image data containing sample piece Q according to this embodiment. In Figure 19, sample piece Q71 is shown as an example of sample piece Q, along with a circle indicating the pickup position.

[0085] Figure 20 shows an example of a training image of a sample piece Q according to this embodiment. Training images Z11, Z12, and Z13 are used to train the position of the tip of the sample piece Q. In training images Z11, Z12, and Z13, information indicating the pickup position of the sample piece Q is shown as a circle. In training images Z11, Z12, and Z13, the size and surface shape of the sample piece are different. On the other hand, in training images Z11, Z12, and Z13, the shape of the sample piece at the pickup position is the same.

[0086] The actual surface shape of each sample piece differs. Since the machine learning model M is generated based on machine learning using training images that include the pickup position of the sample piece Q, the shape of the pickup position of the sample piece Q is learned as a feature in the machine learning model M. Therefore, even when the surface shape of the sample pieces differs, the accuracy of determining the pickup position of the sample piece Q is improved in the charged particle beam apparatus 10.

[0087] Returning to Figure 14, we will continue the explanation of the movement process of needle 18.

[0088] Step S540: The control computer 22 moves the needle 18 to the detected pickup position.

[0089] With this, the control computer 22 terminates the process of moving the needle 18.

[0090] Returning to Figure 13, we will continue the explanation of the sample pickup process.

[0091] Step S60: The control computer 22 connects the needle 18 to the sample piece Q. Here, the control computer 22 makes the connection using a deposition film.

[0092] Step S70: The control computer 22 processes and separates the sample S and the sample piece Q. Figure 21 shows the processing and separation process, and is a diagram showing the cutting position T1 of the support portion Qa of the sample S and the sample piece Q in the SIM image data according to the embodiment of the present invention.

[0093] In this embodiment, the sample piece pickup process and the sample piece mounting process may also be performed on a sample piece Q0 that has been prepared and processed separately in advance. In this case, the pickup position of the sample piece Q0 may be specified and input to the control computer 22, and after the position of the sample piece transfer means (needle 18) and the sample piece Q0 is adjusted, the cutting position T1 in Figure 21 may be determined by machine learning. In this case, the machine learning uses an image as the first image that shows the position (cutting position) where the sample piece transfer means is brought closer to the sample piece in the sample extraction process in which the sample piece is extracted.

[0094] In this case, even if processing size and shape information indicating the processing size and shape of the sample piece Q0 is not input to the control computer 22, the sample piece Q0 can be extracted and separated. Furthermore, after the extraction of the sample piece Q0, the subsequent sample piece mounting process may be carried out in the same manner.

[0095] Step S80: The control computer 22 retracts the needle 18. Here, the control computer 22 detects the position of the tip of the needle 18 in the same manner as the needle 18 movement process in step S50, and moves the needle 18 to retract it.

[0096] Step S90: The control computer 22 moves the sample stage 12. Here, the control computer 22 moves the sample stage 12 by the stage drive mechanism 13 so that the specific columnar portion 44 registered in step S20 above comes into the observation field of view of the charged particle beam.

[0097] (Sample mounting process) Figure 22 shows an example of the sample piece mounting process according to this embodiment. Here, the sample piece mounting process is the process of transferring the extracted sample piece Q to the sample piece holder P.

[0098] Step S100: The control computer 22 determines the relocation position of the sample piece Q. Here, the control computer 22 determines the specific columnar portion 44 registered in step S20 above as the relocation position.

[0099] Step S110: The control computer 22 detects the position of the needle 18. Here, the control computer 22 detects the position of the tip of the needle 18 in the same manner as in step S520 described above.

[0100] Step S120: The control computer 22 moves the needle 18. Here, the control computer 22 moves the needle 18 by the needle drive mechanism 19 to the relocation position of the sample piece Q determined in step S100. The control computer 22 stops the needle 18, leaving a predetermined gap between the columnar part 44 and the sample piece Q.

[0101] Step S130: The control computer 22 connects the sample piece Q connected to the needle 18 to the columnar portion 44.

[0102] Step S140: The control computer 22 separates the needle 18 and the sample piece Q. Here, the control computer 22 performs the separation by cutting the deposition film DM2 that connects the needle 18 and the sample piece Q.

[0103] Step S150: The control computer 22 retracts the needle 18. Here, the control computer 22 moves the needle 18 away from the sample piece Q by a predetermined distance using the needle drive mechanism 19.

[0104] Step S160: The control computer 22 determines whether or not to perform the next sampling. Performing the next sampling means continuing to sample from a different location of the same sample S. Since the number of samples to be sampled was registered in advance in step S10, the control computer 22 checks this data and determines whether to perform the next sampling.

[0105] If the control computer 22 determines that it will perform the next sampling (No), it performs the needle trimming process S230 shown in Figure 23, then proceeds to step S50 and continues the subsequent steps as described above to perform the sampling operation. In this example, step S160 includes the needle trimming process S230. The needle trimming process S230 may be performed between step S160 and step S50.

[0106] On the other hand, if it is determined that no further sampling should be performed (Yes), the control computer 22 terminates the series of flows of the automatic MS.

[0107] Next, the needle trimming process (step S230) will be explained using Figure 23. Figure 23 is a diagram showing an example of the needle trimming process according to the embodiment.

[0108] (Step S230: Needle trimming process) The control computer 22 executes the needle trimming process S230.

[0109] Step S231: In automatic sample sampling, after sampling, that is, after separating the sample piece Q separated and extracted from the sample S by the needle 18, the control computer 22 performs needle cleaning by trimming the needle 18. This allows the control computer 22 to repeatedly use the needle 18 when separating and extracting the sample piece Q from the sample S. The control computer 22 removes any deposits such as the deposition film DM2 and sample piece residue attached to the needle 18 by etching using a focused ion beam. These deposits such as the deposition film DM2 and sample piece residue attached to the needle 18 can be described as foreign matter.

[0110] Step S232: The control computer 22 moves the needle 18 and the stage 13 to a location where there are no structures behind the needle 18 and stops them.

[0111] Step S233: The control computer 22 acquires image data including the needle 18 by irradiation with a focused ion beam. The acquired image including the needle 18 is transmitted to the image processing computer 30. The determination image acquisition unit 303 acquires the image from the image processing computer 30 as a determination image.

[0112] Step S234: The determination unit 304, based on the machine learning model M, determines whether the tip of the needle 18 is included in the determination image acquired by the determination image acquisition unit 303. If it is determined that the tip of the needle 18 has been detected (Yes), the determination unit 304 outputs position information indicating the position of the determined needle 18 to the control computer 22 and proceeds to step S235. If it is determined that the tip of the needle 18 has not been detected (No), the process proceeds to step S232 and the subsequent steps are executed as described above.

[0113] Here, referring to Figures 24 and 25, we will explain an example of training images for needle 18 used to generate the machine learning model M. Note that training images Y31-Y33 shown in Figure 18, which were explained earlier, can also be used.

[0114] Figure 24 shows an example of a training image of the needle according to this embodiment. Training images Y41, Y42, and Y43 are used to train the position of the tip of the needle 18. In training images Y41, Y42, and Y43, the information indicating the position of the tip of the needle 18 is shown as a circle. In training images Y41, Y42, and Y43, the thickness of the needle tip is different in each case. On the other hand, the shape of the needle tip is the same in training images Y41, Y42, and Y43.

[0115] The actual thickness of the tip of the needle 18 changes due to cleaning. Figure 25 shows an example of a training image of a needle whose tip thickness has changed due to cleaning according to this embodiment. Training images Y51, Y52, and Y53 are examples of training images of a needle 18 whose tip thickness has changed due to cleaning, and are used to learn the position of the tip of the needle 18. In training images Y51, Y52, and Y53, the information indicating the position of the tip of the needle 18 is shown as a circle. Training images Y51, Y52, and Y53 show the case when the needle shown in training images Y41, Y42, and Y43 has changed in tip thickness due to cleaning. Note that the tip of the needle 18 may also become thicker if the needle 18 is accidentally cut near the base during use. Training images Y51, Y52, and Y53 should be created taking into account the possibility of accidentally cutting the needle 18.

[0116] Returning to Figure 23, we will now explain step S235.

[0117] Step S235: The control computer 22 moves the needle 18 and the stage 13 to move the tip of the needle 18 to the center of the field of view.

[0118] Step S236: The control computer 22 moves the tip of the needle 18 to the center of the field of view, and then stops the movement of the needle 18.

[0119] Step S237: The control computer 22 acquires image data including the tip of the needle 18 by irradiation with a focused ion beam. The acquired image including the needle 18 is transmitted to the image processing computer 30.

[0120] Step S238: The control computer 22 determines whether or not the needle 18 needs to be cleaned. The determination image acquisition unit 303 acquires an image from the image processing computer 30 as the determination image. The determination unit 304 determines whether or not foreign matter is attached to the tip of the needle 18 based on the machine learning model M. If it is determined that no foreign matter is attached to the tip of the needle 18, it determines that cleaning of the needle 18 is unnecessary (No) and proceeds to step S239. The machine learning model M used here has been trained on the training images of the needle 18 described in Figures 24 and 25, so it can accurately determine the shape of the needle 18 and therefore can accurately determine whether or not foreign matter is attached to the tip of the needle 18.

[0121] If it is determined that foreign matter is attached to the tip of the needle 18, it is determined that cleaning of the needle 18 is necessary (Yes), and the process proceeds to step S240. At this time, the position of the foreign matter attached to the tip of the needle 18, which is included in the determination image acquired by the determination image unit 303, is determined as the position of the target object. The determination unit 304 outputs position information indicating the determined position of the needle 18 to the control computer 22. The machine learning model M used here includes a learning model for foreign matter area determination and a learning model for needle cleaning necessity determination, which are described below.

[0122] Here, referring to Figures 26, 27, 28, 29, and 30, we will explain an example of training images of needle 18 used to generate the machine learning model M used to determine whether cleaning is necessary. Figures 26 to 29 show an example of a training model for determining the presence of foreign matter. Figure 26 shows an example of the original image of a needle with foreign matter attached. Figure 27 shows an example of a training image showing the region of foreign matter in the original image of Figure 26. Figure 28 shows an example of the original image with changes in contrast, angle, magnification, and position of the needle. Figure 29 shows an example of a set of training data for the original image and its training image, showing a needle with foreign matter attached to the tip whose thickness has changed due to cleaning. Figure 30 shows an example of a training model for determining whether needle cleaning is necessary.

[0123] As shown in Figure 26, the original images Y61 to Y66 of needle 18 show foreign matter FB attached to the tip of needle 18. Figure 27 shows training images Y61T to Y66T, which indicate the region of foreign matter FB in the original images Y61 to Y66. For the generation of the machine learning model M, the original image Y61 and the training image Y61T are used as a pair of training data. Similarly, the original image Y62 and the training image Y62T, the original image Y63 and the training image Y63T, the original image Y64 and the training image Y64T, the original image Y65 and the training image Y65T, and the original image Y66 and the training image Y66T are used as a pair of training data.

[0124] The training data used to generate the machine learning model M consists of original images and training images. The training images contain information that shows only the foreign object portion of the original image. For example, training images are created by specifying where the foreign object region is in the original image. By incorporating multiple training data into the machine learning model M, it becomes possible to determine the foreign object region from the original image of a needle to which a foreign object has been attached. It is also possible to add training data to the machine learning model M later, based on image data of the tip of the needle 18 acquired during the use of the charged particle beam apparatus 10.

[0125] Figure 28 shows the original images Y67-Y70 with varying contrast, angle, magnification, and position of the needle 18. It is preferable to use images with varying tip shapes of the needle 18, attachment patterns, contrast, magnification, angle, and position of the needle 18 as source images for image training data, as illustrated in Figure 28. By creating a training model incorporating images under various conditions, robust detection of foreign object regions becomes possible.

[0126] Figure 29 shows the original image Y71 with foreign matter FB attached to the tip of needle 18 whose tip thickness has changed due to cleaning, and its training image Y71T. Even if the tip shape of needle 18 loses its sharpness due to repeated use, it is advisable to prepare another machine learning model for detecting foreign matter, consisting of a training set of the original image (Y71) of the needle tip in a worn state and the training image (Y71T) of the foreign matter attached to it, in order to accurately determine the area of ​​foreign matter. By using different machine learning models depending on the frequency of needle cleaning, the detection of foreign matter can be performed stably.

[0127] Figure 30 shows an example of a machine learning model for determining whether needle cleaning is necessary. The machine learning model for determining whether needle cleaning is necessary includes image Y81 showing an example of needle 18 that does not need cleaning, and image Y82 showing an example of needle 18 that needs cleaning. The determination of whether needle cleaning is necessary may be made using a machine learning model for detecting foreign matter areas (Figures 26-29) or a machine learning model for determining whether needle cleaning is necessary (Figure 30).

[0128] Furthermore, in step S238, if the area (number of pixels) of the region identified as a foreign object is smaller than a certain value, it is possible to determine that "there is no impact on microsampling" and skip the cleaning process described later (steps S240, S241). This determination can be made by applying the determination results of the foreign object region determination model (Figures 26, 27, etc.).

[0129] Returning to Figure 23, we will now explain step S239.

[0130] Step S239: The control computer 22 proceeds to step S50 for processing the next sample piece Q.

[0131] Step S240: The control computer 22 determines the cleaning area of ​​the needle 18. Based on the position information indicating the location of foreign matter attached to the tip of the needle 18, which has been determined by the image processing computer 30, the control computer 22 sets a processing frame (also called a processing area) for performing etching using a focused ion beam. The setting of the processing frame will be described later.

[0132] Step S241: The control computer 22 etches within the set processing frame range with a focused ion beam to remove foreign matter such as deposition films and residues of sample pieces attached to the needle 18, and then shapes the tip of the needle 18 into the desired shape with the focused ion beam. After that, the control computer 22 proceeds to step S237 and executes the subsequent steps.

[0133] In this manner, the control computer 22 uses a machine learning model M, which has learned information including an image of the needle, to control the charged particle beam irradiation optical system to shape the needle 18 based on the judgment of the machine learning model M. As a result, the control computer 22 can remove foreign matter such as deposition films and residues of sample pieces attached to the needle 18, and then shape the tip of the needle 18 to the desired shape using a focused ion beam.

[0134] If an abnormality occurs in the foreign object detection process of the image processing computer 30, the control computer 22 initializes the position coordinates of the needle 18, moves the needle 18 to its initial position, and then moves the needle 18 to a location where there are no structures behind it. Furthermore, even after initializing the position coordinates of the needle 18, if an abnormality occurs in the foreign object detection process of the image processing computer 30, the control computer 22 determines that there is an abnormality such as deformation in the shape of the needle 18, displays a warning message on the screen, and terminates automatic sample sampling. Alternatively, it executes an automatic needle replacement sequence.

[0135] The control computer 22 may perform the needle trimming process (step S230) each time automatic sample sampling is performed. By performing the needle trimming process (step S230) periodically, the automatic sample sampling process can be stabilized. In particular, by removing deposits such as the deposition film DM2 and residue from the sample piece Q attached to the tip of the needle 18 that is close to the sample piece Q, the tip of the needle 18 can be exposed, thereby maintaining the fixing strength between the needle 18 and the sample piece Q. By performing the needle trimming process (step S230), the needle 18 can be used repeatedly without changing it to perform sampling of the sample piece Q, so that multiple sample pieces Q can be sampled consecutively using the same needle 18.

[0136] The needle trimming process (S230) can be performed at any time, not just during automated sample sampling, but also at the first time the needle 18 is replaced.

[0137] Next, we will explain how to set up the processing frame.

[0138] (Setting of processing frame by artificial intelligence (AI)) As described in the needle trimming process (S230), the image processing computer 30 uses the image data generated by irradiation with a focused ion beam (step S237) to perform image recognition (step S238) of the position of the tip of the needle 18, and then determines the area of ​​foreign matter attached to the needle 18 (step S240). The area of ​​foreign matter determined by the image processing computer 30 is transmitted to the control computer 22, which sets a processing frame (processing area) for the focused ion beam corresponding to the area of ​​foreign matter and performs sharpening of the tip of the needle 18 (step S241). Here, sharpening can be considered to include a removal process to remove foreign matter from the tip of the needle 18 and a shaping process to shape the tip of the needle 18 into a desired shape.

[0139] Figures 31 to 34 illustrate the processing frame for a focused ion beam. Figure 31 illustrates an example of a processing frame related to a comparative example. Figure 32 illustrates the problems with the processing frame related to the comparative example. Figure 33 illustrates an example of a processing frame related to an embodiment. Figure 34 illustrates an example of the setting of the processing frame related to an embodiment.

[0140] Figure 31 shows the tip of a needle 18 with a processing frame 40a set up in a comparative example. The processing frame 40a is a rectangular-shaped processing frame that assumes an ideal tip position C by linearly approximating the part of the needle 18 from the tip to the base end. In this example, processing frames 40a are set on both the upper and lower sides of the needle 18. By performing etching using a focused ion beam within the range of the processing frame 40a (the area inside the frame), foreign matter FB adhering to the upper and lower sides of the needle 18 can be removed.

[0141] Figure 32 shows a case where foreign matter FB adheres to the tip of the needle 18. In this example, the foreign matter FB is located between the processing frames 40a set on the upper and lower sides of the needle 18. Therefore, even if etching is performed using a focused ion beam, the foreign matter FB is not located within the range of the processing frames 40a (the area inside the frame), and thus there is a problem in that the foreign matter FB cannot be removed. In other words, it is not possible to remove the foreign matter FB that has adhered to the very tip of the needle 18.

[0142] Figure 33 shows an example of a processing frame 40 according to the embodiment. As shown in Figure 33, in this example, the processing frame 40 is a rectangular-shaped processing frame set to surround the foreign matter FB. Therefore, when etching is performed using a focused ion beam, the foreign matter FB is located within the range of the processing frame 40 (the area inside the frame), and unlike the case of processing frame 40a described in Figure 32, the foreign matter FB can be removed. In other words, the foreign matter FB attached to the tip of the needle 18 can be removed.

[0143] Figure 34 shows examples of how the processing frame is set up according to the embodiment. (A) shows an example image of a needle 18 with foreign matter FB attached to its tip. (B) shows an example where the processing frame 40, which has the same shape as the foreign matter FB, is set around the foreign matter FB. (C) shows an example where the rectangular processing frame 40 is set around the foreign matter FB so as to surround it. In this way, the shape of the processing frame 40 can be set to any shape, so it can be selected according to the shape of the foreign matter.

[0144] Unlike the rectangular machining frame 40a shown in Figures 31 and 32, which assumes an ideal tip position C by linearly approximating the area from the tip to the base of the needle 18, the machining frame 40 set by the control computer 22 is set in the area where foreign matter adheres, as shown in Figures 33 and 34.

[0145] Therefore, the processing frame 40 can be installed even for foreign matter FB adhering to the gap between the ideal tip position C and the actual tip of the needle 18. As a result, it is possible to remove the foreign matter FB adhering to the gap between the ideal tip position C and the actual tip of the needle 18. Through this foreign matter removal and sharpening process of the needle 18, the tip of the needle 18 can be restored to a state where the body is exposed and free of foreign matter. This makes it possible to process the needle 18 into a shape that is optimal for transferring the sample piece Q.

[0146] Furthermore, if the image processing computer 30 determines, for example, in step S238, that there is no region corresponding to foreign matter FB in the image of the needle 18, the control computer 22 does not set the processing frame 40 with the focused ion beam, and the needle cleaning process (S240, S241) is skipped.

[0147] Furthermore, if the image processing computer 30 determines, for example, in step S238, that it cannot identify the shape of the needle 18 in the image of the needle 18 and that the needle 18 is deformed, the control computer 22 does not set the processing frame 40 with the focused ion beam, and the needle cleaning process (S240, S241) therein is skipped. Alternatively, the automatic needle replacement process may be performed thereafter, or if it is determined that the needle 18 is deformed, the needle 18 may be reshaped to correct the deformation. Reshaping the deformed needle 18 results in a needle with an optimal shape for transferring the sample piece. Figure 35 illustrates the reshaping of the needle 18. As shown in Figure 35, the initial needle 18a may deform to look like needle 18b after use or after being significantly worn down for some reason. That is, the tip of the initial needle 18a is sharply pointed like the corner of a triangle, while the tip of the deformed needle 18b has a rectangular shape. Therefore, artificial intelligence (AI) is used to recognize the shape of needle 18a, and the control computer 22 places a processing frame 40 (in this case, two processing frames 40) on needle 18b and reshapes the needle by sharpening it. In other words, based on the judgment of the machine learning model, the control computer 22 irradiates the needle 18b with a charged particle beam from the charged particle beam irradiation optical system 14 and shapes the needle 18b based on the processing frame 40. By shaping the tip of needle 18b, for example, a needle 18c with a sharpened tip can be obtained. This makes it possible to obtain a needle 18c with an optimal shape for transferring a sample piece. Although Figure 35 illustrates the case where the shape of the needle tip is sharply pointed like the corner of a triangle, it is also possible to set the processing frame 40 so that a prismatic shape is formed at the tip of the needle. Even if the tip of the needle is prismatic, it can still be said to be an optimal shape for transferring a sample piece.

[0148] Furthermore, the image processing computer 30 also recognizes the size of the foreign matter FB attached to the needle 18 and transmits the result of the size determination of the attached foreign matter FB to the control computer 22. Based on the information on the size of the foreign matter received from the image processing computer 30, the control computer 22 controls the focused ion beam irradiation optical system 14, which is the irradiation unit for the focused ion beam, and changes the beam conditions of the focused ion beam irradiating the foreign matter FB. In other words, the amount of current of the focused ion beam irradiating the foreign matter FB may be changed. Large foreign matter FB can be removed with a large amount of focused ion beam to shorten the processing time. Small foreign matter FB can be accurately removed with a small amount of focused ion beam with low aberrations.

[0149] Furthermore, when the control computer 22 places the processing frame 40 in the area that the image processing computer 30 has determined to be the area of ​​foreign matter FB, it may use the coordinate information of the area of ​​foreign matter FB, or it may place the processing frame 40 based on the image of the area that has been determined to be the area of ​​foreign matter FB.

[0150] The control computer 22 rotates the needle 18 around its central axis using the rotation mechanism of the needle drive mechanism 19, performing etching at multiple different specific rotation positions to shape the needle 18 into the desired form. The operator can select whether or not to perform etching at multiple different specific rotation positions of the needle 18 using the rotation mechanism of the needle drive mechanism 19.

[0151] Furthermore, in addition to the processing frame 40 installed by the control computer 22 in the area determined by the image processing computer 30 to be a foreign object FB, the conventional method of detecting the edge of the needle 18 may also be used in combination.

[0152] Furthermore, if the area of ​​the region identified by the image processing computer 30 as a foreign object FB is smaller than a specified value, the needle cleaning process (S240, S241) can be skipped.

[0153] Furthermore, the operator can choose whether or not to perform cleaning on the area identified by the image processing computer 30 as containing foreign matter FB, depending on the type and importance of the next sample to be sampled.

[0154] Furthermore, since the shape of needle 18 gradually changes through the processes of bonding and cutting the sample, the learning model can be selected according to the number of times it has been used (see Figure 29).

[0155] The completion of the needle cleaning process may be determined by the processing time, or it may be determined again by artificial intelligence (AI) using a machine learning model M. When using AI for determination, the needle trimming process (step S230) is terminated if no area of ​​foreign matter FB is detected, or if the area detected as having foreign matter FB is smaller than a specified size. Alternatively, a learning model of an ideally shaped needle 18 free of foreign matter FB may be prepared for termination determination (see Y81 in Figure 30). Furthermore, the AI ​​determination method and the determination method using the learning model of an ideally shaped needle 18 free of foreign matter FB may be used in combination.

[0156] If the cleaning of the needle 18 to which the foreign substance FB is attached is successful, the image data may be imported into the image processing computer 30 for self-training to be used as training data. In this case, whether or not sampling with the needle 18 after cleaning was successful will be added as a criterion for whether or not it can be used as training data.

[0157] Furthermore, a self-trained model containing training data can also be used as a training model for another device.

[0158] The computer 22 can easily recognize the needle 18 by pattern matching when driving the needle 18 in three-dimensional space, by matching the tip shape of the needle 18 to a predetermined ideal shape, and can accurately detect the position of the needle 18 in three-dimensional space.

[0159] (modified version) In the above description of the embodiment, techniques for controlling the initial setup process, sample pickup process, sample mounting process, and needle trimming process using a machine learning model M were described, but the invention is not limited thereto. At least, by controlling the needle trimming process using a machine learning model M, a charged particle beam apparatus that enables the removal of foreign matter FB adhering to the leading edge of the needle 18 can be provided.

[0160] In this embodiment, an example has been described in which the training data is a pair of a training image and information indicating the position of an object within the training image, but it is not limited to this. In addition to the training image, the training data may also include parameter information such as the type of sample, scan parameters (such as the acceleration voltage of the focused ion beam irradiation optical system 14 and the electron beam irradiation optical system 15), the number of times the needle 18 has been used since cleaning, and whether or not foreign matter is attached to the tip of the needle 18.

[0161] In this case, the machine learning model M is generated by performing machine learning based on the training images and parameter information. The determination unit 304 also obtains parameter information in addition to image data of SIM images and SEM images from the control computer 22, and determines the position of the object in the image based on the image data, parameter information, and the machine learning model M.

[0162] Furthermore, the parameter information may also include the directional information mentioned above. If the training data includes directional information, the relationship between the object and the direction from which the object is viewed (the direction relative to the sample stage 12) is learned and a machine learning model M is generated, so the determination unit 304 does not need to use directional information to determine the position of the object.

[0163] As described above, the computer (in this embodiment, the control computer 22) controls the position of the second object (in this embodiment, the columnar portion 44, the needle 18, and the sample piece Q) based on the results of the image processing computer 30 determining the position of the second object (in this embodiment, the columnar portion 44, the needle 18, and the sample piece Q) based on the machine learning model (in this embodiment, the machine learning model M) and second information including the second image (in this embodiment, the SIM or SEM image of the columnar portion 44, the needle 18, and the sample piece Q). The image processing computer 30 and the control computer 22 may be integrated and provided together in the charged particle beam apparatus 10.

[0164] The present invention has been described in detail above based on examples, but it goes without saying that the present invention is not limited to the above embodiments and examples, and can be modified in various ways. [Explanation of symbols]

[0165] 10: Charged particle beam apparatus, 18: Needle, 22: Control computer, 30: Image processing computer, M: Machine learning model, FB: Foreign object

Claims

1. A charged particle beam apparatus that repeatedly transports a sample piece using a needle and cleans the needle as needed, A charged particle beam irradiation optical system that irradiates with a charged particle beam, A sample stage on which the sample is placed and moved, A sample piece transfer means comprising a needle for holding and transporting the sample piece to be separated and extracted from the sample, and a needle drive mechanism for driving the needle, A holder fixing base that holds the sample piece holder to which the sample piece is transferred, A machine learning model that has learned information including an image of a foreign object attached to the needle and an image of the portion of the needle containing the foreign object, The system includes a computer that, based on the judgment of the machine learning model, determines the area of ​​foreign matter attached to the needle, irradiates the area with the charged particle beam, and controls the charged particle beam irradiation optical system to process the foreign matter, The thickness of the tip of the needle changes due to the cleaning. The computer uses the judgment of another machine learning model that has learned information including images of foreign matter attached to the needle whose tip thickness has changed according to the cleaning frequency, and images of the portion of the image containing the foreign matter. Charged particle beam device.

2. The image of the needle learned by the machine learning model includes an image showing the shape of the needle. The charged particle beam apparatus according to claim 1.

3. Based on the machine learning model's judgment, the computer skips irradiation with the charged particle beam if it determines that there is no area of ​​foreign matter attached to the needle. The charged particle beam apparatus according to claim 1.

4. Based on the machine learning model's judgment, the computer determines that an area of ​​foreign matter is attached to the needle, and if the area of ​​the area is smaller than a specified value, it skips the irradiation of the charged particle beam. The charged particle beam apparatus according to claim 3.

5. Based on the judgment of the machine learning model, the computer changes the beam conditions of the charged particle beam based on the area of ​​the region determined to be a foreign substance attached to the needle. The charged particle beam apparatus according to claim 1.

6. If the computer determines, based on the machine learning model's judgment, that it cannot identify the shape of the needle and that the needle is deformed, it will skip the irradiation of the charged particle beam without setting a processing area, or perform the step of replacing the needle. The charged particle beam apparatus according to claim 1.

7. The process of extracting the sample piece from the sample using the needle and transferring it to the sample piece holder, The steps include moving the needle to the irradiation position of the charged particle beam, A step of irradiating the needle with the charged particle beam to acquire an image, The process involves processing the aforementioned image using the machine learning model to determine the processing area from which foreign matter adhering to the needle is removed. The process includes irradiating the processing area with the charged particle beam to remove foreign matter adhering to the needle, The charged particle beam apparatus according to claim 1.

Images