Focused ion beam apparatus

Abstract

The objective of the present disclosure is to provide a focused ion beam device that is capable of obtaining information pertaining to a cut section before cut processing for a sample is completed. The focused ion beam device according to the present disclosure trains, in advance, the correspondence relationship between the three-dimensional structure of a sample and a processing condition by using a learning apparatus, and inputs the processing condition and a position on the three-dimensional structure to the learning apparatus, thereby acquiring a predicted result of cross-section information pertaining to the sample from the learning apparatus (see fig. 2).

Description

Technical Field

[0001] The present disclosure relates to a focused ion beam apparatus that irradiates a sample with a focused ion beam.

Background Art

[0002] A focused ion beam (FIB) apparatus processes a sample by irradiating the sample with a focused ion beam. Further, there is also an apparatus that has a function of obtaining an observation image of a sample by detecting secondary particles generated from the sample by irradiating the sample with, for example, an electron beam (FIB-SEM: Scanning Electron Microscope). For example, a sample can be cut by an ion beam, and an observation image of the cut cross section can be acquired by an electron beam.

[0003] The following Patent Document 1 describes a technique for performing cross-section processing of a sample using an ion beam. This document aims to "realize an automatic cross-section processing observation apparatus that terminates cross-section processing observation when a desired observation target is exposed on the cross section." As a solution, it describes a technique of "having a control unit 11 that repeatedly executes a process consisting of slice processing by an ion beam 9 and acquisition of a SIM image by secondary electrons emitted from the cross section formed by the processing, and the control unit 11 divides an observation image into a plurality of areas, and when a change occurs between an image of one area among the plurality of areas and an image of an area corresponding to one area of an observation image of another cross section acquired in the process, the process is terminated." (See the abstract).

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] Conventional focused ion beam systems, such as those described in Patent Document 1, require the actual cutting process to be performed and the cutting surface to be irradiated with an electron beam for observation in order to obtain an observation image of the cut surface of a sample. However, in actual operation, there are cases where it is desirable to observe the cut surface before the cutting process is completed. Such pre-observation has been difficult with conventional focused ion beam systems.

[0006] This disclosure has been made in view of the above-mentioned problems, and aims to provide a focused ion beam apparatus that can obtain information about the cut surface before completing the cutting process on the sample. [Means for solving the problem]

[0007] The focused ion beam apparatus according to this disclosure pre-learns the correspondence between the three-dimensional structure of a sample and the processing conditions using a learning device, and obtains a prediction result of the cross-sectional information of the sample from the learning device by inputting the position on the three-dimensional structure and the processing conditions to the learning device. [Effects of the Invention]

[0008] The focused ion beam apparatus described herein allows information about the cut surface to be obtained before the cutting process on the sample is completed. Further issues, configurations, and advantages of this disclosure will become apparent from the following description of the embodiments. [Brief explanation of the drawing]

[0009] [Figure 1] This is a diagram showing the configuration of a focused ion beam apparatus 1 according to Embodiment 1. [Figure 2] This is a block diagram showing the configuration of the learning device 161 provided by the control unit 16. [Figure 3] This is a schematic diagram illustrating a method for determining whether or not the end of the cutting process has been reached. [Figure 4] This diagram illustrates the procedure for updating a cross-sectional image to one that more closely resembles an actual cutting process. [Figure 5]This diagram illustrates an alternative method for determining whether or not the cutting process has reached its end. [Modes for carrying out the invention]

[0010] <Embodiment 1> Figure 1 is a configuration diagram of a focused ion beam apparatus 1 according to Embodiment 1 of the present disclosure. The focused ion beam apparatus 1 is an apparatus that processes a sample 2 by irradiating the sample 2 with a focused ion beam. The sample 2 is, for example, a device in which semiconductor elements are arranged in three dimensions. The focused ion beam apparatus 1 comprises an ion beam tube 11 (irradiation unit), an electron beam tube 12 (irradiation unit), a sample chamber 13, a stage 14, a detector 15, and a control unit 16.

[0011] The ion beam tube 11 irradiates the sample 2 with a focused ion beam. The electron beam tube 12 irradiates the sample 2 with an electron beam. The sample chamber 13 houses the sample 2. The stage 14 can be used to move the sample 2 on top of it. The detector 15 detects secondary particles generated from the sample 2 by irradiating the sample 2 with a focused ion beam or electron beam, and outputs a detection signal representing its intensity. The control unit 16 controls each part of the focused ion beam apparatus 1.

[0012] Figure 2 is a block diagram showing the configuration of the learner 161 provided in the control unit 16. The learner 161 can be constructed by pre-learning the correspondence between design data 201 describing the three-dimensional structure of the sample 2, processing conditions 202 for the sample 2, and the predicted result of the cross-sectional view of the sample 2 resulting from the processing conditions 202 (predicted cross-sectional image 203) using machine learning.

[0013] Design data 201 describes the material at each three-dimensional position of sample 2, the shape / type of the semiconductor element placed at that position, and so on. Design data 201 is data describing the design structure of sample 2, created, for example, by a CAD (Computer Aided Design) system.

[0014] Processing conditions 202 are used to determine how far the cutting process progresses (i.e., the processing speed) when a cutting process is performed on the sample 2 using a focused ion beam, for example. Examples of processing conditions 202 include: the type of gas used when performing the processing; the acceleration voltage of the focused ion beam; the type of focused ion; the probe current value of the focused ion; the irradiation time of the focused ion beam at the same position; the tilt correction angle of the focused ion beam (the tilt angle when tilting the stage 14 or the beam to uniformly apply the beam), and the mounting angle of the sample.

[0015] The learning process can be carried out as follows: The control unit 16 specifies a three-dimensional position on the design data 201 and inputs the machining conditions 202 to the learner 161, thereby obtaining a predicted cross-sectional image 203 from the learner 161. If an incorrect predicted cross-sectional image 203 is obtained (for example, a cross-sectional image of a different position, a cross-sectional image of a depth that cannot be reached by the input machining conditions 202, etc.), the correct predicted cross-sectional image 203 is fed back to the learner 161. By repeating this process, the learner 161 can be trained.

[0016] The learner 161 can be composed of, for example, a machine learning model constructed using a neural network and trained model data describing its learning results. Alternatively, a learner constructed using other appropriate machine learning methods can be used. This data can be stored in the memory device provided by the control unit 16.

[0017] The control unit 16 obtains a predicted cross-sectional image 203 from the learner 161 by specifying the three-dimensional position on the design data 201 and inputting the processing conditions 202 to the learner 161 after it has completed learning. This makes it possible to obtain a prediction result for the cross-sectional view assuming that the sample 2 has been cut, without actually performing the cutting process. The control unit 16 outputs the prediction result in an appropriate output format, such as screen display or data output.

[0018] The advantages of obtaining a cross-sectional view without actually performing cutting processing are as follows. For example, depending on the skill level of the operator, there is a possibility of performing cutting processing using processing conditions that are not necessarily appropriate. By inputting the processing conditions 202 into the learning device 161, the cross-sectional surface obtained under those processing conditions 202 can be predicted. In other words, it is possible to predict, without actual processing, whether the assumed cutting processing can be appropriately performed under the processing conditions 202. If the predicted cross-sectional image 203 is not as expected, the processing conditions can be modified and an appropriate processing step can be performed again under appropriate processing conditions.

[0019] In the present embodiment, instead of or in combination with the design data 201, data describing the three-dimensional structure obtained by processing the sample 2 using a device capable of observing the three-dimensional structure of the sample 2 (e.g., a charged particle beam device capable of three-dimensional observation) may be used in the learning process of the learning device 161.

[0020] In the present embodiment, instead of the predicted cross-sectional image 203, the learning device 161 may learn other cross-sectional information representing the cross-sectional shape. For example, it is conceivable to learn the feature amount of the cross-sectional shape. As an example of the feature amount, information that can represent unique information for each cross-section, such as the type, number, size, position, shape, etc. of semiconductor elements arranged on the cross-section, can be used.

[0021] <Embodiment 2> In Embodiment 1, obtaining the predicted cross-sectional image 203 from the learning device 161 for which learning has been completed was described. In addition to this, the learning result of the learning device 161 can also be updated using a cross-sectional image actually obtained by cutting the sample 2.

[0022] First, a focused ion beam is irradiated onto sample 2 to perform a cutting process. Next, an electron beam is irradiated onto the cut area, and secondary particles are detected by detector 15. The control unit 16 uses the detection signal from detector 15 to generate an observation image (cross-sectional image) of the cut area. The control unit 16 inputs the generated cross-sectional image and the processing conditions during the cutting process into the learner 161, thereby relearning the correspondence between them. Based on the results of the relearning, the learner 161 updates the trained model data.

[0023] <Embodiment 3> The predicted cross-sectional image 203 described in Embodiments 1 and 2 can also be used to determine whether or not the end of the cutting process has been reached. In particular, when cutting a fine sample 2, it has conventionally been common practice to determine whether or not the end of the process has been reached based on the operator's skill. Embodiment 3 of this disclosure describes an example of a configuration that automatically performs such end determination. The configuration of the focused ion beam apparatus 1 is the same as in Embodiments 1 and 2.

[0024] Figure 3 is a schematic diagram illustrating a method for determining whether or not the end of the cutting process has been reached. As described in Embodiment 1, the control unit 16 can obtain a predicted cross-sectional image 203 from the learner 161. The predicted cross-sectional image 203 is a prediction of the cross-sectional view obtained by completing the cutting of the sample 2 using the current processing conditions. In parallel with this, the control unit 16 acquires an actual cross-sectional image 301 obtained by actually processing the sample 2 by irradiating the sample 2 with an electron beam.

[0025] The control unit 16 compares the predicted cross-sectional image 203 with the actual cross-sectional image 301. If the two images match, the control unit 16 can determine that the actual cross-sectional image 301 is the cross-sectional image at the end of the cutting process, and thus can determine that the end has been reached. The comparison of the two images can be performed, for example, by obtaining a matching score using an appropriate pattern matching method, and if the score is above a threshold, the two images are considered to match.

[0026] Figure 4 illustrates the procedure for updating a cross-sectional image to an image that more closely resembles an actual cutting process. The learner 162 is configured to receive the cross-sectional image 401 of sample 2 as input and output an image that more closely resembles an actual cutting process. The learner 162 can be trained by repeatedly training it with the cross-sectional image 401 and images of actual cutting processes. Like the learner 161, the learner 162 can be composed of any machine learning model and trained model data that describes its training results.

[0027] The control unit 16 inputs the cross-sectional image obtained by actually processing the sample 2 into the learner 162, thereby obtaining an image 402 that more closely resembles the actual cutting process. This allows for updating the predicted cross-sectional image used in the method for determining whether or not the end of the cutting process shown in Figure 3 has been reached.

[0028] Figure 5 illustrates an alternative method for determining whether the cutting process has reached its end. The learner 161 can output a predicted cross-sectional image 203 assuming that the cutting process is complete, or it can output a predicted cross-sectional image during the cutting process. For example, by inputting the material of the sample, the etching rate of the material, and the irradiation time for each irradiation position of the focused ion beam as processing conditions 202 to the learner 161, and pre-training the learner 161 with the predicted cross-sectional image 203 at that time, the learner 161 can obtain a predicted cross-sectional image 203 during the processing process.

[0029] The control unit 16 inputs the predicted cross-sectional image 203 during the machining process to a learner 163 configured in the same way as the learner 162, thereby obtaining a determination result 501 from the learner 163 indicating whether or not the cutting process has reached its end. The learner 163 can be configured in the same way as the learner 162, or it can be configured to accept equivalent cross-sectional information (e.g., cross-sectional feature quantities) as input instead of the predicted cross-sectional image 203.

[0030] According to the judgment method shown in Figure 5, by checking the progress of the machining process during the machining process, any machining abnormalities can be detected before the end of the machining process is reached.

[0031] <Embodiment 4> In the above embodiments, the learner 161 may perform the learning process while taking into account the processing speed of the sample 2. For example, the processing speed will differ depending on the relationship between the etching rate of the material of the sample 2 described in the design data 201 and the acceleration voltage, ion beam current, etc., specified by the processing conditions 202. By performing the learning process after specifying the irradiation time of the focused ion beam to the same irradiation position as the processing condition 202, it is possible to predict how far the processing will progress based on the processing speed. As a result, a predicted cross-sectional image 203 at any intermediate processing point can be obtained based on the processing speed of the sample 2.

[0032] Furthermore, the processing speed of sample 2 may also vary depending on the beam profile (beam shape) of the focused ion beam. The learner 161 can predict how far the processing will progress based on the processing speed determined by the beam profile by specifying the beam profile as the processing condition 202 and then performing the learning process.

[0033] <Regarding variations of this disclosure> This disclosure is not limited to the embodiments described above, but includes various modifications. For example, the embodiments described above are described in detail for the purpose of illustrating this disclosure, and do not necessarily have to include all the configurations described. Furthermore, parts of one embodiment can be replaced with the configurations of another embodiment. Furthermore, configurations of other embodiments can be added to the configuration of one embodiment. Furthermore, parts of the configuration of each embodiment can be added, deleted, or replaced with parts of the configurations of other embodiments.

[0034] In the embodiments described above, the control unit 16 can be configured by hardware such as a circuit device that implements its function, or by a computing device such as a CPU (Central Processing Unit) executing software that implements its function.

[0035] In the embodiments described above, an electron beam was used as a means of observing the sample 2, but other charged particle beams may be used as long as an observation image can be obtained. For example, an observation image may be obtained by irradiating the sample with an ion beam different from the ion beam used for processing. [Explanation of symbols]

[0036] 1: Focused ion beam apparatus 11: Ion beam tube 12: Electron beam tube 13: Sample Room 14: Stage 15: Detector 16: Control Unit 161: Learning device

Claims

1. A focused ion beam apparatus that irradiates a sample with a focused ion beam, An irradiation unit for processing the sample by irradiating the sample with the focused ion beam. Control unit for controlling the irradiation unit, Equipped with, The control unit is configured to obtain a prediction result of cross-sectional information representing the cross-sectional shape of the sample by inputting the three-dimensional structure and processing conditions of the sample into a learner that has been trained by machine learning, and then outputting the prediction result. The control unit inputs the position on the three-dimensional structure and the processing conditions to the learning device, thereby obtaining the prediction result at the position from the learning device. A focused ion beam apparatus characterized by the following features.

2. The learning device is configured to perform machine learning using the three-dimensional data describing the three-dimensional structure and the processing conditions. The aforementioned three-dimensional data describes at least one of the following: the material at the three-dimensional position of the sample, or the elements arranged at the three-dimensional position of the sample. The learning device is configured to learn about at least one of the material or element at the three-dimensional position of the sample by inputting the three-dimensional data and the processing conditions and learning by machine learning. The control unit obtains from the learner, as a prediction result, at least one of the material or element at the three-dimensional position of the sample learned by the learner. The focused ion beam apparatus according to claim 1, characterized by its features.

3. The irradiation unit irradiates the sample with a charged particle beam used to acquire a cross-sectional image of the sample, The control unit acquires the cross-sectional image by detecting secondary particles generated from the sample by irradiating the sample with the charged particle beam, The control unit updates the learning results of the learning device by having the learning device learn the acquired cross-sectional image and processing conditions again. The focused ion beam apparatus according to claim 1, characterized by its features.

4. The irradiation unit irradiates the sample with a charged particle beam used to acquire a cross-sectional image of the sample, The control unit acquires the cross-sectional image by detecting secondary particles generated from the sample by irradiating the sample with the charged particle beam, The control unit determines the progress of the machining by comparing the cross-sectional image with the prediction result. The focused ion beam apparatus according to claim 1, characterized by its features.

5. The control unit acquires the prediction result for the time when the processing is completed, If the control unit determines that the acquired prediction result matches the cross-sectional image, it controls the irradiation unit to terminate the processing. The focused ion beam apparatus according to claim 4.

6. The control unit compares the cross-sectional image and the prediction result by performing pattern matching between the cross-sectional image and the prediction result, The control unit determines that the acquired prediction result matches the cross-sectional image if the matching score in the pattern matching is equal to or greater than a threshold. The focused ion beam apparatus according to claim 5, characterized in that it is a feature of the present invention.

7. The control unit determines whether the processing is complete by inputting the cross-sectional image into a second learner, which has previously learned the correspondence between the cross-sectional image and whether the processing is complete or not through machine learning. The focused ion beam apparatus according to claim 5, characterized in that it is a feature of the present invention.

8. The control unit determines whether the processing has been completed by inputting the prediction result into a third learner, which has previously learned the correspondence between the prediction result and whether the processing has been completed through machine learning. The focused ion beam apparatus according to claim 5, characterized in that it is a feature of the present invention.

9. The aforementioned learning device is The material of the sample, the processing conditions, and the processing speed of the sample, The correspondence between them has been learned in advance using machine learning. The control unit inputs the material and processing conditions to the learning device, thereby acquiring the cross-sectional information reflecting the processing speed from the learning device. The focused ion beam apparatus according to claim 1, characterized by its features.

10. The learning device has previously learned the correspondence between the beam shape of the focused ion beam and the processing speed of the sample through machine learning. The control unit inputs the beam shape to the learner, thereby acquiring the cross-sectional information reflecting the processing speed from the learner. The focused ion beam apparatus according to claim 1, characterized by its features.

11. The aforementioned three-dimensional data is design data describing the results of designing the structure of the sample. The focused ion beam apparatus according to claim 2, characterized by its features.

12. The aforementioned three-dimensional data is obtained by irradiating the sample or another sample having the same structure as the aforementioned sample with a charged particle beam and analyzing its three-dimensional structure using a charged particle beam apparatus. The focused ion beam apparatus according to claim 2, characterized by its features.

13. The aforementioned processing conditions are: The type of gas used when performing the processing, the acceleration voltage of the focused ion beam, the type of focused ion, the probe current of the focused ion beam, the irradiation time of the focused ion beam at the same position, the tilt correction angle of the focused ion beam, or the mounting angle of the sample. It is at least one of the following. The focused ion beam apparatus according to claim 1, characterized by its features.

Images