Computer system, sample observation method, and program

Abstract

The invention relates to a technology of a sample observation device and the like, and provides a technology capable of realizing appropriate image quality quantification. A computer system in a sample observation device for observing a sample, the computer system having one or more processors and one or more memories, the processors performing an operation of determining a plurality of evaluation partial regions (Si) for an image (201) obtained by imaging the sample, and performing an operation of evaluating the evaluation partial regions (Si) on the basis of the plurality of evaluation partial regions (Si), a plurality of image quality evaluation values (Ei) for evaluating a plurality of different image qualities are calculated by calculating an image quality evaluation value (Ei) for each evaluation partial region (Si).

Description

Technical Field

[0001] The present disclosure relates to a specimen observation technique, for example, to a specimen observation apparatus having a function of observing defects, abnormalities, etc. (sometimes collectively referred to as defects) and circuit patterns, etc. in a specimen such as a semiconductor wafer. Background Art

[0002] In the manufacture of semiconductor wafers, in order to ensure profitability, it is particularly important to quickly start the manufacturing process and shift to a mass production system with a high yield rate at an early stage. For this purpose, various inspection devices, observation devices, measurement devices, etc. are introduced into the production line. A semiconductor wafer as a specimen is, for example, subjected to a defect inspection in an inspection device. The inspection device outputs coordinate information indicating the position / part of a defect in the specimen as defect candidate coordinates. The output defect candidate coordinates are supplied to a specimen observation device, which is an observation device for observing defects.

[0003] The specimen observation device takes a high-resolution photograph of the defect candidate on the wafer surface based on the defect candidate coordinates and outputs a photographed image. As the specimen observation device, an observation device using an optical microscope and an observation device using a scanning electron microscope (SEM) are widely used. In other words, these devices are charged particle beam devices having a mechanism for irradiating a charged particle beam and have functions of photographing, measuring, observing, evaluating, inspecting, etc. (sometimes collectively referred to as observation) of a specimen.

[0004] Observation operations using the observation device are expected to be automated in the mass production line of semiconductor wafers. For the automation of the mass production line, the observation device sometimes has a function of automatically collecting images of defect positions in a specimen, that is, automatic defect review (ADR), in other words, and a function of automatically classifying the collected defect images, that is, automatic defect classification (ADC). Through these functions, classified defect images can be obtained automatically.

[0005] The defect candidate coordinates output by the inspection device contain errors. There is a difference between the coordinate system of the defect candidate coordinates in the inspection device and the coordinate system in the observation device. Therefore, due to the deviation caused by this difference, sometimes the defect candidate cannot be found even when photographing the defect candidate coordinates using the observation device. Therefore, in ADR, the observation device takes a wide-field photograph centered on the defect candidate coordinates and detects the defect candidate part from the image obtained by this photograph. The observation device takes a high-magnification and high-resolution photograph of the obtained defect candidate part and outputs the photographed image as an observation image.

[0006] In addition, regarding the observation image obtained by photographing a semiconductor wafer, sometimes for the purpose of improving visual recognition, a process of improving the image quality (in other words, high-definition processing) using an image quality improvement engine is performed.

[0007] As a prior art example, Japanese Unexamined Patent Application Publication No. 2012-142299 (Patent Document 1) can be cited. In Patent Document 1, a technique is described in which, for an image of a specimen obtained by photographing using a scanning type charged particle microscope, a high-quality image with reduced noise components is obtained to improve the accuracy of image processing.

[0008] Prior Art Documents

[0009] Patent Documents

[0010] Patent Document 1: Japanese Unexamined Patent Application Publication No. 2012-142299 Summary of the Invention

[0011] Problems to be Solved by the Invention

[0012] In Patent Document 1, a technical example of a process for improving the image quality of an image obtained by photographing a semiconductor wafer is described. In Patent Document 1, the following is described: a degradation function of an image is calculated based on photographing conditions, specimen information, etc., and a high-quality image is obtained through image restoration processing using the degradation function.

[0013] Conventionally, the quality of an observation image obtained by using a photographing function or an image quality improvement engine has been judged by visual evaluation by a user. Therefore, there are problems such as an increase in the user's operation cost and a blur in human judgment criteria.

[0014] Therefore, in a system such as an observation device, if the image quality of an image obtained by photographing a wafer can be quantified, these problems can be solved.

[0015] However, when quantifying image quality, the required image quality, in other words, the evaluation value, parameter value, etc. of the image quality, is not limited to one, and sometimes consists of multiple ones. For example, regarding the image quality required for appropriate observation or inspection of an image showing the circuit pattern of a wafer, multiple elements and parameter values such as sharpness, visual recognition of defects, and noise amount (in other words, noise suppression degree) are considered.

[0016] Conventional image quality quantification methods do not consider such multiple elements and parameter values of the image quality for an image photographed by a microscope such as a SEM (in other words, a charged particle beam device, a photographing device). In an image obtained by photographing a wafer, it is considered that there are cases where the appropriate region for calculating the image quality evaluation value also varies according to such elements and parameter values of the image quality.

[0017] An object of the present disclosure is to provide a technique capable of achieving appropriate quantification of image quality for the above-described sample observation devices and other technologies.

[0018] Means for Solving the Problem

[0019] A representative embodiment in the present disclosure has the following structure. A computer system according to one embodiment is a computer system in a sample observation device for observing a sample, the computer system including one or more processors and one or more memories, and the processor performs the following operations: for an image obtained by photographing the sample, determine a plurality of evaluation partial regions; based on the plurality of evaluation partial regions, calculate an image quality evaluation value for each evaluation partial region, thereby calculating a plurality of image quality evaluation values for evaluating a plurality of different image qualities.

[0020] Advantageous Effects of the Invention

[0021] According to a representative embodiment in the present disclosure, for the above-described sample observation devices and other technologies, appropriate quantification of image quality can be achieved. Other problems, structures, effects, etc. are also presented in the mode for carrying out the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 FIG. is a diagram showing the structure of a system including the sample observation device of Embodiment 1.

[0023] Figure 2 FIG. is a diagram showing the processing flow and functional block structure in the sample observation device and method of Embodiment 1.

[0024] Figure 3 FIG. is a diagram showing a data example of image quality evaluation value information and an image quality evaluation result in Embodiment 1.

[0025] Figure 4 FIG. is a diagram showing an example of an image quality evaluation value in Embodiment 1.

[0026] Figure 5 FIG. is a diagram showing an example of an image quality evaluation value in Embodiment 1.

[0027] Figure 6 FIG. is a diagram showing Step S6 of the setting in Embodiment 1.

[0028] Figure 7 FIG. is a diagram showing a processing example of Steps S2 and S3 in Embodiment 1.

[0029] Figure 8 FIG. is a diagram showing an example of the correspondence between an evaluation partial region Si and an image quality evaluation value Ei in Embodiment 1.

[0030] Fig. 9It is a diagram showing the cases where the evaluation partial regions Si in Embodiment 1 overlap and do not overlap.

[0031] Fig.10 It is a diagram showing a detailed processing example of Step S2 in Embodiment 1.

[0032] Fig.11 It is a diagram showing a processing example of the classification using design data (Step S21) in Embodiment 1.

[0033] Fig.12 It is a diagram showing a processing example of the classification using shooting conditions (Step S21) in Embodiment 1.

[0034] Fig.13 It is a diagram showing a processing example of the determination of the evaluation partial region Si according to the classification (Step S22) in Embodiment 1.

[0035] Fig.14 It is a diagram showing an example of the evaluation partial region Si for each image quality evaluation value Ei in Embodiment 1.

[0036] Fig.15 It is a diagram showing an example of the classification criteria for the classification partial regions in Embodiment 1.

[0037] Fig.16 It is a diagram showing a processing example and a GUI example of the determination of the internal parameters in Step S63 in Embodiment 1.

[0038] Fig.17 It is a diagram showing a data example of the internal parameters of the image quality evaluation value Ei and the elemental image quality evaluation values in Embodiment 1.

[0039] Fig.18 It is a diagram showing a processing example of the calculation of the image quality evaluation value using the elemental image quality evaluation values in Embodiment 1.

[0040] Fig.19 It is a diagram showing an example of the machine learning model used in Step S3 in Embodiment 1.

[0041] Fig. 20 It is a diagram showing a processing example of calculating the comprehensive evaluation value C in Step S3 in Embodiment 1.

[0042] Fig.21 It is a diagram showing the flow and functional block structure including the image quality improvement processing in Embodiment 1.

[0043] Fig. 22 It is a diagram showing the determination of the parameter P of the image quality improvement processing using the regional target value in Embodiment 1.

[0044] Fig.23 This is a diagram showing the determination of parameter P for the image quality improvement process using the comprehensive evaluation value C in Embodiment 1.

[0045] Fig.24 This is a diagram showing the determination of parameter P when using a rule-based image quality improvement engine in the first embodiment.

[0046] Fig.25 This is a diagram showing the determination of parameter P when using a machine learning-based image quality improvement engine in Embodiment 1.

[0047] Fig.26 This is a diagram showing an example of a GUI screen related to the image quality improvement process in Embodiment 1.

[0048] Fig. 27 This is a diagram showing an example of a GUI screen related to the correction of the evaluation partial area in Embodiment 1.

[0049] Fig.28 This is a diagram showing a flowchart including steps such as the correction of the evaluation partial area in Embodiment 1.

[0050] Fig.29 This is a diagram showing an example of a GUI screen related to the evaluation step in Embodiment 1.

[0051] Fig.30 This is a diagram showing an example of the processing of the defect detection step in Embodiment 1.

[0052] Fig.31 This is a diagram showing an example of the processing of the shape measurement step in Embodiment 1.

[0053] Fig.32 This is a diagram showing a flowchart including the step of determining the shooting conditions using the image quality evaluation value in Embodiment 1.

[0054] Fig.33 This is a diagram showing an example of the processing of the step of monitoring the device using the image quality evaluation value in Embodiment 1. Detailed Embodiment

[0055] Hereinafter, embodiments of the present disclosure will be described in detail with reference to the drawings. In the drawings, the same reference numerals are generally assigned to the same parts, and repeated descriptions are omitted. In the drawings, for ease of understanding of the invention, the representation of the components sometimes does not show the actual position, size, shape, range, etc.

[0056] In the description, when describing program-based processing, sometimes the description is made with the program, function, processing unit, etc. as the main body. However, as the main body of the hardware related to them is a processor, or a controller, device, computer, system, etc. composed of the processor and the like. The computer executes the processing according to the program read into the memory while appropriately using resources such as the memory and communication interface by the processor. Thereby, a specified function, processing unit, etc. are realized. The processor is composed of semiconductor devices such as CPU / MPU, GPU, etc., for example. The processing is not limited to software program processing and can also be installed by a dedicated circuit. The dedicated circuit can apply FPGA, ASIC, CPLD, etc.

[0057] The program can be pre-installed in the target computer as data, or can be distributed to the target computer as data from a program source. The program source can be a program distribution server on a communication network or a non-transitory computer-readable storage medium, such as a memory card, disk. The program can also be composed of multiple modules. The computer system can also be composed of multiple devices. The computer system can also be composed of a client / server system, a cloud computing system, an IoT system, etc. Various data and information are composed of structures such as tables and lists, but are not limited thereto. The expressions of identification information, identifiers, IDs, names, numbers, etc. can be mutually replaced.

[0058] <Embodiment 1>

[0059] Use Figure 1 to Figure 33 The sample observation device and method of Embodiment 1 will be described. The sample observation method of Embodiment 1 is a method having steps executed by the sample observation device of Embodiment 1. Figure 1 The sample observation device 1 of Embodiment 1 shown in etc. is a device having the function of observing a semiconductor wafer 10 as a sample 10.

[0060] The sample observation device 1 of Embodiment 1 uses an optical microscope or SEM, etc. as a microscope or imaging device to capture an image of the wafer 10 as the sample 10. In Embodiment 1, the description is made in the case where the sample observation device 1 has an SEM2 ( Figure 1 ). The sample observation device 1 of Embodiment 1 has a computer system 3 (in other words, a controller, control device) for controlling the SEM2. In Embodiment 1, the description is made in the case where the computer system 3 calculates the image quality evaluation value Ei, etc.

[0061] The computer system 3 of the sample observation device 1 has a photographing mechanism for controlling the SEM2 to obtain an image 201 as a photographed image (in other words, an observed image) ( Figure 2 ), an observation mechanism for observing the observation image 201, such as a defect detection mechanism for detecting a defective part in the observation image. Each mechanism is composed of a control processor 102 ( Figure 1) etc.

[0062] Figure 2 The sample observation method of Embodiment 1 shown etc. has a step S1 of taking an image 201 of a semiconductor wafer 10 as a sample 10 by SEM2 ( Figure 1 ) and a step S5 of observing the semiconductor wafer 10 as a sample 10 using the taken image 201. The sample observation method of Embodiment 1 has a step S3 of calculating a plurality of image quality evaluation values Ei related to the taken image 201 for this observation. The sample observation method of Embodiment 1 has a step S2 of determining an evaluation partial region Si for calculating each image quality evaluation value Ei based on the taken image 201 by SEM2 ( Figure 1 ). Then, the sample observation method of Embodiment 1 uses this evaluation partial region Si, and in step S3, calculates a plurality of image quality evaluation values Ei as a plurality of elements constituting the image quality of the image 201. Based on the image quality evaluation values Ei obtained as a result of steps S2 and S3, an image 201 suitable for observation (step S5), that is, an image 201 with as high an image quality as possible, can be selected from the taken image 201 as an observation image.

[0063] In Embodiment 1, as an example of a sample observation device 1 and method, the case of observing / detecting defects etc. of a semiconductor wafer 10 as a sample 10 is described. Figure 2 Step S5 is a step of performing an observation action on the observation image as a prescribed processing action, but is not limited to the observation action. Also, in Embodiment 1, the sample observation device 1 refers to defect detection information 8 (including the above-mentioned defect candidate coordinates) generated and output by an external defect inspection device 5 ( Figure 1 ), and the case of taking an image 201 ( Figure 2 ) with the defect candidate coordinates as the shooting position is described.

[0064] Also, in Embodiment 1, an example of using SEM2 ( Figure 1 ) as a microscope or a shooting device used in the sample observation device 1 is described. Without being limited to this, as a microscope or a shooting device, other types of microscopes and shooting devices such as an optical microscope and a charged particle beam device using charged particles such as ions can also be applied. The processing in the system including the sample observation device 1 is performed using an appropriate image consistent with the type of microscope / shooting device to be used and shooting information (such as shooting position, shooting magnification, etc.).

[0065] Also, in Embodiment 1, as the taken image 201 based on SEM2 ( Figure 2),Basically, the description is made in the case of using an image taken from a direction perpendicular to the upper surface of the wafer, in other words, a top view image as the shooting direction. However, this is not limited thereto, and as the image 201, an image taken from a direction inclined with respect to the vertical direction of the upper surface of the wafer, in other words, an inclined direction, etc. can also be applied.

[0066] [Specimen Observation Device]

[0067] Figure 1 Shows the structure of a system including the specimen observation device 1 of Embodiment 1. The specimen observation device 1 is generally composed of an SEM 2 (scanning electron microscope) and a computer system 3. The SEM 2 is an example of a shooting device or a microscope, in other words, a charged particle beam device. The computer system 3 is a higher-level control device of the SEM 2, in other words, a controller. As a specific example, the specimen observation device 1 is a review SEM having the above-described ADR function. The computer system 3 is combined with the SEM 2 through a connection line (in other words, a signal line, a communication line), etc. The computer system 3 is a device having functions such as controlling the SEM 2. The specimen observation device 1 at least has a function of controlling the SEM 2 to obtain an image taken by the SEM 2, and a function of observing the specimen 10 based on the taken image.

[0068] The specimen observation device 1 has necessary functional blocks and various devices, but only a part of the essential elements is shown in the drawings. The overall specimen observation device 1 including Figure 1 is configured as a semiconductor inspection system, etc.

[0069] On a communication network 9 (e.g., LAN), as an external device for the specimen observation device 1, for example, a defect inspection device 5, a defect classification device 6, etc. are connected. In this example, as a result of inspecting the wafer as the specimen 10 in the external defect inspection device 5 in advance, defect detection information 8 is produced. The defect detection information 8 is information including defect candidate coordinates. The defect detection information 8 output from the defect inspection device 5 is stored, for example, in an external storage device 4 in advance. However, this is not limited thereto, and the defect detection information 8 may also be stored in a database such as a server on the communication network 9. The computer system 3 reads and refers to the defect detection information 8 from the external storage device 4 during observation.

[0070] The defect classification device 6 is a device or system having the above-described ADC function. The defect classification device 6 performs ADC processing based on the data / information of the processing result of the observation based on the ADR function in the specimen observation device 1 to obtain a result of classifying defects and defect images. In addition, not limited to Figure 1 the structural example of, it may also be a mode in which the defect classification device 6 is incorporated in the specimen observation device 1.

[0071] In Figure 1In the figure, a case is shown where the upper control device of the SEM2 is composed of one computer system 3. However, the upper control device may also be composed of multiple computer systems, such as multiple server devices.

[0072] The computer system 3 includes a control processor 102, a storage processor 103, an arithmetic processor 104, an input / output interface 105, a communication interface 107, a user interface control processor 106, etc. These components are connected to the bus 114 and can communicate and perform input / output with each other.

[0073] The control processor 102 controls the whole. The control processor 102 is composed of, for example, a hardware circuit, or a processor such as a CPU, MPU, or GPU. When the control processor 102 has a processor such as a CPU, the processor executes processing according to a program read from the storage processor 103. The control processor 102 realizes various functions according to program processing, for example. The arithmetic processor 104 is composed of, for example, a processor such as a CPU, MPU, or GPU, and memories such as a ROM and a RAM. The arithmetic processor 104 performs arithmetic operations according to a program read from the storage processor 103 by the processor. In addition, the control processor 102 and the arithmetic processor 104 may be of an integrated structure.

[0074] The storage processor 103 is composed of a device that stores various information and data including programs, and can be composed of, for example, a storage medium device having a disk, a semiconductor memory, etc. Programs and other data read from the external storage device 4 or the communication network 9 may also be stored in the storage processor 103. Defect detection information 8 and the like read from the external storage device 4 or the like may also be stored in the storage processor 103. Image data and the like obtained from the SEM2 may also be stored in the storage processor 103.

[0075] The input / output interface 105 is a device equipped with interfaces for input devices, output devices, and the external storage device 4, and performs input / output of data and information between these devices. The computer system 3 is connected to, for example, the external storage device 4 through the input / output interface 105. Various programs / data can be stored in the external storage device 4. Image data and processing result information can be stored in the external storage device 4.

[0076] The communication interface 107 is a device installed with a communication interface corresponding to the communication network 9 such as a LAN, and transmits and receives data and information to and from the communication network 9. The computer system 3 is connected to the communication network 9 via the communication interface 107. The computer system 3 is connected to external systems and devices via the communication network 9 and can communicate. For example, the communication network 9 is connected to the defect inspection device 5 and the defect classification device 6. Other examples of external devices include database servers, manufacturing execution systems (MES), etc. The computer system 3 can also refer to the design data of the sample (in other words, the sample information) and the information of the manufacturing process from the external device. Examples of design data include CAD data, etc. Examples of manufacturing processes include etching, deposition, etc.

[0077] The user interface control processor 106 is a part that provides / controls the user interface, and the user interface includes a graphical user interface (GUI) for inputting and outputting information / data with the user, in other words, the operator. The computer system 3 can also be connected to the user terminal 7 (in other words, the client terminal) as an input and output terminal through the user interface control processor 106. The user terminal 7 can also be a device connected to the communication network 9. The user terminal 7 and other input and output devices can also be built into the computer system 3 as a whole. The user interface control processor 106 provides the user terminal 7 with data of the screen (such as a web page) corresponding to the GUI. The input and output interface 105 or the user interface control processor 106 can be connected to other input and output devices, such as a display device, a sound output device, an operating device, etc. The user interface control processor 106 can also be integrated with the control processor 102, etc.

[0078] Can not be limited to Figure 1 The structure of the computer system 3 may also include more than one computer system. Figure 1 The computer system 3 has a plurality of processors, but any configuration having more than one processor will suffice.

[0079] The user operates the user terminal 7 or other input / output devices to input information such as instructions and settings to the sample observation device 1, especially the computer system 3, and also confirms information such as output on the screen. The user terminal 7 can also be applied to a general PC, for example. The user terminal 7 can have a built-in keyboard, mouse, display, etc., or can be connected externally. The user terminal 7 can also be a remote terminal connected to a communication network 9 such as the Internet. The user interface control processor 106 creates data of a screen having a GUI and provides it to the user terminal 7 through communication, and the user terminal 7 displays the screen on the display.

[0080] The system including the specimen observation device 1 can also be configured in the following manner. The computer system 3 can also be configured as a server in a client / server system, a cloud computing system, an IoT system, etc. The user terminal 7 can also be configured as a client computer for the server. In the case of the client / server system, for example, the user operates the user terminal 7 as a client computer, and the user terminal 7 sends a request to the computer system 3 as the server. The server receives the request and performs processing corresponding to the request (such as shooting and observation processing). The server sends the data of a screen (such as a Web page) including the processing result of the request, etc. as a response to the user terminal 7. The user terminal 7 receives the data of the response and displays the screen (such as a Web page) on the display. The user can confirm the processing result, etc. in this screen.

[0081] In addition, for example, the computer system 3 or an external device can also perform machine learning. In machine learning, a large amount of computer resources are sometimes required. In this case, the processing related to machine learning can also be performed in a server group such as a cloud computing system. In addition, the functions can also be shared between the server group and the client computer. In the case of performing machine learning, the computer system 3 uses learning images, etc. in the learning stage to train a learning model. Through training, the parameters of the learning model are adjusted. The learning model can apply, for example, CNN (Convolutional Neural Network), etc. In the estimation stage, the computer system 3 inputs an object image into the trained learning model and obtains determination result information, etc. as the output, which is the result of the estimation based on the learning model. In addition, the operator and the computer system implementing the learning / training of the model in the learning stage and the operator and the computer system implementing the estimation using the trained model in the estimation stage can also be different entities.

[0082] [SEM]

[0083] In Figure 1In the SEM2, there are a workbench 109, an electron gun 110, an electron lens (not shown), a deflector 112, a detector 111, etc. inside the housing 101. The workbench 109 is, in other words, a specimen stage that mounts / holds a semiconductor wafer as the specimen 10. The workbench 109 is a mechanism capable of performing movements such as horizontal directions (the illustrated X and Y directions), vertical direction (Z direction), rotation, inclination, etc. The electron gun 110 irradiates the specimen 10 on the workbench 109 with an electron beam b1, in other words, a charged particle beam b1. The electron lens (not shown) converges the electron beam b1 on the surface of the specimen 10. The deflector 112 scans the electron beam b1 on the surface of the specimen 10. The detector 111 detects electrons / particles b2 such as secondary electrons and backscattered electrons generated from the specimen 10 by the irradiation of the electron beam b1 as an electrical signal. In other words, the detector 111 detects the state of the surface of the specimen 10 as an image. In this example, the detector 111 has a plurality of detectors at a plurality of positions.

[0084] In this example, the SEM control function in the computer system 3 controls elements such as the workbench 109, the electron gun 110, the deflector 112, and the detector 111 of the SEM2. In addition, illustration of elements such as a drive circuit for driving mechanisms such as the workbench 109 is omitted.

[0085] The SEM2 takes a picture of the wafer 10 as the specimen 10 according to the control from the computer system 3 and other factors, in accordance with the set imaging conditions. The image signal detected as an electrical signal by the detector 111 of the SEM2 is supplied to the computer system 3 through a connection line. In addition, a circuit such as an analog / digital conversion circuit may be provided inside or at a subsequent stage of the detector 111. The computer system 3 processes the image signal supplied from the detector 111 through the control processor 102, the arithmetic processor 104, etc., thereby obtaining an image ( Figure 2 image 201 in the figure). In addition, during imaging by the SEM2, multiple images may be obtained through the multiple detectors 111 provided in the SEM2. The computer system 3 may also process the obtained multiple images to generate the image 201. The computer system 3 stores data / information such as this image in the storage processor 103, etc. The computer system 3 performs observation processing on this image, for example, determines / detects a defective part, and stores the processing result information in the storage processor 103, etc.

[0086] Figure 1The sample observation device 1 has an SEM 2 that takes pictures of a sample 10 based on the defect candidate coordinates of defect detection information 8 from a defect inspection device 5. Alternatively, the sample observation device 1 can also perform the same operation using an external microscope or imaging device. The computer system 3 can control an external microscope or imaging device or obtain a captured image from an external microscope or imaging device. In this system, multiple microscopes or imaging devices can coexist, such as an SEM and an optical microscope. The computer system 3 can also appropriately select the microscope or imaging device to be used for differential use. Figure 1 The overall structure of such a system is not limited to a single sample observation device 1. It can also be a system having one or more microscopes / imaging devices that capture images of the sample 10 and one or more computer systems or sample observation devices that perform processing such as observation on the captured images.

[0087] [Basic processing flow]

[0088] Use Figure 2 The sample observation device 1 and method of Embodiment 1 will be described. Figure 2 Indicates the basic processing flow executed by the computer system 3 ( Figure 1 ) in the sample observation device 1 and method of Embodiment 1, which has steps S1 to S7. The sample observation method of Embodiment 1 is a method for observing the sample 10, and has steps S2 and S3 for calculating the image quality evaluation value Ei. The computer system 3 in the sample observation device 1 of Embodiment 1 performs processing such as calculating its image quality evaluation value Ei. Figure 2 The processing flow also corresponds to the basic functional block structure related to the function of the Figure 1 computer system 3.

[0089] [Calculation of the image quality evaluation value of the observed image]

[0090] In Figure 2 , examples related to the calculation of the image quality evaluation value Ei in the sample observation device 1 and method of Embodiment 1 will be described. In Figure 2 , one of the features of the sample observation method of Embodiment 1 is that it has: a region determination step (step S2) for determining an evaluation partial region Si for an image 201 (also called an SEM image) obtained by photographing a semiconductor wafer 10 as the sample 10 with the SEM 2; and an image quality evaluation value calculation step (step S3) for calculating different image quality evaluation values Ei for the evaluation partial region Si.

[0091] As problems related to the calculation of the image quality evaluation value of SEM images, there are the following problems. Most of the image features related to the image quality evaluation value of SEM images exist in specific regions of the SEM image. Therefore, when the entire SEM image is used as the calculation region for the image quality evaluation value, the image features in regions that are irrelevant or have little relation to the image quality evaluation value also affect the image quality evaluation value as the calculation result. For example, when evaluating the sharpness of the image quality of an SEM image, it is desired to evaluate only the amount of change in brightness near the edge of the circuit pattern of the wafer. When the entire SEM image is set as the calculation region, the noise generated in the flat part of the circuit pattern affects the image quality evaluation value, i.e., sharpness.

[0092] Therefore, in the sample observation device and method of Embodiment 1, when there are multiple image quality evaluation values Ei (in other words, parameter values, evaluation item values) as elements constituting the image quality for the captured image 201, for each image quality evaluation value Ei, a specific region suitable for the calculation of that image quality evaluation value Ei is determined as the evaluation partial region Si (step S2). Then, the sample observation device and method of Embodiment 1 use this evaluation partial region Si to calculate the image quality evaluation value Ei (step S3).

[0093] In Figure 2 first, in step S1 as the shooting step, the computer system 3 controls the shooting based on the SEM 2 to obtain an image 201 for observing the wafer 10 of the sample 10 (in other words, a captured image, an object image, an SEM image, etc.). In addition, at this time, the captured image 201 is shot using the shooting conditions 205 ( Figure 6 ) described later, etc.

[0094] Next, in step S2 as the region determination step, the computer system 3 performs a region determination process on the captured image 201, thereby determining the evaluation partial region Si for each image quality evaluation value Ei of the multiple image quality evaluation values Ei applied to this image 201 (i = 1 to N, N: the number of partial regions). In this step S2, in addition to the image 201, the shooting conditions 205, design data 206 ( Figure 6 ) described later, manufacturing process information, etc. can also be used as input information.

[0095] Next, in step S3 as the image quality evaluation value calculation step, the computer system 3 uses the evaluation partial region Si corresponding to each image quality evaluation value Ei applied to the image 201 obtained in step S2 to calculate different respective image quality evaluation values Ei (i = 1 to N, N: the number of image quality evaluation values) for each evaluation partial region Si.

[0096] Next, in step S4, the computer system 3 stores the image quality evaluation value Ei calculated in step S3 in a memory resource (for example Figure 1 of the storage processor 103), and displays it to the user ( Figure 1 ). For example, this display can be achieved by the user interface control processor 106 of Figure 1 providing a GUI screen (described later) containing information such as the image quality evaluation value Ei to the user of the user terminal 7. The user can confirm the image quality evaluation value Ei and the like on this screen. The result of step S4 is output as the image quality evaluation result 204 including the image quality evaluation value Ei.

[0097] Next, in step S5, the computer system 3 uses the image quality evaluation result 204 including the image quality evaluation value Ei calculated in the above step S3 to perform a specified processing action such as observation, and generates and saves the processing result information. In one example, in the case of the observation action, the computer system 3 selects the image 201 determined to have high image quality based on the image quality evaluation value Ei as the observation image and performs the observation action. After step S5, this process ends.

[0098] In addition, step S6 is a method for the processing actions related to steps S2 and S3. In other words, it is a step for determining the set value, or a setting step and a method determination step. This step S6 includes the determination / setting of the method for determining the evaluation partial area Si in step S2 and the method for calculating the image quality evaluation value Ei in step S3. In step S6, the computer system 3 determines its method and set value according to the user setting information 202. The computer system 3 determines the control parameter values and the like under the method selected from several methods for the processing of each step (S21, S22, S31, S32), and applies them as set values to the processing of each step. The content determined in this step S6 can be preset as the design of this system, or can be determined according to the user setting information 202 based on the user setting.

[0099] In the case of using the user setting information 202, in advance in step S7, a setting based on the user ( Figure 1 ) is performed on the GUI screen provided by the computer system 3, such as the selection of the applied method and control parameter values, and the user setting information 202 is saved in the memory resource. Steps S6 and S7 can also be combined into one.

[0100] In step S6, the computer system 3 can also, like in step S2, input and use the object image 201, the shooting conditions 204 described later, the design data 205, etc. ( Figure 6 ) to determine the method and set value of each step. Step S6 can be performed before the execution of steps S2 and S3, or can be performed after the user confirms the execution results of steps S2 and S3.

[0101] Figure 2The details of the process are as follows. Step S2 specifically includes two steps, step S21 and step S22. Step S21 is a segmentation / classification step for generating a classification segmentation region Rj obtained by segmenting and classifying the image 201. Step S22 is an evaluation partial region determination step for determining the evaluation partial region Si using the classification segmentation region Rj as the classification result.

[0102] Step S3 specifically includes two steps, step S31 and step S32. Step S31 is a step of calculating the image quality evaluation value Ei for each evaluation partial region Si using the evaluation partial region Si. Step S32 is a step of calculating the comprehensive evaluation value C obtained by combining multiple image quality evaluation values Ei.

[0103] In step S2, the evaluation partial region Si can be calculated using rule-based image processing or machine learning. In addition, in step S3, the image quality evaluation value Ei can be calculated using rule-based image processing or machine learning.

[0104] [Image quality (1) evaluated using the image quality evaluation value Ei]

[0105] The image quality evaluation value Ei and the image quality evaluated using the image quality evaluation value Ei will be described. In the first embodiment of the present disclosure, there are multiple different image qualities related to the image 201 photographed by the SEM2 and evaluated using multiple different image quality evaluation values Ei. The image quality evaluation value Ei and the image quality, in other words, the elements, evaluation items, and parameter values of the image quality, are as follows.

[0106] In Figure 3 (A) thereof, the image quality evaluation value information 203 is represented in a table form. The image quality evaluation value information 203 is setting information / data regarding the image quality evaluation value Ei and the image quality. The computer system 3 preliminarily stores the image quality evaluation value information 203 in the memory resource according to the setting. In Figure 2 it is shown that the image quality evaluation value information 203 is used in the set step S6. The image quality evaluation value information 203 can also be set in the user-set step S7. The set image quality evaluation value information 203 is referred to in steps S2 and S3.

[0107] In Figure 3In the table of the image quality evaluation value information 203 of (A), as columns, there are "image quality evaluation value (Ei)", "image quality", and "evaluation / non-evaluation". The "image quality evaluation value (Ei)" column is the identification information of the image quality evaluation value. The "image quality" column is the image quality evaluated by the image quality evaluation value Ei corresponding to the image quality evaluation value Ei. In other words, it is the image quality name and image quality description. The "evaluation / non-evaluation" column indicates whether to evaluate or not evaluate the image quality based on the image quality evaluation value Ei of the target image 201. In other words, whether to apply this evaluation item / parameter. "Evaluation" (applied) is indicated by a circle.

[0108] According to the target image 201 ( Figure 2 ), it is possible to appropriately set "evaluation / non-evaluation" according to the image quality evaluation value Ei. Regarding this setting, it can also be performed by the user on the GUI screen ( Figure 2 step S7 of user setting). Or, this system can also automatically determine (step S6). For example, the computer system 3 can also automatically select the image quality evaluation value Ei used in the past process (manufacturing process of semiconductor devices) that has processed similar patterns for the semiconductor pattern seen in the image 201 and apply it to the image 201. The computer system 3 can also appropriately refer to the manufacturing process information, past performance information, statistical information, etc. related to the specimen 10.

[0109] In addition, in Figure 3 (A), for the sake of explanation, only the parameters of the image quality evaluation value Ei are shown. Regarding the parameter values of the specific calculation, they are appropriately stored in the memory resource in association with the target image 201. In Figure 3 (B), a data example of the image quality evaluation result 204 ( Figure 2 ) is shown in tabular form. In this image quality evaluation result 204, for each target image 201, the parameters (evaluation items) and values of the image quality evaluation value Ei are stored. The parameters are, for example, E1 to E9 the same as (A) and the comprehensive evaluation value C calculated in step S32.

[0110] In Embodiment 1, for each image 201 ( Figure 2 ), or for a specified target set (for example, the same semiconductor wafer 10), two or more of the above-mentioned multiple image quality evaluation values Ei are used. In Figure 3 the example, a case of such a setting is shown in which the image quality evaluation values E1, E2, E3, E5, and E9 are used to evaluate each image quality for a certain image 201.

[0111] In Figure 3 the example of (A), the multiple image quality evaluation values Ei have the following nine parameter values of E1 to E9.

[0112] (E1): Sharpness of the image

[0113] (E2): Visual identifiability of defects

[0114] (E3): Degree of noise suppression

[0115] (E4): Interlayer contrast

[0116] (E5): Degree of dark part emphasis

[0117] (E6): Shape preservation degree (length measurement value preservation degree)

[0118] (E7): Roughness preservation degree

[0119] (E8): Degree of ringing suppression

[0120] (E9): Naturalness of the image

[0121] The summary of each image quality evaluation value Ei and the image quality evaluated by using it is as follows.

[0122] The sharpness of the image of (E1) indicates the ease of distinguishing regions with different structures and materials from each other. Generally, it is calculated as the degree of brightness change at the boundary (edge) between regions with different structures and materials.

[0123] The visual identifiability of defects of (E2) indicates the ease of distinguishing the defective region from the surrounding region. For example, there are the contrast between the defective region and the surrounding region, the contrast inside the defective region, the intensity of the shadow caused by the defect, etc.

[0124] The degree of noise suppression of (E3) indicates the amount of noise generated by shooting, etc.

[0125] The interlayer contrast of (E4) indicates the contrast between different layers in the semiconductor pattern.

[0126] The degree of dark part emphasis of (E5) indicates the contrast of the dark part such as the lower layer of the semiconductor pattern, and the contrast between the dark part region and the defective region.

[0127] (E6)'s shape preservation degree (in other words, length measurement preservation degree) indicates the similarity between the length measurement value in the reference image where a similar pattern appears and the length measurement value in the image to be evaluated. The larger this evaluation value is, the more the shape of the pattern of the specimen is preserved.

[0128] (E7)'s roughness preservation degree indicates the similarity between the roughness of the pattern in the reference image where a similar pattern appears and the roughness of the pattern in the image to be evaluated. The larger this evaluation value is, the more the roughness of the pattern of the specimen is preserved.

[0129] (E8)'s degree of ringing suppression indicates the amount of ringing (referring to wavy artifacts) generated in the image.

[0130] (E9) The naturalness of the image represents the degree of discomfort felt by a person when viewing the visual image.

[0131] [Image quality (2) evaluated using the image quality evaluation value Ei]

[0132] For Figure 3 each image quality evaluation value Ei, a supplementary explanation of its definition is provided. Figure 4-5 It is a schematic explanatory diagram showing the definition of each image quality evaluation value Ei.

[0133] (1) Figure 4 (1a) and (1b) of

[0134] (2) Figure 4 (2a) and (2b) of

[0135] (3) Figure 4 (3a) and (3b) of

[0136] (4) Figure 4 (4a) and (4b) of

[0137] (5) Figure 4Figs. (5a) and (5b) are explanatory diagrams regarding the "dark part emphasis degree" which is the image quality evaluation value E5. The image in (5a) is an example of an image when the dark part emphasis degree is relatively small, and the image in (5b) is an example of an image when the dark part emphasis degree is relatively large. The computer system 3 calculates, for example, the contrast within the dark part area 451 or the contrast between the dark part area 451 and the defect area 452 as the value of the dark part emphasis degree.

[0138] (6) Figure 5 Figs. (6a), (6b), and (6c) are explanatory diagrams regarding the "shape preservation degree" which is the image quality evaluation value E6, showing examples of images. The image in (6a) is an example of a reference image, the image in (6b) is an example of an image when the shape preservation degree is relatively small, and the image in (6c) is an example of an image when the shape preservation degree is relatively large. The shape preservation degree is a value for evaluating whether the shape of the pattern captured in the reference image is also preserved in the observation image which is the object of image quality evaluation. This evaluation value can be calculated by comparing the edges of the reference image and the edges of the object image to be evaluated. For example, a certain edge 461 in the reference image in (6a) is compared with the corresponding edge 462 in the image in (6b). In addition, a certain edge 461 in the reference image in (6a) is compared with the corresponding edge 463 in the image in (6c). In the image in (6b), the change in the shape of the edge 462 is relatively large with respect to the edge 461, but in the image in (6c), the change in the shape of the edge 463 is relatively small with respect to the edge 461.

[0139] In this example, an example of the shape change of the wiring pattern is shown, but in the calculation of the shape preservation degree, not only the shape of the wiring pattern but also whether the shape of the defect (defect area 464) can be preserved is evaluated. In this example, a case where the shape of a part of the pattern is not preserved is shown, but even when the shape of the edge remains unchanged and the overall width of the pattern shrinks, due to the change in the position of the edge, the measured length value of the pattern width changes and the shape preservation degree becomes smaller.

[0140] (7) Figure 5(7a), (7b), and (7c) are explanatory diagrams regarding the "roughness preservation degree" as the image quality evaluation value E7, showing examples of images. The image in (7a) is an example of a reference image, the image in (7b) is an example of an image when the roughness preservation degree is relatively small, and the image in (7c) is an example of an image when the roughness preservation degree is relatively large. During the manufacturing process of the semiconductor pattern, fluctuations occur in the shape of the pattern edges. The computer system 3 evaluates whether the roughness of the pattern edges captured in the reference image is also preserved in the observation image that is the object of image quality evaluation, as the value of the roughness preservation degree. In this example, there are fluctuations and roughness in the edge 471 of a certain pattern in the reference image of (7a). In contrast, in the image of (7b), the corresponding edge 472 is close to a straight line, and in the image of (7c), the corresponding edge 473 has fluctuations and roughness.

[0141] (8) Figure 5 (8a) and (8b) are explanatory diagrams regarding the "ringing suppression degree" as the image quality evaluation value E8, showing examples of images. The image in (8a) is an example of an image when the ringing suppression degree is relatively small, and the image in (8b) is an example of an image when the ringing suppression degree is relatively large. Due to the influence of shooting conditions and image quality improvement processing, wavy artifacts (referred to as ringing) appear around the edges of the specimen pattern in the observation image. The computer system 3 evaluates whether ringing appears in the evaluation object image as the value of the ringing suppression degree. At this time, similar to the above shape preservation degree and roughness preservation degree, the computer system 3 prepares a reference image, and for example, by detecting brightness changes that do not exist in the reference image, the ringing suppression degree can be calculated. In the image of (8a), for example, with respect to a certain edge 481, ringing occurs in the transverse direction (represented by multiple vertical lines). In contrast, in the image of (8b), the ringing is suppressed at the corresponding edge 482.

[0142] (9) As the image quality evaluation value E9, the "naturalness of the image" represents the degree of discomfort felt by people when viewing the image. For example, in a normal observation image, although there is a quantity size, due to the generation of noise, the naturalness of the image may become smaller in an image where the noise has been completely removed.

[0143] [Setting step (S6)]

[0144] Use Figure 6 , to explain the details of the setting (method determination) step S6. Specifically, as Figure 6 shown, step S6 has, for example, steps S61, S62, S63, S64, and S65 as five steps.

[0145] In step S61, the computer system 3 sets Figure 2The classification method of the classification division region Rj in step S21. In step S61, the computer system 3 sets, for example, the classification method of the classification division region Rj shown in Fig.10 etc.

[0146] In step S62, the computer system 3 sets the determination method of the evaluation partial region Si in Figure 2 step S22.

[0147] As the methods of steps S21 and S22, a method based on rules or machine learning can be generally set.

[0148] In step S63, the computer system 3 determines the calculation method of the image quality evaluation value Ei in Figure 2 step S31.

[0149] In step S64, the computer system 3 determines the calculation method of the comprehensive evaluation value C in Figure 2 step S32.

[0150] As the methods of steps S31 and S32, a method based on rules or machine learning can be generally set.

[0151] Based on the above steps, step S65 can also be used. The image quality evaluated using the image quality evaluation value Ei is not limited to the 9 image quality evaluation values E1 to E9 listed in Figure 3 , and a new image quality evaluation value can also be defined and added. In this case, in step S65, the computer system 3 adds a new image quality evaluation value Ei according to the user's needs. The computer system 3 adds a new image quality evaluation value Ei, for example, according to the user's operation on the GUI screen, and sets the information of the added image quality evaluation value Ei in the image quality evaluation value information 203. More specifically, the added image quality evaluation value Ei corresponds to the calculation method, definition information, program, etc. for calculating the image quality evaluation value Ei. The added image quality evaluation value Ei then becomes a candidate for the image quality evaluation value Ei applied to the image 201, and can be selected Figure 3 in the "evaluate / not evaluate" in. In addition, the added image quality evaluation value Ei can also be Figure 3 a change in the image quality evaluation values E1 to E9. For example, regarding a certain image quality evaluation value Ei, if other calculation formulas, etc. can be implemented in the definition for calculating its value, other calculation formulas, etc. can also be set as the change. For example, two added image quality evaluation values Eia and Eib are set.

[0152] Regarding the example of the shooting condition 205, as based on SEM2 ( Figure 1During shooting, various parameter values, such as acceleration voltage, probe current, electron beam opening angle, electron beam inclination angle, focal position, etc., can be listed. The design data 206, in other words, examples of specimen information can include information such as the height corresponding to the shape of the semiconductor wafer 10 ( Figure 1 ) and, in particular, the position in the horizontal direction, and information about the material.

[0153] [Determination of the Si region for evaluation and calculation of the image quality evaluation value Ei]

[0154] As an issue related to the evaluation of image quality, when the entire image is set as the calculation region for the image quality evaluation value, the image characteristics of regions irrelevant to the image quality also affect the image quality evaluation value. Therefore, there are points such as the inability to quantify the image quality, difficulty in quantification, or a decrease in the quality of quantification.

[0155] In contrast, in the first embodiment, from the target image 201 ( Figure 2 ), only the region that reflects the image characteristics related to the image quality evaluation value Ei corresponding to the image quality to be evaluated is determined as the evaluation partial region Si (step S2). Thus, for each image quality, the image quality evaluation value Ei can be calculated using the appropriate evaluation partial region Si, and the quantification of the image quality considering multiple image qualities can be performed.

[0156] Figure 7 Shows a processing example of the determination of the evaluation partial region Si and the calculation of the image quality evaluation value Ei corresponding to steps S2 and S3 of Figure 2 . For the input target image 201, in step S701, the computer system 3 obtains / refers to the information of the image quality evaluation values Ei of two or more image qualities applied to the image quality evaluation from the image quality evaluation value information 203 like Figure 3 . Here, the multiple (N) image quality evaluation values Ei applied are represented by E1 to EN.

[0157] Next, in step S702 corresponding to step S2, the computer system 3 determines multiple (N) evaluation partial regions Si (S1 to SN) corresponding to the multiple (N) image quality evaluation values Ei (E1 to EN) applied. That is, the computer system 3 determines the evaluation partial region Si for each image quality evaluation value Ei. Here, the multiple (N) evaluation partial regions Si are represented by S1 to SN.

[0158] In this example, as Figure 7As shown, the correspondence between multiple image quality evaluation values Ei and multiple partial regions Si for evaluation is one-to-one. The partial regions Si for evaluation are one or more regions, where i = 1 to N, and N is the number of partial regions. The image quality evaluation values Ei are one or more parameter values, where i = 1 to N, and N is the number of parameters of the evaluation items. In this example, the number N of the partial regions Si for evaluation and the number N of the parameters of the image quality evaluation values Ei are the same value, but it is not limited to this.

[0159] Next, in step S703 corresponding to step S3, the computer system 3 calculates the image quality evaluation value Ei for each partial region Si for evaluation. The calculation result is stored as the image quality evaluation result 204.

[0160] As a variant example, Figure 8 An example showing a case where the correspondence between the partial regions Si for evaluation and the image quality evaluation values Ei is not one-to-one is presented. Any correspondence is possible. (A) shows a case where the correspondence between Si and Ei includes one-to-many. For example, using the partial region S1 for evaluation, the image quality evaluation values E1, E2, and E3 are calculated respectively. (B) shows a case where the correspondence between Si and Ei includes many-to-one. For example, using the partial regions S1, S2, and S3 for evaluation, the image quality evaluation value E1 is calculated.

[0161] In addition, Fig. 9 It is an explanatory diagram of the case where the partial regions Si for evaluation of each image quality evaluation value Ei do not overlap and the case where they overlap. Each partial region Si for evaluation can be in a non-overlapping case or an overlapping case, and any case can be applied. The example of (A) shows a case where the partial region Sa for a certain image quality evaluation value Ea and the partial region Sb for a certain image quality evaluation value Eb do not overlap. In this example, the hatched pattern region in image 901 is the partial region Sa for evaluation, and the white region in image 902 is the partial region Sb for evaluation, and they do not overlap.

[0162] (B) shows a case where the partial region Sa for a certain image quality evaluation value Ea and the partial region Sc for a certain image quality evaluation value Ec overlap and are the same, and a case where the partial region Sa for a certain image quality evaluation value Ea and the partial region Sd for a certain image quality evaluation value Ed partially overlap. In this example, the hatched pattern region in image 911 is the partial region Sa for evaluation, and the hatched pattern region in image 912 is the partial region Sb for evaluation, and they are the same. In addition, the region shown by the dashed box in image 913 is the partial region Sd for evaluation, and this partial region Sd for evaluation partially overlaps with the partial region Sa for evaluation. For example, the partial region Sd for evaluation is a region obtained by magnifying the horizontal width of the partial region Sa for evaluation.

[0163] In the determination of the partial regions Si for evaluation ( Figure 2In step S22), different image quality evaluation values Ei, such as E1 and E2, are values of evaluation items (parameters) for different image qualities. Therefore, basically, the evaluation partial regions Si (such as S1 and S2) used as the regions for calculating these image quality evaluation values Ei also often become different regions. For example, as in the example shown in Fig. 9 (A) of, the evaluation partial region Sa for a certain image quality evaluation value Ea and the evaluation partial region Sb for another image quality evaluation value Eb do not overlap and become different regions. Not limited to this, each evaluation partial region Si corresponding to each image quality evaluation value Ei can also be as in Fig. 9 (B) of the example shown, there are cases where they become the same region or cases where they become partially overlapping regions.

[0164] [Region determination step (S2)]

[0165] Describe the details of step S2 for region determination of Figure 2 . In step S21, the computer system 3 classifies the classified segmentation regions Rj (j = 1 to Nr) after dividing the image 201. Here, the classified segmentation regions are set as Rj, j = 1 to Nr, and Nr is the number of segmentation regions. In step S22, the computer system 3 determines the evaluation partial regions Si (S1 to SN) based on the classified segmentation regions Rj and their classification results.

[0166] As a problem related to the image quality evaluation of an image, such as Figure 3 the sharpness (E1) and the visual recognition of defects (E2) in the example, when calculating the image quality evaluation value (Ei) focusing on the structure of a semiconductor device, it is difficult to calculate this image quality evaluation value without information on the structure (circuit pattern, etc.) in the captured image and the position of this structure. For example, the image quality evaluation value E1 related to sharpness represents the degree of brightness change such as the edge of a circuit pattern as a structure. However, without information on the structure and position in the captured image 201, information on regions with large brightness changes such as noise in flat parts, which does not contribute to the evaluation of sharpness, is also included in the calculation of the image quality evaluation value. As a result, the accuracy of the image quality evaluation value decreases.

[0167] Therefore, in Embodiment 1, in step S21, the computer system 3 performs region division so that each region in the image 201 becomes a structural element of a semiconductor device, and on this basis, classifies which structural element each divided region, that is, the classified segmentation region Rj, is. Then, in step S22, the computer system 3 uses the information of the classified segmentation regions Rj and the classification results to determine the regions suitable for the calculation of each image quality evaluation value Ei as the evaluation partial regions Si.

[0168] Fig.10 Shows a detailed processing example of step S2. First, the computer system 3 performs the processes of steps S1001 and S1002 corresponding to the segmentation / classification step S21 on the evaluation object image 201. In step S1001, the computer system 3 divides the entire area of the image 201 into respective areas in a manner that matches the structural elements of the semiconductor device observed in the image 201, and temporarily obtains the segmented areas in the state before classification. In step S1002, the computer system 3 classifies by structural elements based on these segmented areas so that each segmented area in the image 201 becomes a structural element of the semiconductor device, and determines the segmented areas in the classified state, that is, the segmentation areas Rj for classification.

[0169] In steps S1001 and S1002, as inputs, not only the image 201 of the specimen 10 but also the design data 206 of the semiconductor pattern corresponding to the image 201, the shooting conditions 205, and the classification information 207 from the user can be used. The design data 206 has information such as the position and shape of, for example, the circuit pattern of the wafer 10 of the specimen 10. The shooting conditions 205 are, for example, information such as the conditions for controlling Figure 1 the irradiation of the electron beam b1 of the SEM 2 and the conditions related to detection. The classification information 207 from the user is, for example, information set by the user operating by observing the captured image, etc. in the GUI screen (for example, Figure 6 in step S7) for segmentation into the segmentation areas Rj for classification and classification.

[0170] In step S2, specifically, rule-based segmentation methods such as the Watershed method or machine learning-based segmentation methods such as CNN can also be used. In addition, in step S2, semantic segmentation methods such as FCN (Fully Convolutional Network) can also be used to perform region segmentation and region classification simultaneously.

[0171] Through the segmentation / classification step S21 ( Fig.10 S1001 and S1002 therein), the classification result 1010 of the segmentation areas Rj for classification is obtained. In this example, as shown in the figure, the classification result 1010 has a defect area 1011, a lower layer area 1012, an edge area 1013, and a wiring upper surface area 1014 as the segmentation areas Rj for classification, but is not limited thereto. These areas are areas corresponding to the structural elements of the semiconductor device and are respectively corresponded to the segmentation areas Rj for classification. If an example of the correspondence, identification, and classification of the data / information representing the classification result 1010 with respect to the segmentation areas Ri for classification, then it becomes R1 = defect area 1011, R2 = lower layer area 1012, R3 = edge area 1013, R4 = wiring upper surface area 1014.

[0172] The defective area 1011 is an area where a defect is roughly suspected, which is different from the result of defect observation in step S5. The lower layer area 1012 is an area on the lower layer side in the shooting direction and the semiconductor device structure, in other words, a dark area. The upper surface area 1014 of the wiring is an area on the upper surface side in the shooting direction and the semiconductor device structure, which is a vertical line pattern area in this example. The edge area 1013 is an area that forms the edge of the upper surface area 1014 of the wiring.

[0173] After that, in step S1003 corresponding to step S22, the computer system 3 extracts the evaluation partial area Si corresponding to the image quality evaluation value Ei according to the classification result 1010, thereby determining the evaluation partial area Si (1020). In the illustrated evaluation partial area 1020, in each of the images G1 to GN, the white area shown in the figure represents the area corresponding to the evaluation partial area Si and used for the calculation of the image quality evaluation value Ei, and the black area shown in the figure represents the area that does not conform to the evaluation partial area Si and is not used for the calculation of the image quality evaluation value Ei. In the calculation of the image quality evaluation value Ei in the subsequent step S31, the computer system 3 only uses the white area as the evaluation partial area Si in the image 201 for the calculation of the image quality evaluation value Ei, and does not use the black area.

[0174] For example, the evaluation partial area S1 for the image quality evaluation value E1 "sharpness" is constituted by using the extracted classification segmentation area R3 = edge area 1013 (represented by a slanted pattern). Specifically, the evaluation partial area S1 is constituted by widening the edge area 1013 (R3). In addition, for example, the evaluation partial area S5 for the image quality evaluation value E5 "dark part emphasis degree" is constituted by using the extracted classification segmentation area R2 = lower layer area 1012 (represented by a slanted pattern).

[0175] Regarding the evaluation partial area Si output as a result of step S2, it can also be manually corrected by the user after being saved in the memory resource. As described later, the computer system 3 displays the evaluation partial area Si on the GUI screen, for example, and the user observes and confirms the evaluation partial area Si and makes corrections, determination / saving as needed.

[0176] [Classification method of classification segmentation area Rj using design data]

[0177] Use Fig.11 , for the design data 206 using the semiconductor pattern corresponding to the image 201, Figure 2 The detailed processing example of the classification method of the classification segmentation area Rj in the segmentation / classification step S21 (steps S1001 and S1002 in Fig.10 ) will be described.

[0178] In Embodiment 1 (especially Fig.11 in the illustrated embodiment), one of the features of the segmentation / classification step S21 is that the design data 206 of the semiconductor circuit corresponding to the image 201 is used as input and classified according to the segmentation regions Rj for classification.

[0179] As a problem, sometimes it is impossible to classify the evaluation object image into the structural elements of the semiconductor, or the classification is difficult. In contrast, in Embodiment 1 ( Fig.11 ), in the classification of the segmentation regions Rj for classification, the design data 206 of the semiconductor is used as the classification criterion for the structural elements. Thus, the classification of the segmentation regions Rj for classification can be performed.

[0180] Fig.11 FIG. shows a processing example of the classification using the design data 206. First, the computer system 3 inputs the design data 206 of the semiconductor corresponding to the evaluation object image 201. In the design data 206 of this example, the illustrated white portion 1103 corresponds to the upper surface wiring region, and the black portion 1104 corresponds to the dark region. The design data 206 of this example includes regions of structural elements corresponding to the circuit patterns in the image 201. In the design data 206, not only the wiring patterns captured in the image 201 but also other wiring patterns are included. Therefore, in step S1101, the computer system 3 inputs the image 201 and the design data 206 and matches them to obtain the matched design data 206B. The matched design data 206B is the data in the design data 206 corresponding to the portion of the wiring pattern captured in the image 201.

[0181] After that, in step S1102 as the classification step, the computer system 3 classifies the segmentation regions Rj for classification of the image 201 according to the matched design data 206B to obtain the classification result 1110. In addition, regarding the defective region 1111 in the classification result 1110, it can be determined by extracting a region with a large difference from the average brightness of the upper surface wiring region or the average brightness of the dark region.

[0182] [Classification method of the segmentation regions Rj for classification using the shooting conditions]

[0183] Using Fig.12 , a detailed processing example of the classification method of the segmentation regions Rj for classification in the segmentation / classification step S21 ( Fig.10 steps S1001 and S1002 therein) regarding the shooting conditions 205 using the evaluation object image 201 will be described.

[0184] In Embodiment 1 (especially Fig.12In the embodiment of [], one of the features of the classification step S21 is that the shooting condition 205 of the image 201 is used as an input, and classification is performed according to the classification segmentation region Rj.

[0185] As a problem, for the evaluation target image, it may not be possible to classify it as a semiconductor structural element, or the classification may be difficult. In contrast, in Embodiment 1, in the classification of the classification segmentation region Rj, the shooting condition 205 of the evaluation target image 201 is used as the classification criterion for the structural element. Thereby, the classification of the classification segmentation region Rj can be performed.

[0186] Fig.12 (A) of [] shows a processing example of classification using the shooting condition 205. In step S1201 corresponding to the segmentation / classification step S21, the computer system 3 uses the evaluation target image 201 and the shooting condition 205 when shooting this image 201 as inputs, classifies the classification segmentation region Rj of the image 201 with respect to the semiconductor structure, and obtains the classification result 1210.

[0187] In this classification step S1201, the computer system 3, for example, previously learns the relationship between the acceleration voltage, which is one of the shooting conditions 205, the brightness value of the image 201, and the material of the semiconductor structure. According to the learning model 1200, based on the input acceleration voltage and brightness value, the material of the structure is estimated as an output, and the classification segmentation region Rj is classified according to the estimation result.

[0188] Fig.12 (B) of [] shows an example of the prior learning stage. In the learning step S1202, the computer system 3 uses the learning shooting condition 1221, the learning image data 1222, and the learning semiconductor structure information 1223 (including information on the material of the semiconductor structure) to construct the learning model 1200 and perform learning / training. Through training, for example, the parameter values of the learning model 1200 are adjusted to be appropriate.

[0189] In addition, regarding the defect region 1211 in the classification result 1210, it can be determined, for example, by extracting the region where the material cannot be estimated.

[0190] [Steps for determining Si using Rj]

[0191] Use Fig.13 , the details of the step S22 ([[]] Figure 2 ) for determining the evaluation partial region Si using the classification segmentation region Rj will be described. In Embodiment 1, Figure 2 The steps S2 (S21, S22), especially Fig.13One of its features is that the evaluation partial region Si for each image quality evaluation value Ei is generated by combining, dividing, dilating, contracting, or a combination thereof of the classification division regions Rj.

[0192] As a problem, depending on the image quality evaluation value Ei, the evaluation partial region Si, which is a region suitable for the calculation of the image quality evaluation value Ei, and the classification division region Rj divided / classified according to the semiconductor structure may not always match. In this case, even if the image quality evaluation value Ei is simply calculated according to the classification division region Rj, the desired image quality cannot be evaluated. In other words, it is difficult to improve the accuracy of image quality evaluation. For example, Figure 3 The image quality "visual recognition of defects" shown by the image quality evaluation value E2 of Figure 4 represents the ease of distinguishing the defect region from the surrounding region as shown in (2a) and (2b) of

[0193] In contrast, in Embodiment 1, one process selected from processes such as combining, dividing, dilating, and contracting the classification division regions Rj, or a combination of multiple selected processes is applied, thereby generating the evaluation partial region Si, which is a region suitable for the calculation of the image quality evaluation value Ei. Thus, the image quality evaluation value Ei can also be calculated in a region including multiple classification division regions Rj or a partial region of the classification division region Rj.

[0194] Fig.13 shows a processing example of generating the evaluation partial region Si by processing such as combining, dividing, dilating, and contracting the classification division regions Rj. The input classification result 1010 is the same as Fig.10 First, take the setting of the evaluation partial region Si when the image quality evaluation value Ei is the image quality evaluation value E2 "visual recognition of defects" as an example. When calculating the image quality evaluation value E2 "visual recognition of defects", the contrast between the defect region and the surrounding region affects the evaluation value. Therefore, not only the defect region but also the region adjacent to the defect region, that is, the surrounding region, needs to be included in the evaluation partial region Si.

[0195] Therefore, the computer system 3 selects the classification division region Rj corresponding to the image quality evaluation value Ei (E2) in the classification result 1010 in step S1301. The computer system 3 selects the defect region 1011 (R1) and the dark region 1012 (R2), which is the region adjacent to the defect region 1011 (R1), as the classification division region Rj. In Fig.13Among them, the classified divided region Rj selected here is also represented as the shaded pattern region in the image 1304. Then, in step S1302 of the computer system 3, the selected defect region 1011 (R1) is combined with the region 1012 (R2), whereby the evaluation partial region S2 for generating the image quality evaluation value E2 of "visual recognition of defects" is generated. The white part in the generated image 1305 is the evaluation partial region S2 thereof.

[0196] The image 1311 shows another example when calculating the evaluation partial region S2 for the image quality evaluation value E2. In this example, the following situation is shown: when taking the peripheral region with respect to the defect region 1011 (R1), not all of the adjacent dark part regions 1012 (R2) are used, but only a part of the range close to the defect region 1011 (R1) within the adjacent dark part regions 1012 (R2) is used, and in this example, only the range 1312 represented by the rectangle is used.

[0197] As another example, when the image quality evaluation value Ei is Figure 3 the determination of the evaluation partial region S1 for the "sharpness" of the image quality evaluation value E1 is as follows. When the computer system 3 calculates the image quality evaluation value E1 of "sharpness", it calculates the degree of change in brightness at the boundary (in other words, the edge, etc.) between regions with different semiconductor structures and materials. In the calculation of the degree of change in brightness, it is necessary to observe the change amount when comparing the brightness of the edge region with the brightness of the peripheral region. Therefore, in the evaluation partial region S1 for the image quality evaluation value E1, not only the edge region but also the peripheral region of the edge region needs to be included.

[0198] Therefore, the computer system 3 selects the classified divided region Rj corresponding to the image quality evaluation value E1 in the classification result 1010. The computer system 3 first selects the region 1013 (R3) as the edge region as the classified divided region Rj. The selected region here is also represented as the shaded pattern region in the image 1307. In step S1303, the computer system 3 generates the evaluation partial region S1 for the image quality evaluation value E1 related to "sharpness" by expanding the region 1013 (R3) as the edge region. In this example, the computer system 3 expands each region 1013 (R3), which is the edge of the vertical line pattern extending in the y-axis direction, in the left and right directions of the x direction, which is the direction orthogonal thereto, as shown by the arrow in the expansion direction 1309 in the image 1307. The white part in the generated image 1308 is the evaluation partial region S1 thereof.

[0199] Regarding the evaluation partial regions Si for other image quality evaluation values Ei, similar to the above example, they can be generated by the process using the classification segmentation regions Rj according to the definition of the process for each image quality evaluation value Ei. Additionally, depending on the different image quality evaluation values Ei, not only the combination or dilation of the classification segmentation regions Rj in the above example, but also the evaluation partial regions Si can be generated by segmentation, contraction, or a combination thereof. As an example of segmenting the classification segmentation regions Rj, the upper surface region 1014 (R4) of the wiring can be segmented into multiple regions according to relationships such as far / near from the edge region 1013 (R3). As an example of contracting the classification segmentation regions Rj, the upper surface region 1104 (R4) of the wiring can be contracted to exclude the region near the edge region 1103 (R3).

[0200] Regarding the above method of generating the evaluation partial regions Si from the classification segmentation regions Rj according to the image quality evaluation value Ei, in Figure 6 step S62, it can be set by the user as a generation method according to rules, or it can be set by the user as a generation method using machine learning.

[0201] Fig.14 Examples including other image quality evaluation values Ei show examples of the generation results of the evaluation partial regions Si (S1 to S8) for each image quality evaluation value Ei (E1 to E8). In each image, the white part represents the evaluation partial region Si. For example, in the case of the image quality evaluation value E3 "noise suppression degree", the computer system 3 generates the evaluation partial region S3 based on the upper surface region 1014 (R4) of the wiring and the dark part region 1012 (R2). Additionally, in the case of the image quality evaluation value E4 "interlayer contrast", the computer system 3 generates the evaluation partial region S4 based on the upper surface region 1014 (R4) of the wiring and the dark part region 1012 (R2). Additionally, in the case of the image quality evaluation value E5 "dark part emphasis degree", the computer system 3 generates the evaluation partial region S5 based on the upper surface region 1014 (R4) of the wiring and the defect region 1011 (R1).

[0202] For example, in the case of the image quality evaluation value E6 "shape preservation degree", the computer system 3 generates the evaluation partial area S6 in the following manner: not only the peripheral area of the pattern edge but also whether the shape of the defective area is preserved can be evaluated. The area 1406 is an area obtained by expanding the width to a certain extent in the form of a line including the shape (e.g., an ellipse) of the defective area 1011 (R1). In addition, in the case of the image quality evaluation value E7 "roughness preservation degree", the computer system 3 generates the evaluation partial area S7 based on the edge area 1013 (R3) described above. In addition, in the case of the image quality evaluation value E8 "ringing suppression degree", the computer system 3 generates an area around the edge other than the edge area (e.g., an area within a certain distance from the edge area) as the evaluation partial area S8.

[0203] [Classification criteria for the classification division area Rj]

[0204] Use Fig.15 , to Figure 2 explain the classification criteria for the classification division area Rj related to the step S21. In the first embodiment, one of the characteristics of the classification division area Rj is that it includes at least one or more areas among the upper surface area of the wiring, the edge area, the dark area, and the defective area as categories and classifications. These classification division areas Rj correspond to Fig.10 the defective area 1011, the dark area 1012, the edge area 1013, and the upper surface area of the wiring 1014 in the example of

[0205] As a problem, the appearance varies greatly depending on the image, so there is a point that it is sometimes difficult to determine the classification criteria for the classification division area Rj used to determine the evaluation partial area Si. In contrast, in the first embodiment, focusing on the structural information of the semiconductor, the classification division area Rj is classified in such a way as to include at least one or more areas among the upper surface area of the wiring, the edge area, the dark area, and the defective area. For example, as described above, the defective area 1011 is classified as the first classification division area R1, the dark area 1012 is classified as the second classification division area R2, the edge area 1013 is classified as the third classification division area R3, and the upper surface area of the wiring 1014 is classified as the fourth classification division area R4.

[0206] Fig.15It is an explanatory diagram regarding the above classification criteria. The wiring upper surface region is the region occupied by the wiring pattern at the highest position in the vertical direction (Z direction) with respect to the wafer surface in the image of the top view taken from above the wafer surface. The dark region is the region occupied by the wiring pattern or the background portion, etc. at the lowest position in the vertical direction (Z direction) with respect to the wafer surface in the image of the top view taken from above the wafer surface. The edge region is the region at the boundary between the wiring upper surface region and the dark region. The defect region is a general region suspected of defects such as unwanted patterns and foreign objects.

[0207] In Fig.15 the example of, the image 1501 shown on the left is the same image as the aforementioned image 201. The classification result 1502 represents the result of classifying the image 1501 into a plurality of classification segmentation regions Rj. In the classification result 1502, the regions within the image 1501 are classified into four classification segmentation regions Rj (for example, R1 to R4) which are the wiring upper surface region 1503, the edge region 1504, the dark region 1505, and the defect region 1506. In addition, as shown in the illustrated example, a certain classification segmentation region Rj may also have a plurality of regions at separate positions.

[0208] In other examples, the image 1507 shown on the right is an image obtained by photographing a pattern with the same structure but a different appearance from the image 1501. The classification result 1508 represents the result of classifying the image 1507. The classification result 1508 is also the same as the classification result 1502, and the regions within the image 1507 are classified into four classification segmentation regions Rj which are the wiring upper surface region 1509, the edge region 1510, the dark region 1511, and the defect region 1512.

[0209] As in the above examples, for images that capture the same structure but have different appearances, by focusing on the semiconductor structure information, unified classification can be performed. The semiconductor structure information here can be the structure information known through image analysis based on the captured image 201, or it can also be the structure information known based on the design data 206.

[0210] In the example of Embodiment 1, the image is classified into four regions which are the wiring upper surface region, the edge region, the dark region, and the defect region, but it is not limited to this. For example, in the case where the wiring has a structure of 3 layers or more, other regions such as the wiring middle layer region may also be added to the classification result. In addition, the edge region can be set not only as the boundary between the wiring upper surface region and the dark region, but also as the boundary between the wiring upper surface region and the wiring middle layer region, the boundary between the wiring middle layer region and the dark region, etc., that is, the boundary of the regions of each layer.

[0211] [Method for determining internal parameters related to the calculation of Ei]

[0212] Use Fig.16 , the method for determining the internal parameter Qi_m related to the calculation of the image quality evaluation value Ei in step S63 in Figure 6 will be described.

[0213] In Embodiment 1, there is an interface for inputting an evaluation from a user related to the image quality of an image, and there is step S63 of determining a calculation method for the image quality evaluation value Ei using this user evaluation. In this step S63, one of the features of the computer system 3 is that the internal parameters Qi_m (m = 1 to Nm) are used to calculate the image quality evaluation value Ei, and the internal parameters Qi_m are determined in such a way that the image quality evaluation value Ei is consistent with the user evaluation or as close to it as possible. The internal parameter Qi_m is a parameter that constitutes the image quality evaluation value Ei, in other words, it is a parameter of the process in the method for calculating the image quality evaluation value Ei. m = 1 to Nm, and Nm is the number of internal parameters.

[0214] As a problem, in the evaluation of image quality, the criteria for good or bad image quality vary depending on the user, and sometimes the calculation result of the pre-designed image quality evaluation value Ei does not match the user's evaluation. In contrast, in Embodiment 1, it is noted that the image quality evaluation value Ei changes depending on the internal parameter Qi_m used in the calculation of the image quality evaluation value Ei. In Embodiment 1, the internal parameter Qi_m is determined so that the calculation result of the image quality evaluation value Ei is consistent with the user evaluation. Thereby, the situation where the calculation result of the image quality evaluation value Ei does not match the user evaluation is eliminated.

[0215] Fig.16 is an explanatory diagram regarding the determination of the above internal parameter Q. (A) shows a processing example, and (B) shows a GUI screen example corresponding to (A). In (A), the computer system 3 inputs the evaluation target image group 1600. The image group 1600 is a plurality of images with different image qualities taken at the same position of the wafer. In step S1601, the computer system 3 provides the GUI 1601 to the user, and the user inputs an evaluation of the image quality through the GUI 1601.

[0216] In Embodiment 1, as shown in (B), as the GUI 1601 for inputting the evaluation from the user related to the image quality, it has the GUI screen 1601. The computer system 3 displays, in the GUI screen 1601, an evaluation partial area 1601A for calculating the image quality evaluation value Ei of the object and a plurality of images (user evaluation images) 1601B with different image qualities. The evaluation partial area 1601A corresponds to the evaluation partial area Si determined as described in the step S2 for the images in the image group 1600. In this example, as shown by the white part, the wiring upper surface area and the dark part area become the calculation areas for the image quality evaluation value Ei. The user evaluation images 1601B display, for example, 3 images (#1, #2, #3) with different image qualities corresponding to the image group 1600.

[0217] An image quality evaluation interface 1601C is provided in the user evaluation images 1601B. The user visually confirms the user evaluation images 1601B in the GUI screen 1601, makes a subjective evaluation, and inputs the ranking of each image quality in the image quality evaluation interface 1601C. In other words, the user determines which image has a relatively higher or lower image quality for the user evaluation images 1601B and inputs the ranking. The image quality evaluation interface 1601C can, for example, select the ranking in a list box. Without limitation, the image quality evaluation interface 1601C can also be an interface where the user drags the images and arranges them in the order of image quality. After the input of the image quality ranking is completed, the user presses the evaluation result output button 1601D. Thereby, the computer system 3 saves / outputs the user evaluation result including the image quality ranking set by the user as the user setting 1605 ( Figure 2 the user setting information 202 in). The step S1601 of the image quality user evaluation in the GUI screen 1601 corresponds to Figure 2 the step S7 of the user setting in.

[0218] After saving / outputting the user setting 1605, in step S1602, the computer system 3 calculates the image quality evaluation value Ei for each image for the user evaluation images 1601B (the corresponding image group 1600) displayed to the user on the GUI screen 1601. At this time, the computer system 3 uses the initially prepared internal parameters 1607 to calculate the image quality evaluation value Ei of the object for each image.

[0219] Next, in step S1603, the computer system 3 compares the image quality evaluation value Ei of each image obtained in step S1602 with the user evaluation result in the user setting 1605, that is, the image quality ranking. In this comparison, the computer system 3 confirms whether the magnitude relationship of the image quality evaluation value Ei of each image is consistent with the user-set image quality ranking. In the case of inconsistency, it is determined that the internal parameters need to be adjusted.

[0220] Then, in step S1604, the computer system 3 uses the obtained comparison result to determine the internal parameter Qi_m of the image quality evaluation value Ei. At this time, the computer system 3 determines the value of the internal parameter Qi_m so that the magnitude relationship of the image quality evaluation value Ei of the user evaluation image 1601B displayed to the user is consistent with the image quality ranking set by the user, and there are significant differences in the image quality evaluation values Ei of these respective images.

[0221] Examples of the internal parameter Qi_m of the image quality evaluation value Ei are as follows. In the case of calculating the image quality evaluation value Ei using rule-based processing, for example, in the case of calculating the sharpness of the edges in the image (image quality evaluation value E1) based on the luminance gradient, examples of the internal parameter Qi_m include the filter size of the Sobel filter. In addition, in the case of calculating the image quality evaluation value Ei using a machine learning model, examples include model parameters such as the weights (in other words, coupling coefficients) between nodes of the network and biases.

[0222] Fig.17 (A) shows an example of the data structure of the data on the internal parameter Qi_m of the image quality evaluation value Ei. Such internal parameter Qi data / information is, for example, managed as data in the form of being included in the image quality evaluation value information 203 ( Figure 3 ). Fig.17 The table of (A) has a column for the image quality evaluation value Ei and a column for the internal parameter Qi_m. For example, the image quality evaluation value E1 has internal parameters Q1_1 to Q1_Nm as multiple internal parameters Qi_m. Parameter values are set in each internal parameter Qi. In addition, it is also possible to set the use status (in other words, on / off) in each internal parameter Qi.

[0223] [Element image quality evaluation value related to the calculation of Ei]

[0224] Use Fig.18 to explain the "element image quality evaluation value" related to the calculation of the image quality evaluation value Ei. In Embodiment 1, one of the features of the image quality evaluation value Ei is that it can be expressed by a combination of element image quality evaluation values Di_k (k = 1 to Nd). Nd is the number of element image quality evaluation values. In addition, in Embodiment 1, one of the features of the internal parameter Qi_m of the image quality evaluation value Ei is that it can be a parameter that specifies the combination method of the element image quality evaluation values Di_k.

[0225] As a problem, it is difficult to change the elements of the image features considered in the calculation of the image quality evaluation value Ei, and thus it is difficult to correspond to the changes in the semiconductor structure to be evaluated. In contrast, in the first embodiment, the image quality evaluation value Ei is expressed by a combination of the element image quality evaluation values Di_k. Thereby, the image quality evaluation value Ei can be changed in accordance with the changes in the semiconductor structure and its changes, and appropriate image quality quantification can be performed.

[0226] Fig.18 It is an explanatory diagram related to the element image quality evaluation value Di. Fig.18 The processing example of can be applied to step S3 ( Figure 2 ). In addition, this processing example can be applied according to the image quality evaluation value Ei. It is assumed that an evaluation target image 1801 and an evaluation partial area 1802 for calculating the image quality evaluation value Ei corresponding to the image 1801 are given. In this case, in step S1801, the computer system 3 takes the image 1801 and the evaluation partial area 1802 as inputs, and calculates a plurality (Nd) of element image quality evaluation values Di_k related to the target image quality evaluation value Ei. Then, in step S1802, the computer system 3 takes the Nd element image quality evaluation values Di_k and the internal parameters (Qi_m) 1804 related to the target image quality evaluation value Ei as inputs, and calculates the image quality evaluation value Ei formed by combining these Nd element image quality evaluation values Di_k. This calculation is expressed as Ei = H(Di_k, Qi_m). H is a function of Di. The internal parameter (Qi_m) 1804 referred to in step S1802 is information specifying the combination method of the element image quality evaluation values Di_k. As a result of step S1802, an image quality evaluation result 204 including the image quality evaluation value Ei is obtained.

[0227] Fig.17 (B) of shows a structural example of the data of the element image quality evaluation value Di_k regarding the image quality evaluation value Ei. Such data / information of the element image quality evaluation value Di_k is, for example, data-managed in a manner included in the aforementioned image quality evaluation value information 203 ( Figure 3 ). Fig.17 The table of (B) has a column for the image quality evaluation value Ei and a column for the element image quality evaluation value Di_k. For example, the image quality evaluation value E1 has element image quality evaluation values D1_1 to D1_Nk as a plurality of element image quality evaluation values Di_k. Parameter values are set in each element image quality evaluation value Di_k. In addition, it is also possible to set the use state (in other words, on / off, etc.) in each element image quality evaluation value Di_k.

[0228] When one of the internal parameters Qi_m of the image quality evaluation value Ei is set as the parameter of the combination method for the specified element image quality evaluation value Di_k, it is as follows. For example, one of the internal parameters Q1_m of the image quality evaluation value E1 is set as the parameter of the combination method for the specified element image quality evaluation value D1_k. An example of the combination method is the method of using all of the element image quality evaluation values D1_k (k = 1 to Nk). In this case, one internal parameter Q1_m specifying this combination method is information indicating the use of all of the element image quality evaluation values D1_k (k = 1 to Nk). For example, when the use or non-use of each element image quality evaluation value D1_k is represented by on (1) / off (0), this internal parameter Q1_m becomes a value such as {1, 1, …, 1}.

[0229] Not limited to this, the internal parameter specifying the combination method may also be the information of the function of Fig.18 step S1802 specified. In addition, setting information may be set separately from the internal parameter. In Figure 6 step S63 for example, the calculation method of the image quality evaluation value Ei including the combination method or function of the above-mentioned element image quality evaluation value D1_k is set.

[0230] A specific example related to the above-mentioned element image quality evaluation value Di is as follows. In the case where the image quality evaluation value E3 “noise suppression degree” calculated Figure 3 is used as the image quality evaluation value Ei, the case of obtaining the element image quality evaluation values D3_k (for example, D3_1, D3_2) constituting this image quality evaluation value E3 is described. First, in order to calculate the amount of brightness change caused by noise in the object image 1801, the computer system 3 calculates the local average value of the brightness of the object pixel and the surrounding pixels for each pixel of the image 1801 in the evaluation partial area 1802, and calculates the difference between the brightness of the object pixel and the local average value. The computer system 3 obtains the brightness change caused by noise as the first element, that is, the element image quality evaluation value D3_1, by calculating the differences obtained by the above process for all pixels.

[0231] On the other hand, the impression given to the user by the noise generated in the image 1801 changes not only according to the brightness change caused by the noise but also according to the contrast of the image. Therefore, the computer system 3 calculates the contrast of the image 1801 as the other second element, that is, the element image quality evaluation value D3_2.

[0232] The image quality evaluation value E3 “degree of noise suppression” is constituted by, for example, combining the above-described elements, the image quality evaluation value D3_1 (“brightness change caused by noise”) and the element image quality evaluation value D3_2 (“contrast of the image”). This combination method can be specified by one of the internal parameters Qi_m. Thereby, the computer system 3 can calculate a preferable image quality evaluation value E3 “degree of noise suppression” that takes into account both the brightness change caused by noise and the contrast of the image. In the above example, in order to calculate the “degree of noise suppression”, the “brightness change caused by noise” (the above difference) and the contrast of the image are used, but it is not limited thereto, and other element image quality evaluation values Di can be used to calculate the image quality evaluation value Ei.

[0233] [Calculation of Ei using machine learning]

[0234] Refer to Fig.19 , and a method for calculating the image quality evaluation value Ei using machine learning will be described. In Embodiment 1, Figure 2 One of the features of the step S3 of calculating the image quality evaluation value is that a machine learning model can also be used to calculate the image quality evaluation value Ei.

[0235] As a problem, it is difficult to determine in advance the image features related to the calculation of the evaluation value for image quality such as the naturalness of the image, which is determined subjectively by humans, and it is difficult to perform quantitative evaluation. In contrast, in Embodiment 1, in step S3, when using machine learning, a machine learning model that has learned human subjectivity and the like as teacher data is used to calculate the image quality evaluation value Ei. Therefore, the evaluation value can be calculated without determining the image features in advance.

[0236] Fig.19 Indicates an example of the machine learning model applied to Figure 2 step S3, particularly step S31. As a method for installing this machine learning, for example, well-known deep learning can be applied. In a specific example, this model can apply a convolutional neural network (CNN). In the Fig.19 example, a machine learning model using a three-layer structured CNN is shown. In this model, g represents the input image, and F(g) is the output based on the estimation result of the model. In addition, F1(g) and F2(g) represent the intermediate data between the input and the estimation result. The intermediate data F1(g) and F2(g) and the estimation result F(g) are calculated by the following equations 1 to 3.

[0237] Equation 1: F1(g) = max(0, W1 * g + B1)

[0238] Equation 2: F2(g) = max(0, W2 * F1(g) + B2)

[0239] Equation 3: F(g) = average(W3 * F2(g) + B3)

[0240] Here, * represents a convolution operation. W1 represents n1 filters of size u0 × f1 × f1. u0 represents the number of channels of the input image, and f1 represents the size of the spatial filter. By convolving the input image g with the u0 × f1 × f1 filters nu1 times, a nu1 - dimensional feature map is obtained. B1 is an n1 - dimensional vector, which is the bias component corresponding to the n1 filters. Similarly, W2 is nu2 filters of size nu1 × f2 × f2, B2 is an n2 - dimensional vector, W3 is 1 filter of size nu2 × f3 × f3, and B3 is a one - dimensional vector. u0 is a value determined by the number of channels of the evaluation target image. In addition, f1, f2, nu1, and nu2 are values determined by the user before the learning sequence. For example, f1 = 9, f2 = 5, nu1 = 128, and nu2 = 64.

[0241] In the example of Embodiment 1, the estimated result F(g) obtained by Equation 3 is used as the image quality evaluation value Ei. In addition, the evaluation partial region Si can also be used as a mask in F1, F2, and F. In addition, not limited to the above - mentioned CNN structure example, other structures can also be used. For example, the number of layers can be changed, a network with 4 or more layers can be used, etc., and it can also be a structure with skip connections. In addition, in Equation 3, an example where the average value is used in the calculation of F(g) is shown, but the maximum value or the minimum value can also be calculated.

[0242] [Calculation of the comprehensive evaluation value C (Step S32)]

[0243] Use Fig. 20 , to explain the method of calculating the comprehensive evaluation value C from multiple image quality evaluation values Ei in Step S32 of Figure 2 . In Embodiment 1, one of the features is that in Step S32, the comprehensive evaluation value C composed of a combination of the image quality evaluation values Ei is calculated.

[0244] Fig. 20 is an explanatory diagram related to the calculation of the comprehensive evaluation value C. As a result of the processing (especially Step S703) as described above for the computer system 3, information on multiple image quality evaluation values Ei of the target image 201 is obtained. These multiple image quality evaluation values Ei are the multiple image quality evaluation values Ei required for calculating the comprehensive evaluation value C. Regarding which image quality evaluation value Ei to use for calculating the comprehensive evaluation value C, it is set as part of the calculation method in Step S64 of Figure 7 Figure 6 .

[0245] ​Next, in step S2001, the computer system 3 calculates a comprehensive evaluation value C by combining the above-mentioned multiple image quality evaluation values Ei according to the calculation method 2005 of the comprehensive evaluation value C. The calculation method 2005 of the comprehensive evaluation value C is Figure 6 the set value of step S64 of Figure 6 , for example, information specifying which image quality evaluation value Ei among the multiple image quality evaluation values Ei (i = 1 to N) to use and the combination method of the multiple image quality evaluation values Ei used. The comprehensive evaluation value C can be expressed as a function of the image quality evaluation value Ei, for example. The calculated comprehensive evaluation value C is saved together with the associated image quality evaluation value Ei in the form included in the image quality evaluation result 204.

[0246] The comprehensive evaluation value C is calculated using, for example, the following formula 4. The following Fc is a function with the image quality evaluation value Ei as the independent variable. In this case, the calculation method 2005 of the comprehensive evaluation value C becomes a parameter related to the function Fc.

[0247] Formula 4: C = Fc(Ei)

[0248] A specific example of the calculation of the comprehensive evaluation value C is as follows. For example, generally, it is desirable to have high sharpness of the observed image. However, in the case of applying shooting and image quality improvement processing to improve sharpness, ringing may be amplified. That is, sharpness and ringing are in a trade-off relationship. The above sharpness is associated with the image quality evaluation value E1 "sharpness" in Figure 3 and the above ringing is associated with the image quality evaluation value E8 "degree of ringing suppression" in Figure 3 . Therefore, it is sometimes difficult to comprehensively judge the quality of the image by observing only each image quality evaluation value Ei.

[0249] Therefore, in Embodiment 1, for example, a comprehensive evaluation value C is calculated by combining the image quality evaluation value E1 of sharpness and the image quality evaluation value E8 of the degree of ringing suppression. Through this comprehensive evaluation value C, the image quality can be comprehensively observed / judged. In addition, for example, by making the combination method of the image quality evaluation value E1 and the image quality evaluation value E8 variable according to a parameter, it is also possible to set which image quality is emphasized as a priority. Thus, a comprehensive evaluation value C that matches the user's preference can be calculated.

[0250] [Calculation of Comprehensive Evaluation Value C: Priority]

[0251] In Embodiment 1, one of the features is that the comprehensive evaluation value C can also be calculated using the priority assigned / set for the image quality evaluation value Ei.

[0252] As a problem, there are cases where it is desired to emphasize a specific image quality evaluation value Ei among multiple image quality evaluation values Ei, etc., and it is difficult to perform a comprehensive evaluation corresponding to such cases. In contrast, in Embodiment 1, priorities are assigned / set to each image quality evaluation value Ei according to the image quality emphasized by the user, and a comprehensive evaluation value C is calculated through a combination of these image quality evaluation values Ei.

[0253] In Fig. 20 as well, the calculation of the comprehensive evaluation value C using priorities is also shown. It is possible to set in Figure 6 step S64 Fig. 20 the calculation method 2005 of the comprehensive evaluation value C, which combination method of which image quality evaluation value Ei, etc. is used in the calculation of the comprehensive evaluation value C, and it is also possible to set the priority of the used image quality evaluation value Ei and the calculation method of the comprehensive evaluation value C using priorities, etc. Here, the priority of each image quality evaluation value Ei is set as ri (i = 1 to N).

[0254] As an example, the comprehensive evaluation value C is defined as a combination of using E1 and E8. C = Fc(E1, E8). Priorities ri (r1, r8) are set for each used image quality evaluation value Ei (E1, E8). For example, weight coefficients corresponding to the priorities ri are set for each used image quality evaluation value Ei. For example, it is the weight w1 corresponding to the priority r1, the weight w8 corresponding to the priority r8, etc. The calculation formula of the comprehensive evaluation value C is set as a function Fc defined using the weight (in other words, coefficient) corresponding to the priority ri. For example, C = Fc(E1, E8) = E1×w1 + E8×w8. In addition, in this example, it is defined such that the larger the value of Ei, the higher the evaluation, the higher the priority ri, the larger the value of the weight wi, and the higher the value of the comprehensive evaluation value C, the higher the comprehensive evaluation. In addition, this example is set to add weights, but it is not limited to this, and definitions using arithmetic operations, etc. can be performed. Similarly, the comprehensive evaluation value C can be defined through operations using multiple image quality evaluation values Ei and priorities ri.

[0255] In the case where the user, for example, wants to emphasize the image quality evaluation value E1 "sharpness" for evaluation, it is only necessary to set a larger value for the priority r1 (corresponding weight w1). In addition, in the case where the user, for example, wants to emphasize the image quality evaluation value E8 "ringing suppression degree" for evaluation, it is only necessary to set a larger value for the priority r8 (corresponding weight w8). The user settings related to the calculation method 2005 of the comprehensive evaluation value C including the above setting of the priority ri can also be realized through the GUI screen. In addition, as in the above example, in the case where there are multiple calculation methods of the comprehensive evaluation value C according to the difference in the emphasized image quality, by pre - defining / setting these multiple calculation methods in advance, the user can select and apply from them during evaluation. In addition, it can also be in Figure 3 Items for setting the priority ri in the image quality evaluation value information 203.

[0256] [Examples of utilization of the image quality evaluation value Ei]

[0257] In Figure 2 The image quality evaluation value Ei or the comprehensive evaluation value C calculated in step S3 can be utilized in various ways. As examples of the utilization of the image quality evaluation value Ei and the comprehensive evaluation value C, the following can be cited. The details of each will be described later.

[0258] (Example 1) Adjustment / Determination of the parameter P of the image quality improvement process (image quality improvement engine) ( Fig.21 )

[0259] (Example 2) Correction of a partial area for evaluation ( Fig.28 )

[0260] (Example 3) Evaluation (comparison of image quality between images) ( Fig.29 )

[0261] (Example 4) Defect detection ( Fig.30 )

[0262] (Example 5) Shape measurement ( Fig.31 )

[0263] (Example 6) Determination of shooting conditions ( Fig.32 )

[0264] (Example 7) Device monitoring ( Fig.33 )

[0265] At least one of the functions corresponding to the above utilization examples is installed in the specimen observation system and method of Embodiment 1. When multiple functions are installed, the user can select functions through user settings or the like for utilization.

[0266] [Adjustment / Determination of the parameter P of the image quality improvement process]

[0267] Using Fig.21 etc., the adjustment / determination of the parameter P of the image quality improvement process (image quality improvement engine) using the image quality evaluation value Ei will be described. In Embodiment 1, one of the features is that it may also have a step of determining the parameter P of the image quality improvement engine based on the image quality evaluation value Ei, and a step of high-qualityizing an image using the image quality improvement engine with the determined parameter P.

[0268] As a problem, in the adjustment of the parameters of the image quality improvement engine, it is impossible to quantify the image quality of the image output with a certain parameter. Therefore, it is impossible to judge the quality of the output image, and it is difficult to adjust the parameters. On the contrary, in Embodiment 1, by using the image quality evaluation value Ei, it is possible to quantify the image quality, and it is possible to judge the quality of the image according to the magnitude of the image quality evaluation value Ei. Therefore, the parameters of the image quality improvement engine are adjusted according to the image quality evaluation value Ei. Thereby, it is possible to determine the parameters of the image quality improvement engine that outputs an image with excellent image quality.

[0269] First, Fig.21 Fig. shows a structural example of the overall process and functional blocks including the image quality improvement process (image quality improvement engine) and the calculation of the image quality evaluation value Ei. In Fig.21 it particularly shows the adjustment / determination of the parameter P of the image quality improvement process (corresponding image quality improvement engine) using the image quality evaluation value Ei. First, as shown on the left, the step S2101 of calculating the image quality evaluation value Ei for the input object image 201 is the same as the steps S2 and S3 of the Figure 2 above. As a result of step S2101, an image quality evaluation result 204 including the image quality evaluation value Ei is obtained.

[0270] When the object image 201 is enhanced to high image quality, in other words, the image quality is improved, the computer system 3 performs an image quality improvement process using the image quality improvement engine in step S2100. As a result of step S2100, an image 2101 is obtained as an image quality improved image (in other words, an image quality improved image). The image quality improvement process in step S2100 includes, for example, processes for improving resolution and S / N (signal-to-noise ratio). As an example, in image restoration processing, a clear and high-S / N image is estimated from the captured image in order to remove resolution degradation and noise overlap.

[0271] Regarding the image quality improvement engine for the high image quality process in step S2100, the computer system 3 has, for example, a rule-based image quality improvement engine 2100A and a machine learning-based image quality improvement engine 2100B. In step S2100, the computer system 3 performs the high image quality process using the image quality improvement engine selected by the setting in step S6 of the Figure 6 above. This example is a case where there are two types of image quality improvement engines, but it is not limited to this, and it may be a case where the system has only one type of image quality improvement engine according to the system.

[0272] The image quality improvement process of the image quality improvement engine in step S2100 has one or more parameters constituting the image quality improvement process, and this parameter (in other words, the image quality improvement process parameter) is set to P. A plurality of parameters P are set to Pi (i = 1 to Np). Np is the number of parameters.

[0273] In step S2100, when using the rule-based image quality improvement engine 2100A, image processing parameters are used as parameter P. When using the machine learning-based image quality improvement engine 2100B, machine learning model parameters, such as the parameters of a CNN, are used as parameter P.

[0274] In the image quality improvement process using the image quality improvement engine in step S2100, first, the image quality improvement process is performed according to the image quality improvement process setting information 2110 including parameter P set in Figure 6 step S6 (setting step).

[0275] Furthermore, the computer system 3 can also use the image 201 or the image 2101 as the high-definition image to calculate the image quality evaluation value Ei in step S2101, and use the calculated image quality evaluation value Ei to adjust / determine the parameter P of the image quality improvement process in step S2102. In this example, after temporarily obtaining the image 2101 as the high-definition image, the computer system 3 calculates the image quality evaluation value Ei for the image 2101 and adjusts / determines the parameter P in step S2102. As a result of step S2102, the adjusted parameter P2104 is obtained. In the adjustment of step S2102, the adjusted parameter P2104 is obtained according to the setting information 2110 of parameter P. The setting information 2110 including parameter P can also be updated by the adjusted parameter P2104.

[0276] When the adjusted parameter P2104 is obtained in step S2102, in step S2100, the input image 201 is subjected to image quality improvement processing using the adjusted parameter P2104, and as a result, the image 2101 as the high-definition image is obtained.

[0277] In Embodiment 1, there is a function of adjusting / determining the parameter P2104 of the image quality improvement process in step S2100 as described above Fig.21 to a more appropriate value, that is, the image 2101 becomes a more appropriate image quality (especially step S2102).

[0278] For example, assume that the image 2101 is obtained as a result of the image quality improvement process for the rule-based image quality improvement engine 2100A in step S2100. In step S2101, as a result of weighting the image quality evaluation value E1 related to the sharpness of the edge, a small value (i.e., a poor sharpness evaluation) is output as the sharpness. In this case, in step S2102, the computer system 3 selects, as an adjustment of the parameter P for the image quality improvement process, a parameter P that increases the number of times of the image restoration process for removing blur, or a parameter P that strongly applies the edge enhancement process. The adjusted parameter P is applied to the image quality improvement process in step S2100. After that, the computer system 3 performs the image quality improvement process in step S2100 with the adjusted parameter P on the same image 201 as the initial one (or it can also be other images in the future), thereby obtaining an image 2101 with improved image quality (e.g., sharpness) compared to the previous time.

[0279] The computer system 3 repeats as needed Fig.21 such steps as the image quality improvement step S2100, the step S2101 of calculating the image quality evaluation value Ei, and the step S2102 of adjusting / determining the parameter P. Through such a cycle, the parameter P2104 of the image quality improvement process can be optimized so that an image 2101 with the desired image quality by the user can be obtained according to the result of the image quality improvement process in step S2100.

[0280] In the above example, the computer system 3 calculates one or more image quality evaluation values Ei in step S2101 and uses the image quality evaluation value Ei to adjust the parameter P, but it is not limited to this. The computer system 3 can also calculate, in step S2101, a comprehensive evaluation value C (e.g., a comprehensive evaluation value C considering the image quality improvement process) as in the step S32, and use the comprehensive evaluation value C to adjust the parameter P.

[0281] In addition, examples of the parameter P of the image quality improvement process (e.g., image restoration process) corresponding to the shooting condition 205 of the image 201 include the chromatic aberration coefficient, the spherical aberration coefficient, etc.

[0282] [Adjustment / Determination of Parameter P Using the Target Value of Ei]

[0283] Use Fig. 22 , regarding the above Fig.21 step S2102 of adjusting / determining the parameter P, a method of using the target value of the image quality evaluation value Ei (recorded as the region target value 2201) will be described. In Embodiment 1, one of the features of the computer system 3 is that for some or all of the image quality evaluation values Ei among the multiple image quality evaluation values Ei, the region target value 2201 can be set individually, and the parameter P of the image quality improvement process can be adjusted / determined so that the image quality evaluation value Ei approaches the region target value 2201.

[0284] As a problem, even when generating / outputting an image with the highest (or lowest, under the definition that the smaller the value of Ei, the better the image quality) image quality evaluation value Ei in the high image quality step S2100, an image with the desired image quality by the user may not be obtained. For example, when setting the noise suppression degree of the noise generated in the image as one of the image quality evaluation values Ei ( Figure 4 the image quality evaluation value E3 in ), the image quality evaluation value Ei is the best in an image without noise. However, since the user usually observes an image with noise, an image with the best image quality evaluation value Ei may be judged as unnatural. Therefore, in Embodiment 1, the computer system 3 individually sets a regional target value for some or all of the image quality evaluation values Ei, and adjusts / decides the parameter P so that the image quality evaluation value Ei approaches the regional target value.

[0285] Fig. 22 is a detailed processing example or a modified example related to Fig.21 step S2102, and is an explanatory diagram of the adjustment / decision of the parameter P of the image quality improvement process using the regional target value as the target value of the image quality evaluation value Ei. Specifically, step S2102 includes steps S2102A and S2102B. Here, the regional target value of each image quality evaluation value Ei is set as Ti.

[0286] First, as Fig.21 shown, for the image 201 taken for observation, the computer system 3 performs an image quality improvement process in step S2100, and as a result, generates / outputs the image 2101. The computer system 3 uses this image 2101 as an input, and in step S2101, calculates the image quality evaluation value Ei of this image 2101 to obtain the image quality evaluation result 204. Then, in step 2102A of step S2102, the computer system 3 calculates the difference between the image quality evaluation value Ei and the previously set regional target value Ti2201. This difference is the difference of each image quality evaluation value Ei. Then, in step S2102B, the computer system 3 adjusts / decides the parameter P2104 of the image quality improvement process so that this difference becomes smaller. Regarding step S2102B, it can be set as an adjustment such that the difference is less than a threshold value, or it can be set as an adjustment such that the difference obtained this time is less than the difference obtained last time. The computer system 3 repeatedly performs such a series of processes, and saves / outputs the parameter P that minimizes the difference between the image quality evaluation value Ei and the regional target value Ti2201 as much as possible.

[0287] In addition, in Fig. 22In it, an example of representing the regional target value Ti by a table is shown. This table has a column for the image quality evaluation value Ei and a column for the regional target value Ti. The setting information of such a regional target value Ti2201 can also be managed as data in a manner included as part of the image quality evaluation value information 203 described above.

[0288] In addition, in Fig. 22 a structural example of the image quality improvement process based on the image quality improvement engine 2100 in step S2100 is illustrated. In this example, the image quality improvement engine 2100 is structured to sequentially perform a plurality of processes (processes 1 to n) 2210 in a process-like manner and has parameters P (P1 to Pn) related to each process 2210. In step S2102, the computer system 3 can also determine the values of these respective parameters P.

[0289] Regarding the method for setting the regional target value 2201, as Figure 6 a part of step S6 of the setting, the user can directly input / set the regional target value 2201 through the GUI screen. Alternatively, a method automatically set by the computer system 3 can also be applied. As the method, for example, the following method can be cited: automatically setting the image quality evaluation value Ei calculated from an image that satisfies the user's desired image quality and has an appearance similar to the observation image 201 as the regional target value 2201.

[0290] The above example shows the case of using the target value of the image quality evaluation value Ei, but similarly, it is also possible to set the target value (comprehensive target value) regarding the comprehensive evaluation value C and use the target value of the comprehensive evaluation value C to determine the parameter P.

[0291] [Adjustment / Determination of Parameter P Using the Comprehensive Evaluation Value C]

[0292] Using Fig.23 , a method for adjusting / determining the parameter P of the image quality improvement process in step S2102 using the comprehensive evaluation value C calculated in step S32 (specifically Figure 2 ) in Fig. 20 is described. In Embodiment 1, Fig.21 one of the features of step S2102 in Fig.21 is that the parameter P of the image quality improvement process can be adjusted / determined so that the value of the comprehensive evaluation value C is improved.

[0293] As a subject, in consideration of cases where multiple image quality evaluation values Ei are taken into account, cases where importance is attached to a specific image quality evaluation value Ei among the multiple image quality evaluation values Ei, etc., appropriate parameters P for image quality improvement processing are determined. On the other hand, in Embodiment 1, the parameters P for image quality improvement processing are adjusted / determined so that the value of the comprehensive evaluation value C is improved. In particular, the parameters P for image quality improvement processing are adjusted / determined so that the value of the comprehensive evaluation value C calculated using the priority of the above-described image quality evaluation value Ei is improved.

[0294] Fig.23 It is an explanatory diagram related to the adjustment / determination of the parameter P of the image quality improvement processing using the comprehensive evaluation value C. Fig.23 It is Fig.21 The deformed process. First, the computer system 3 performs high-quality conversion on the image 201 captured for observation in step S2100, and generates / outputs the image 2101. Next, in step S2301 (corresponding to the above-described steps S2 and S31), the computer system 3 calculates the image quality evaluation value Ei of the image 2101. Next, in step S2302 (corresponding to the above-described step S32), according to the calculation method 2005 of the comprehensive evaluation value C ( Fig. 20 ), the computer system 3 calculates the comprehensive evaluation value C formed by combining the above-calculated image quality evaluation values Ei. The calculated comprehensive evaluation value C is included in the image quality evaluation result 204 together with the image quality evaluation value Ei.

[0295] Next, in step S2303, the computer system 3 adjusts / determines the parameter P2104 of the image quality improvement processing (step S2100) so that the value of the above comprehensive evaluation value C is improved. Regarding step S2302, it can be set as an adjustment such that the C value is less than the threshold value, or it can be set as an adjustment such that the obtained C value is less than the previously obtained C value. The computer system 3 determines the parameter P2104 such that the image 2102 with the best possible comprehensive evaluation value C can be generated / output by repeatedly performing such a series of processes.

[0296] In addition, in the example of Fig.23 , it can also be the same as Fig. 22Similar to the example, a comprehensive target value 2305 is also set for the comprehensive evaluation value C as the target value. In step S2303, the parameter P is adjusted / decided so that the difference between the comprehensive evaluation value C and the comprehensive target value 2305 becomes smaller. In setting the comprehensive target value 2305, similar to the regional target value 2201, the user can also directly input / set the comprehensive target value 2305 through the GUI screen. Or, similar to the regional target value 2201, a method of automatically setting the comprehensive target value 2305 by the computer system 3 can also be applied. In addition, in setting the comprehensive target value 2305, a comprehensive target value 2305 formed by combining the regional target value 2201 can also be set. If the comprehensive target value 2305 is set as Tc, then Tc can be defined by the function Ft(Ti) of the regional target value Ti.

[0297] [Rule-based image quality improvement engine]

[0298] Use Fig.24 , for Fig.21 The image quality improvement engine of step S2100 for image quality improvement processing ( Fig.21 The image quality improvement engine 2100 in) is a method in the case of the image quality improvement engine 2100A composed of rule-based image processing will be described. In this case, the parameter P2104 of the image quality improvement processing is the image processing parameter of the image quality improvement engine 2100A.

[0299] As an issue, in the case of achieving high image quality using a machine learning-based image quality improvement engine, it is difficult to finely adjust the image quality of the output image in the image quality improvement engine of the learned model. In contrast, in Embodiment 1, in the case of using the rule-based image quality improvement engine 2100A, it is possible to finely adjust the image quality of the image 2102 after the image quality improvement processing.

[0300] Fig.24 It is an explanatory diagram of the rule-based image quality improvement engine 2100A. In this example, the computer system 3 performs image quality improvement processing on the image 201 taken for observation using the rule-based image quality improvement engine 2100A, and as a result, obtains the image 2102. In the image quality improvement engine 2100A, according to n processes 2210 (process 1 to process n) designed in advance (the same as Fig. 22 ). Such processes 2210 may include, for example, shadow correction processing, noise removal processing, image restoration processing, etc.

[0301] As Fig.24As shown in the table in , in the parameter P2104 of the image quality improvement process of the computer system 3, the ON (performed) / OFF (not performed) status and the values of the parameters P (P1 to Pn) for each of the n processes 2210 are set. In this example, for the target image 201, process 1, which is a filtering process, is set to ON, and the filter size is set to 3×3 as the parameter P1 of process 1. On the other hand, process 2 is set to OFF. Therefore, in the case of this setting of the parameter P, process 2 is not performed on the image 201 during the image quality improvement process in step S2100.

[0302] As a modification example, it can also be set as follows. In Fig.24 's example, the rule-based image quality improvement engine 2100A generates intermediate images 2401 between the respective processes 2210 of the multiple processes 2210. For example, a first intermediate image 2401 is generated as the output of process 1, and a second intermediate image 2401 is generated as the output of process 2. In the modification example, as Fig.24 shown, the computer system 3 can set the intermediate image 2401 generated in the middle of the image quality improvement process of the image quality improvement engine 2100A as the target for calculating the image quality evaluation value Ei. In addition, it is also possible to set the intermediate image 2401 generated by a specific multiple processes as the target and calculate the image quality evaluation value Ei. For example, it can be cited that the intermediate image 2401 generated by only executing process 2 and process 3 is set as the target to calculate the image quality evaluation value Ei. The setting for this can also be set in step S6. For example, it is possible to set which intermediate image 2401 of which process 2210 is set as the target to calculate which image quality evaluation value Ei, etc. As described above, the image quality evaluation value Ei calculated thereby can be used for the adjustment / decision of the parameter P.

[0303] [Machine learning-based image quality improvement engine]

[0304] Using Fig.25 , for Fig.21 the method in the case where the image quality improvement engine in step S2100 is an image quality improvement engine 2100B composed of machine learning-based image processing given by, for example, a CNN will be described. In this case, the parameter P2104 of the image quality improvement process is, for example, model parameters such as the weights (in other words, coupling coefficients) between nodes of the CNN network, biases, etc.

[0305] As an issue, it is sometimes difficult to design in advance the processes required to improve the image quality. Since the appearance changes of the semiconductor pattern that is the object in the observed image are diverse, it is difficult to design a rule-based image quality improvement engine that can correspond to all observed images. In contrast, in Embodiment 1, it is possible to use the image quality improvement engine 2100B composed of machine learning-based image processing given by, for example, a CNN to handle this.

[0306] Fig.25 This is an explanatory diagram of the machine learning-based image quality improvement engine 2100B. In this example, the computer system 3 performs image quality improvement processing on the image 201 captured for observation using the machine learning-based image quality improvement engine 2100B, and as a result, obtains the image 2102. In this example, a CNN with a three-layer structure is applied to the machine learning-based image quality improvement engine 2100B as the machine learning model 2500. The model 2500 of this CNN calculates the intermediate data Z1(g), Z2(g), and the estimation result Z(g) for the input image g, and outputs the estimation result Z(g) as the high-quality image, that is, the image 2102. The computer system 3 sets the model parameters of the model 2500 of the image quality improvement engine 2100B in the image quality improvement processing parameter P2104. In this example, as shown in the table, in the parameter P2104, the bias of Z1(g) is set to 0.5, and the weight between Y1 - Y2 is set to 0.05.

[0307] In addition, the structure of the neural network of the image quality improvement engine 2100B is not limited to the above example, and other structures can also be applied. For example, the number of layers of the CNN can be changed, a network with four or more layers can be used, etc., or it can be set to a structure with skip connections.

[0308] In the adjustment / determination of the parameter P2104 regarding the machine learning-based image quality improvement engine 2100B ( Fig.21 in step S2102), for example, the image quality evaluation value Ei or the comprehensive evaluation value C is used as the loss function during the learning of the model 2500. Thereby, the model parameters for improving the image quality can be efficiently determined.

[0309] [Correction of the evaluation partial region]

[0310] Next, the function of correcting the evaluation partial region Si will be described. After the computer system 3 displays the image quality evaluation value Ei to the user through the GUI screen in Figure 2 step S4, through the operation of the user, a user setting including region correction (which refers to the correction of the evaluation partial region Si) is performed. The computer system 3 uses the user setting information 202 including the region correction information obtained through this user setting as the input information for step S6 (especially steps S21, S22) of the setting. Then, through the setting in step S6, it is determined that the processing contents of steps S21 and S22 are more appropriate. Through such a cycle, the evaluation partial region Si determined in step S22 is optimized, and as a result, a more appropriate image quality evaluation value Ei can be calculated.

[0311] Fig.28 The flow of step S7 showing the user setting including each function for the function of correcting the evaluation partial region Si. Fig.28 The flow is based on Figure 2 is based on the process of Fig.28 In the process of , according to the image quality evaluation value Ei obtained in the steps S1, S2, and S3 described above, after displaying the image quality evaluation value Ei, etc. on the GUI screen in step S4, step S7 for user settings is set. In this step S7, the computer system 3 provides a GUI screen for user settings, and the user performs an input operation while observing the GUI screen. Thus, using each function related to the setting, the result of the user setting is saved as the user setting information 202.

[0312] In Fig.28 In step S7 of , as an example of the function, there are step S71 for setting the image quality evaluation value Ei, step S72 for correcting the evaluation partial area Si, and step S73 for evaluation (comparing the image quality between images) described later. The user can use the functions selected through the GUI screen. Step S71 is equivalent to the function that allows the user to confirm / set the image quality evaluation information 203 such as Figure 3 . This function includes the function of adding and setting the image quality evaluation value Ei ( Figure 6 step S65 of ). Step S72 is equivalent to the above-mentioned area correction function. The user setting information 202 (including the image quality evaluation information 203 here) of the result of step S7 is used for Figure 6 the setting of step S6 of .

[0313] [GUI(1)]

[0314] First, an example of the GUI related to user settings (step S7), display of the image quality evaluation value Ei (step S4), etc. in the sample observation system of Embodiment 1 will be described.

[0315] In Embodiment 1, the object image 201, the evaluation partial area Si, the image quality evaluation value Ei, etc. are displayed to the user through the GUI screen. And, in Embodiment 1, corresponding to the function of the above-mentioned area correction, through the GUI screen, a GUI is provided that allows the user to manually correct the evaluation partial area Si. In addition, in Embodiment 1, a GUI is provided that allows the user to set the internal parameter Qi_m ( Fig.16 ), the area target value Ti ( Fig. 22 ), etc.

[0316] Fig.26 shows an example of the GUI screen having the above-mentioned GUI. Fig.26 The screen of is an example of the GUI screen related to the image quality improvement process of step S2101 ( Fig.21 ) and the functions associated therewith. Fig.26The screen has an interface area (image list bar) 2602, an interface area (observed image bar) 2603, an interface area (image quality of evaluation object bar) 2604, an interface area (image quality bar) 2605, an interface area (partial area for evaluation bar) 2606, an interface area (internal parameter bar) 2607, an interface area (image quality improvement parameter bar) 2608, an interface area (image quality improvement image bar) 2610, an interface area (evaluation value bar) 2611, an interface area (target value bar) 2612, an interface area (OK / NG bar) 2613, etc.

[0317] In the interface area (image list bar) 2602, a list of the images 201 (herein referred to as observed images) captured in the shooting step S2 ( Figure 2 ) is displayed. In the interface area (observed image bar) 2603, the observed image of the selected ID in the list is displayed.

[0318] In the interface area (image quality of evaluation object bar) 2604, a list of the image quality evaluation values Ei corresponding to the image quality of the evaluation object is displayed. In this list, the user can make a decision by, for example, selecting the image quality evaluation value Ei corresponding to the image quality of the target image in the check box. The image quality evaluation value Ei corresponding to the image quality selected in the check box can be specified in the interface area 2605. In the interface area (image quality bar) 2605, for example, when the user operates the list box, the image quality evaluation value Ei is displayed as an option, and the user can select the desired image quality evaluation value Ei. The image quality evaluation value Ei of the image quality selected in the interface area 2605 becomes the object in the interface area 2603, the interface area 2610, the interface area 2606, etc.

[0319] In the interface area (partial area for evaluation bar) 2606, the partial area Si for evaluation (determined in step S22) used in the calculation of the image quality evaluation value Ei selected in the interface area 2605 is displayed.

[0320] In the interface area (internal parameter bar) 2607, the internal parameter Qi_m of the selected image quality evaluation value Ei is displayed. In the interface area 2607, for example, when the user operates the list box, the displayed internal parameter Qi_m can be selected, and the parameter value (e.g., "0.1") of the internal parameter Qi_m is displayed.

[0321] In the interface area (image quality improvement parameter bar) 2608, regarding the image quality improvement engine ( Fig.21Adjustment / setting of parameter P2104 in step S2100) is performed based on the user's manual operation for adjustment / setting. In interface area 2608, each parameter P, for example, the parameter P2104 of the image quality improvement engine, is displayed by a slider. The user changes the position of the slider, thereby enabling adjustment of the values of each parameter P (e.g., #1 to #4). In addition, the default value of each parameter P is represented by a gray slider.

[0322] When the user presses the interface area (image quality improvement processing execution button) 2615, the image quality improvement processing (step S2100) using the value of parameter P represented by the slider in interface area 2608 operates. The image 2102, which is the image quality improved image generated / output through this image quality improvement processing, is displayed in the interface area (image quality improved image bar) 2610. The user can compare and confirm the observation image in interface area 2603, the image quality improved image in interface area 2610, and the evaluation partial area Si in interface area 2606.

[0323] In addition, when the user presses the interface area (automatic adjustment button) 2609, the computer system 3 performs automatic adjustment of the parameter P2104 of the image quality improvement engine (step S2100) to adjust / determine the value of this parameter. According to the value of the adjusted parameter P, the position of the slider of parameter P in interface area 2608 is automatically changed. The image 2102 after high image quality conversion by the image quality improvement engine using the automatically adjusted parameter P is displayed in interface area 2610.

[0324] The value of the image quality evaluation value Ei of the image quality improved image in interface area 2610 for the specified image quality in interface area 2605 is displayed in the interface area (evaluation value bar) 2611. In addition, the area target value Ti related to this image quality evaluation value Ei is displayed in the interface area (target value bar) 2612. The user confirms the image quality improved image in interface area 2610, the image quality evaluation value Ei in interface area 2611, and the area target value Ti in interface area 2612, and determines whether the image quality improved image with satisfactory image quality is obtained (set as OK / NG). The user inputs / selects this judgment result in the interface area (OK / NG bar) 2613.

[0325] In addition, in the interface area (design data bar) 2614, it is possible to specify the classification of the classification division area Rj required for determining the evaluation partial area Si ( Figure 2 in step S21) of the design data 206 used ( Figure 6 ).

[0326] [GUI(2)]

[0327] Fig. 27 Next, an example of a GUI screen related to the correction function of the evaluation partial area Si is shown. Fig. 27 The screen has a GUI related to the calibration of the evaluation partial area Si, and has interface areas 2716, 2717, 2718, 2719, 2720, 2721, 2724, etc. In addition, Fig. 27 The screen of Fig.26 can be integrated with the screen of

[0328] or can be other transformed GUIs. Fig. 27 Due to the influence of the structure of the semiconductor pattern, the size of the defective area in the image, etc., the evaluation partial area Si may not be determined as the area envisioned by the user. Therefore, in this case, the screen of

[0329] can be used to calibrate (in other words, manually correct) the evaluation partial area Si according to the user's input operation. Figure 2 In the interface area (inspection image list column) 2716, a list of images 201 (observation images) captured in the shooting step S1 ( Figure 2 ) is displayed. The observation image of the ID selected by the user in this list is displayed in the interface area (observation image column) 2717. In the interface area (image quality to be evaluated column) 2718, a list of image quality evaluation values Ei corresponding to the image quality to be evaluated is displayed. In the interface area (image quality column) 2719, the image quality evaluation value Ei selected by the user is displayed as an option, with the image quality evaluation value Ei selected in the interface area 2718. Each of these columns has the same function as the corresponding columns of Fig.26

[0330] In the interface area (evaluation partial area column before calibration) 2720, the evaluation partial area Si before calibration is displayed. In this column, the evaluation partial area Si (determined in step S22) used in the calculation of the image quality evaluation value Ei selected in the interface area 2719 is displayed.

[0331] The user specifies the image quality to be evaluated corresponding to the evaluation partial area Si to be corrected in the interface area 2719. In the interface area 2720, the evaluation partial area Si used in the calculation of the image quality evaluation value Ei corresponding to the image quality selected in the interface area 2719 is displayed. In this example, this evaluation partial area Si has a part where two adjacent line patterns are connected horizontally as the part a1 shown by the dotted line in the vertical line pattern (white part), and this part is the part that the user wants to correct. This part a1 corresponds to a part similar to the defective area in the observation image in the interface area 2717.

[0332] In addition, in the interface area (calibration input field) 2721, the area initially the same as the evaluation partial area Si displayed in the interface area 2720 is displayed for calibration input. The user inputs, in the interface area 2721, the boundary of the evaluation partial area Si (white part) as calibration information (in other words, manual correction information), for example, as shown by the dotted line 2722, using an input device such as a mouse. In this example, at the position a1, the dotted line 2722 is input in such a way as to draw a new boundary of the vertical line pattern so that the evaluation partial area Si does not include the part that horizontally connects the vertical line patterns to each other. The method of inputting the calibration information is not limited to this. In other examples, a method of coating the pixel areas included in the evaluation partial area Si or, conversely, a method of coating the pixel areas not included in the evaluation partial area Si may also be used.

[0333] After the above calibration information is input, the user presses the interface area (calibration execution button) 2723. Thereby, the computer system 3 uses the calibration information in the interface area 2721 to determine the evaluation partial area Si again (step S22), and displays the determined evaluation partial area Si in the interface area 2724. The calibrated evaluation partial area Si is displayed in the interface area (calibrated evaluation partial area field) 2724.

[0334] The user confirms the calibrated evaluation partial area Si in the interface area 2724. As a result, when it is judged that the calibration is good, the calibration is completed by pressing the interface area (calibration completion button) 2725. When the user judges that the calibration is not good, the user inputs calibration information again using the interface area 2721 and performs the calibration operation in the same way.

[0335] In this embodiment, an example in which the user can manually correct the boundary of the evaluation partial area Si is shown regarding the determination of the preferred evaluation partial area Si, but it is not limited to this. The user can also change the parameters of the segmentation method in step S21 of area division / classification, the machine learning model ( Figure 6 step S61). In addition, the user can also change the method of generating the evaluation partial area Si from the classification segmentation area Rj ( Figure 6 step S62).

[0336] [Evaluation Using the Image Quality Evaluation Value Ei]

[0337] Using Fig.29 , for the evaluation step (benchmark step) using the image quality evaluation value Ei ( Fig.28Step S73) will be described. In Embodiment 1, one of the features is that for the image quality evaluation value Ei of a certain same image quality, there is an evaluation step and a function for judging the superiority or inferiority of the image quality of the first image and the second image based on the comparison result between the image quality evaluation value Ei_1 of the first image and the image quality evaluation value Ei_2 of the second image.

[0338] As a problem, in the past, when judging the superiority or inferiority of the image quality of two images, people visually confirmed / judged the image quality, so the judgment criteria were vague. In contrast, in Embodiment 1, by using the image quality evaluation value Ei, the comparison of the image quality can be quantitatively performed.

[0339] In Fig.28 , in the evaluation step S73, the computer system 3 compares the specified first image and the second image with the image quality evaluation value Ei of the specified image quality according to the image quality evaluation result 204 including the image quality evaluation value Ei, and outputs the comparison result to the user.

[0340] Fig.29 An example of the GUI screen related to the evaluation step S73 is shown. In Fig.29 's screen, an example of outputting two images and the comparison result, etc. is shown. In Fig.29 's screen, there are GUIs such as evaluation (in other words, comparison result) 2900, recommended image 2910, observed image selection 2920, etc.

[0341] The evaluation 2900 is in tabular form, and for the first image 2901 and the second image 2902 of the two images to be compared in terms of image quality, the respective image quality evaluation values Ei (E1 to EN) and the comprehensive evaluation value C are displayed side by side. The user can compare the two images, the respective image quality evaluation values Ei, and the comprehensive evaluation value C in the evaluation 2900.

[0342] For example, regarding the image quality evaluation value E1 "sharpness" ( Figure 3 ), the E1 value of the second image 2902 is larger than the E1 value of the first image 2901. For example, it is defined that the larger the E1 value, the better the image quality. Therefore, regarding the image quality "sharpness" evaluated by the image quality evaluation value E1, the user can not only make a visual judgment of the two images, but also judge that the second image 2902 is superior to the first image 2901. The same applies to other image quality evaluation values Ei.

[0343] In addition, the comprehensive evaluation value C is compared in the same way between the first image 2901 and the second image 2902. Thereby, it can be judged which image is superior in terms of the overall image quality. For example, it is defined that the larger the value of the comprehensive evaluation value C, the superior the overall image quality. In this example, the comprehensive evaluation value C of the second image 2902 is high, so the user can judge that the image quality of the second image 2902 is superior overall.

[0344] In addition, the computer system 3 may also have the following function: using the comparison result 2900 in the above evaluation step S73, it determines which of the two images is preferably used as the observation image and automatically recommends it to the user. For example, the computer system 3 comprehensively determines, based on the comprehensive evaluation value C in particular in the comparison result 2900, that the second image 2902 is suitable as the observation image. Then, in the area of the recommended image 2910, the computer system 3 displays whether the first image 2901 or the second image 2902 is recommended or not as the observation image. At this time, for example, when the comprehensive evaluation value C of the image is above the threshold, the computer system 3 may determine the image as "recommended", or it may determine the image with the highest comprehensive evaluation value C among multiple (here, two) images as "recommended".

[0345] When using the above recommendation function, the user confirms the advantages and disadvantages of the recommended image (for example, the second image 2902) in the recommended image 2910 on the GUI screen and, as a judgment result, inputs the image to be used as the observation image. For example, in the area of the observation image selection 2920, the user can select and input which of the two images is to be used as the observation image. The computer system 3 saves the image selected by the user through the GUI screen as the user setting information 202. Based on this user setting information 202, the designated image is used as the observation image (the image used in the processing operation of the observation in step S5 of Figure 2 ).

[0346] In addition, the computer system 3 inputs the designated user setting information 202 including the observation image into Fig.28 step S63 in the setting in step S6. In step S63, the computer system 3 automatically determines the calculation method of the image quality evaluation value Ei in step S31 according to this user setting information 202. At this time, for the image with better image quality used as the observation image, the computer system 3 determines the calculation method of the image quality evaluation value Ei in such a way that the image quality evaluation value Ei becomes larger.

[0347] In addition, the computer system 3 inputs the designated user setting information 202 including the above observation image into step S64 in step S6. In step S64, the computer system 3 automatically determines the calculation method of the comprehensive evaluation value C in step S32 according to this user setting information 202. For the image with better image quality set as the observation image, the computer system 3 determines the calculation method of the comprehensive evaluation value C in such a way that the comprehensive evaluation value C becomes larger.

[0348] [Defect Observation Using Image Quality Evaluation Value Ei]

[0349] In the described Figure 2In the example, the case where the image quality evaluation value Ei is used for defect observation of the image 201 in step S5 is shown. In step S5, for example, when the image quality evaluation value Ei of the image 201 is high enough, for example, above a threshold value, the computer system 3 uses this image 201 as an observation image (in other words, an inspection image) and performs defect observation (in other words, defect inspection, defect detection). For example, an image 201 with a sufficiently high "visual recognition of defects" in the image quality evaluation value E2 is selected as the observation image. In this observation image, defects are determined / detected in a specified defect observation method (for example, a method of comparing with a reference image).

[0350] Use Fig.30 , a detailed processing example of defect observation (step S5) using the image quality evaluation value Ei will be described. In Embodiment 1, one of the features is that there is a defect detection step of detecting defects in an image obtained by photographing a semiconductor wafer based on the above image quality evaluation value Ei.

[0351] As a problem, when the image quality of the image to be defect-detected changes, over-detection (in other words, false alarms) and missed detections of defects occur. In contrast, in Embodiment 1, defect detection processing based on the image quality evaluation value Ei is performed.

[0352] Fig.30 Step S3000 representing defect detection using the image quality evaluation value Ei is shown. In Fig.30 , an example of adjusting the sensitivity of defect detection according to the image quality evaluation value Ei is shown. The computer system 3 uses the image quality evaluation value Ei (including the image quality evaluation result 204 thereof) obtained in the above steps S2 and S3 to perform the defect detection processing of step S3000 and obtains the defect detection result 3010 as a result thereof.

[0353] An example of a general defect detection method is as follows: An image obtained by photographing the defect generation site, that is, an inspection image (in other words, an object image, etc.), is compared with an image of a defect-free site, that is, a reference image, and their difference is investigated. Thus, the defect part is determined / detected. In Fig.30 the defect detection processing of step S3000, such a method is used.

[0354] First, the computer system 3 inputs an inspection image 3001 obtained by photographing the defect generation site and a reference image 3002 of a defect-free site. In step S3001, by performing a difference calculation on these images, a difference detection result 3004 is obtained and output.

[0355] In the difference detection result 3004, the difference generated at the defect site is detected, but at the same time, the difference generated due to the influence of noise, etc. is also detected. Therefore, in Embodiment 1 ( Fig.30 ) In it, for the inspection image 3001, in step S3002 corresponding to steps S2 and S3 for calculating the image quality evaluation value, the computer system 3 calculates the image quality evaluation value Ei3006. Figure 2

[0356] Next, based on the image quality evaluation value Ei3006, the computer system 3 adjusts the sensitivity of defect detection in step S3002. This sensitivity is the sensitivity when judging / detecting the defective part in the defect detection method used for defect detection processing. This sensitivity is not particularly limited, and for example, parameter values such as the weight for determining the defective part based on the difference or the threshold for removing false alarms can be cited.

[0357] In this example, the adjusted sensitivity of defect detection in step S3002 is used for false alarm removal in step S3004. In step S3004, the computer system 3 removes false alarms from the difference detection result 3004 before false alarm removal according to the sensitivity, and outputs this result as the difference detection result 3009 after false alarm removal. The defect detection result 3010 is generated / output based on the difference detection result 3009. In this example, false alarms are removed from the areas presumed to be defective shown as white parts included in the difference detection result 3004. As a result, in the difference detection result 3004, the defective area a1 shown as the white part is detected.

[0358] In this embodiment, the case of adjusting the sensitivity of defect detection according to the image quality evaluation value Ei has been described, but instead, or on this basis, changes in the feature quantity, preprocessing, and postprocessing in defect detection according to the image quality evaluation value Ei can also be performed.

[0359] In this embodiment, the method of using a reference image in defect detection has been described, but it is not limited to this. A method of generating a reference image based on the inspection image 3001 by presumption and comparing it with the presumed reference image can also be applied. Or, a method of inputting only the inspection image 3001 to a machine learning model and outputting the defect detection result as the presumption result of the model can also be applied.

[0360] [Shape measurement using the image quality evaluation value Ei]

[0361] Use Fig.31 , the shape measurement using the image quality evaluation value Ei is described. In Embodiment 1, one of the features is that there is a step of shape measurement for measuring the shape (in other words, the length measurement value, CD: Critical Dimension) in the image obtained by photographing a semiconductor wafer according to the image quality evaluation value Ei.

[0362] As a problem, when the image quality of the image to be measured as the shape measurement object changes, the signal used in the measurement changes and the length measurement value changes. In contrast, in the first embodiment, shape measurement is performed based on the image quality evaluation value Ei.

[0363] Fig.31 Step S3100 representing shape measurement using the image quality evaluation value Ei is shown. In this embodiment, Fig.31 Step S3100 of the shape measurement represents an example of adjusting the length measurement threshold for the shape measurement of the image based on the image quality evaluation value Ei obtained from the results of Figure 2 Steps S2 and S3. The computer system 3 inputs the measurement object image 3101 and analyzes the signals of the measurement points in the image 3101, thereby performing shape measurement. In this example, the case of measuring the wiring width between A and B in the image 3101 is described.

[0364] The computer system 3 first analyzes the signals (in other words, the distribution) between A and B in the image 3101 in the signal analysis step S3101, and generates / outputs the signal analysis result 3103.

[0365] Next, the computer system 3 calculates the image quality evaluation value Ei3105 for the image 3101 in step S3102 corresponding to the Figure 2 image quality evaluation value calculation steps S2 and S3. Next, the computer system 3 adjusts / decides the length measurement threshold in step S3103 based on this image quality evaluation value Ei3105. Then, in step S3104, the computer system 3 performs shape measurement on the signal analysis result 3103 using the determined length measurement threshold, and as a result, generates / outputs the shape measurement result 3110 including the length measurement value 3108.

[0366] In this embodiment, an example of adjusting the length measurement threshold for shape measurement based on the image quality evaluation value Ei has been described, but it is not limited thereto, and changes in feature quantities, pre-processing, and post-processing in shape measurement based on the image quality evaluation value Ei can also be performed.

[0367] [Adjustment of shooting conditions using the image quality evaluation value Ei]

[0368] Using Fig.32 , the adjustment of the shooting conditions using the image quality evaluation value Ei is described. In the first embodiment, one of the features is that there is a step of determining the shooting conditions ( Figure 6 shooting condition 205 in) of the image 201 based on the image quality evaluation value Ei. The computer system 3 adjusts / decides the appropriate shooting condition 205 using the image quality evaluation value Ei. In addition, the computer system 3 can also use the target value of the image quality evaluation value Ei ( Fig. 22to determine the shooting condition 205 based on the regional target value Ti) in

[0369] As a problem, when adjusting the shooting conditions of an image, it is necessary for a person to visually evaluate the image quality, so the adjustment takes time and effort. In contrast, in the first embodiment, by quantifying the image quality, visual evaluation of the image quality is not required, and the time and effort related to the adjustment of the shooting conditions are reduced.

[0370] Fig.32 Indicates the adjustment of the shooting condition 205 using the image quality evaluation value Ei. Fig.32 Based on Figure 2 It has parts that are partly different. First, for the image 201 captured in step S1 according to a certain shooting condition 205, the computer system 3 uses the above-described steps S2 and S3 to calculate the image quality evaluation value Ei and obtains the image quality evaluation result 204. The computer system 3 can also display the image quality evaluation value Ei and the shooting condition 205 to the user in step S4.

[0371] Next, in step S3200, the computer system 3 adjusts / determines (in other words, updates) the shooting condition 205 of the image 201 according to the image quality evaluation value Ei of the image 201. As a result, the updated shooting condition 205B is obtained. This shooting condition 205B becomes the new shooting condition 205.

[0372] In step S3200, for example, when the value of the image quality evaluation value E5 "dark part emphasis degree" ( Figure 3 ) is small, in order to improve the visual recognition of the dark part of the image 201, the acceleration voltage, which is one of the parameters of the shooting condition 205, is changed to a larger value.

[0373] The computer system 3 uses the updated shooting condition 205B and repeats Fig.32 A series of processes of. Thus, a more appropriate shooting condition 205 for being able to capture an image 201 with better image quality is determined. In other words, the computer system 3 reflects the image quality evaluation value Ei and optimizes the shooting condition 205.

[0374] In Fig.32 When adjusting / determining the shooting condition 205 of, similar to the adjustment of the parameter P of the above-described image quality improvement engine ( Fig.21 ), a regional target value Ti can be set for a part or all of the plurality of image quality evaluation values Ei, and the shooting condition 205B can be determined in such a way that the image quality evaluation value Ei approaches the regional target value Ti.

[0375] [Monitoring of the specimen observation device using the image quality evaluation value Ei]

[0376] Use Fig.33The monitoring of the sample observation device using the image quality evaluation value Ei will be described. In Embodiment 1, one of the features is that it has a step of monitoring the device monitoring of the sample observation device based on the image quality evaluation value Ei. The sample observation device to be monitored here refers to Figure 1 the sample observation device 1 of Embodiment 1 in

[0377] As a problem, when using a captured image to determine whether an abnormality has occurred in the sample observation device, etc., it is necessary for a person to visually confirm / judge the image, thus incurring operation costs, etc. In contrast, in Embodiment 1, by using the image quality evaluation value Ei, it is quantitatively determined whether an abnormality has occurred in the sample observation device 1, etc. Thus, visual confirmation by a person is not required, and operation costs, etc. can be reduced.

[0378] Fig.33 Step S3300 showing the monitoring of the sample observation device using the image quality evaluation value Ei is presented. First, in step S3301, the computer system 3 inputs Figure 1 the image 201 captured by the sample observation device 1 of Figure 2 by the SEM2. For this image 201, the image quality evaluation value Ei3302 is calculated in step S3301 corresponding to steps S2 and S3 of

[0379] Then, in step S3302, the computer system 3 uses the image quality evaluation value Ei3302 and the normal range information 3303 of the image quality evaluation value Ei as inputs to determine whether an abnormality has occurred in the sample observation device 1, etc., in other words, to judge the device state. For example, the computer system 3 may determine the device state as normal (in other words, no abnormality) when the image quality evaluation value Ei is within the normal range in the normal range information 3303, and determine the device state as abnormal (in other words, there is an abnormality) when it is outside the normal range.

[0380] If the result of step S3302 determines that the device has an abnormality (yes), it proceeds to step S3303. If it is determined that there is no abnormality (no), it proceeds to step S3304. In step S3303, the computer system 3 analyzes the cause of the abnormality related to the above-mentioned abnormality, and as a result, outputs the cause of the abnormality 3306 (in other words, the result of the abnormality analysis).

[0381] For example, as a result of investigating the image quality evaluation value E1 "sharpness" of the image 201 ( Figure 3 ), if it is determined that there is an abnormality, in step S3303, the computer system 3 determines and outputs that there is an abnormality in the electrical axis of the SEM2 ( Figure 1 ) associated with the sharpness of the image 201 (in other words, the cause of the abnormality). The electrical axis refers to the orbit of the electron beam emitted from the electron gun 110.

[0382] In step S3305, the computer system 3 displays the device status (with abnormality) and the cause of the abnormality 3306 based on the cause of the abnormality 3306, for example, through the GUI screen, and the user can confirm them. Based on the cause of the abnormality 3306, the user can, as a countermeasure, perform adjustments, maintenance / replacement, etc. on the positions of the components in the specimen observation device 1 associated with the cause of the abnormality 3306.

[0383] On the other hand, in step S3304, the computer system 3 outputs the normal information 3307 indicating no abnormality as the device status.

[0384] [Effects of Embodiment 1, etc.]

[0385] As described above, according to the specimen observation device and method of Embodiment 1, it is possible to realize the quantification of the observed image quality. According to Embodiment 1, it is possible to use the evaluation partial area Si suitable for the image quality evaluation value Ei corresponding to various image qualities, and it is possible to realize appropriate image quality quantification. According to Embodiment 1, it is possible to use the quantified image quality evaluation value Ei to realize the effects based on various functions (such as the adjustment of parameters for the above-mentioned image enhancement processing, etc.).

[0386] The embodiments of the present disclosure have been specifically described above, but are not limited to the described embodiments, and various changes can be made without departing from the spirit. Except for the essential elements, the components can be added, deleted, replaced, etc. Without particular limitation, each component can be single or multiple. It can also be a combination of the embodiments.

[0387] Symbol Explanation

[0388] 1... Specimen observation device, 2... SEM (scanning electron microscope), 3... Computer system, 10... Specimen.

Claims

1. A computer system in a specimen observation apparatus for observing a specimen, characterized in that the computer system has more than one processor and more than one memory, and the processor performs the following operations: Determine a plurality of evaluation partial regions for an image obtained by photographing the specimen; Based on the plurality of evaluation partial regions, calculate image quality evaluation values for each evaluation partial region, thereby calculating a plurality of image quality evaluation values for evaluating a plurality of different image qualities.

2. The computer system according to claim 1, characterized in that the plurality of image quality evaluation values include at least two or more of the following: sharpness of the image, visual recognition of defects, degree of noise suppression, interlayer contrast, degree of dark part emphasis, shape preservation, roughness preservation, ringing suppression degree, naturalness of the image.

3. The computer system according to claim 1, characterized in that the processor performs the following operations: Classify according to classification segmentation regions obtained by dividing the image; Based on the classification results of the classification segmentation regions, determine the evaluation partial regions.

4. The computer system according to claim 3, characterized in that the processor uses specimen information to perform the classification, where the specimen information represents the structure of a semiconductor device as the specimen.

5. The computer system according to claim 3, characterized in that the processor uses the shooting conditions of the image to perform the classification.

6. The computer system according to claim 3, characterized in that the processor generates the evaluation partial regions by combining, dividing, dilating, contracting, or a combination of these of the classification segmentation regions.

7. The computer system according to claim 3, characterized in that the classification of the classification segmentation regions includes at least two or more of the following: the upper surface region of the wiring, the edge region, the dark part region, and the defect region in a semiconductor device as the specimen.

8. The computer system according to claim 1, characterized in that the processor calculates a comprehensive evaluation value formed by combining the plurality of image quality evaluation values.

9. The computer system according to claim 1, characterized in that According to the setting, use internal parameters to calculate the image quality evaluation value, and the processor determines a calculation method of the image quality evaluation value using the internal parameters according to the input of user evaluation related to the image quality of the image, so that the image quality evaluation value is consistent with or close to the user evaluation.

10. The computer system according to claim 1, characterized in that According to the setting, define the image quality evaluation value by a combination of a plurality of elemental image quality evaluation values.

11. The computer system according to claim 1, characterized in that the processor determines a parameter value for image quality improvement processing for the image according to the image quality evaluation value.

12. The computer system according to claim 11, characterized in that the processor performs the following operations: Individually set target values for one or more of the plurality of image quality evaluation values; Adjust the parameter value of the image quality improvement process so that the image quality evaluation value approaches the target value.

13. The computer system according to claim 11, wherein the processor performs the following operations: Calculate a comprehensive evaluation value formed by combining the multiple image quality evaluation values; Adjust the parameter value of the image quality improvement process to improve the comprehensive evaluation value.

14. The computer system according to claim 1, wherein the processor performs the following operations: Display the image, the evaluation partial area, and the image quality evaluation value on the screen; Correct the evaluation partial area according to the user input to the screen.

15. The computer system according to claim 1, wherein the processor performs the following operations: Perform an evaluation of comparing the image qualities of a first image and a second image as the image; In the evaluation, compare the image quality evaluation value of the first image and the image quality evaluation value of the second image, and display the comparison result on the screen.

16. The computer system according to claim 1, wherein the processor performs the following operations: For the image obtained by photographing a semiconductor wafer as the specimen, calculate the image quality evaluation value; Select an image for defect observation according to the image quality evaluation value; Perform the defect observation on the selected image.

17. The computer system according to claim 1, wherein the processor performs the following operations: For the image obtained by photographing a semiconductor wafer as the specimen, calculate the image quality evaluation value; Adjust the shooting conditions when shooting the image according to the image quality evaluation value; Shoot the image under the adjusted shooting conditions.

18. The computer system according to claim 1, wherein As monitoring the state of the specimen observation device, the processor determines whether an abnormality occurs in the specimen observation device according to the image quality evaluation value.

19. A method corresponding to claim 1, which is a specimen observation method in a specimen observation device for observing a specimen, characterized in that the computer system in the specimen observation device has one or more processors and one or more memories, As steps executed by the processor, there are the following steps: A step of determining a plurality of evaluation partial areas for an image obtained by photographing the specimen; and A step of calculating an image quality evaluation value for each evaluation partial area according to the plurality of evaluation partial areas, thereby calculating a plurality of image quality evaluation values for evaluating different multiple image qualities.

20. A program corresponding to claim 1, characterized in that It is used to make a computer system execute the specimen observation method described in claim 19.

Images