Recommendation system and method for autonomous mode selection in inspection tools and other tools

Abstract

Methods and systems are provided for selecting modes for a mode selection process. A system includes a computer subsystem configured for determining information of a sample and at least one value of a characteristic of the information from an image generated for the sample with an initial subset of different modes of an imaging subsystem. The computer subsystem is also configured to predict a probability of determining a better value of the characteristic from the image generated with the different modes other than the initial subset based on the determined at least one value of the characteristic and a relationship between the different modes and associated values of the characteristic of the information. In addition, the computer subsystem is configured to select an additional subset of the different modes from which the generating and determining steps are then performed by the imaging and computer subsystems, respectively, based on the predicted probabilities.

Description

Technical Field

[0001] This invention generally relates to methods and systems for selecting patterns in a pattern selection process. Some embodiments relate to recommendation systems and methods for autonomous pattern selection in inspection tools and other tools. Background Technology

[0002] The following descriptions and examples are not acknowledged as prior art simply because they are included in this section.

[0003] Fabricating semiconductor devices (such as logic and memory devices) typically involves processing a substrate (such as a semiconductor wafer) using numerous semiconductor fabrication processes to form the various features and multiple layers of the semiconductor device. For example, photolithography is a semiconductor fabrication process involving transferring a pattern from a photomask to a photoresist disposed on a semiconductor wafer. Additional examples of semiconductor fabrication processes include (but are not limited to) chemical mechanical polishing (CMP), etching, deposition, and ion implantation. Multiple semiconductor devices can be fabricated on a single semiconductor wafer in a particular arrangement and then separated into individual semiconductor devices.

[0004] Inspection processes are used at various stages during semiconductor manufacturing to detect defects on samples, leading to higher manufacturing yields and thus higher profits. Inspection has always been a crucial part of semiconductor device fabrication. However, as semiconductor device sizes shrink, inspection becomes even more critical for the successful manufacture of acceptable semiconductor devices, as even small defects can cause device failure.

[0005] Defect re-inspection typically involves re-examining defects detected by the inspection process using high-magnification optical systems or scanning electron microscopy (SEM) to generate additional information about the defects at higher resolution. Therefore, defect re-inspection is performed at discrete locations on the sample where defects were previously detected by inspection. The higher-resolution data generated by defect re-inspection is better suited for determining defect attributes such as profile, roughness, and more accurate size information. Compared to inspection, the information determined by defect re-inspection allows for a more accurate classification of defects into several defect types.

[0006] Metrology processes are also used to monitor and control the process at various steps during semiconductor manufacturing. Metrology differs from inspection processes in that, unlike inspection processes which detect defects on samples, metrology processes measure one or more characteristics of a sample that cannot be determined using currently available inspection tools. For example, metrology processes measure one or more characteristics of a sample during the process (e.g., the dimensions of features formed on the sample, such as linewidth, thickness, etc.) so that the performance of the process can be determined from one or more characteristics. Furthermore, if one or more characteristics of a sample are unacceptable (e.g., outside a predetermined range), the measurement of those characteristics can be used to modify one or more parameters of the process so that additional samples manufactured by the process have acceptable characteristics.

[0007] The difference between metrological processes and defect re-inspection processes lies in the following: unlike defect re-inspection, which re-inspects defects detected by inspection, metrological processes can be performed at locations where defects were not detected. In other words, unlike defect re-inspection, the location for performing metrological processes on a sample can be independent of the results of the inspection process performed on the sample. Specifically, the location for performing metrological processes can be selected independently of the inspection results. Furthermore, because the location for performing metrology on a sample can be selected independently of the inspection results, unlike defect re-inspection (where the location for performing defect re-inspection on a sample cannot be determined until the inspection results of the sample are generated and available for use), the location for performing metrological processes can be determined before performing the inspection process on the sample.

[0008] Many different methods have been tried and developed to set the output production (e.g., imaging) parameters of a process (such as the process described above). The most widely used mode selection method on optical inspection tools involves manually running modes and analyzing the resulting data (e.g., signal-to-noise ratio (SNR) data) in the user interface. The user, physically using the tool, wants to run a set of initial modes (based on experience or from 10,000 or more unexplored possible modes, selecting the best known mode (BKM)). The tool captures images and calculates various metrics for each defect in each mode. The user analyzes this data and selects the next set of potentially good modes. This iterative process continues until the user is satisfied with the "top modes" seen so far, or all modes have been used, or even the user has run out of tool time.

[0009] Another current method for pattern selection is the brute-force approach. In this case, the user can queue a large number of optical patterns running at night on the tool. The user can then analyze the data later after all the optical patterns have run. Based on the data analysis, the user can still perform one or two iterations as needed for further analysis.

[0010] Therefore, current pattern selection methods have many significant drawbacks. For example, existing methods are labor-intensive, require extensive expertise, and produce inconsistent results. Becoming an expert requires qualified training and certification, and several years of experience gained through various in-house and field operations at different locations. At any given time, there may only be a few such experts, which is why manual optical pattern selection is error-prone, costly, and time-consuming. Even without significant errors, the chance of producing suboptimal formulations is quite high.

[0011] Therefore, it would be advantageous to develop systems and methods for selecting patterns in the pattern selection process that do not have one or more of the aforementioned drawbacks. Summary of the Invention

[0012] The following description of various embodiments should in no way be construed as limiting the subject matter of the appended claims.

[0013] One embodiment relates to a system configured to select a mode for a mode selection process. The system includes an imaging subsystem configured to generate images of a sample using different modes of the imaging subsystem. The system also includes a computer subsystem configured to determine information about the sample and at least one value of a characteristic of the information from the images generated using an initial subset of the different modes. The computer subsystem is further configured to predict a probability of determining a better value of the characteristic from images generated using the different modes other than the initial subset, based on the determined at least one value of the characteristic and the relationship between the different modes and the associated values ​​of the characteristic of the information. Additionally, the computer subsystem is configured to select an additional subset of the different modes, based on the predicted probability, for which the imaging and computer subsystems then respectively perform the generation and determination steps. The system may be further configured as described herein.

[0014] Another embodiment relates to a computer-implemented method for selecting a pattern for a pattern selection process. The method includes the above-described generation, determination, prediction, and selection steps. Each of the steps in the above method may be performed as further described herein. Embodiments of the above method may include any other steps of any other method described herein. The above method may be performed by any of the systems described herein.

[0015] Another embodiment relates to a non-transitory computer-readable medium storing program instructions executable on a computer system to perform a computer-implemented method for selecting a mode for a mode selection process. The computer-implemented method includes the steps of the methods described above. The computer-readable medium may be further configured as described herein. The steps of the computer-implemented method may be performed as further described herein. Additionally, the computer-implemented method to which the program instructions are executable may include any other steps of any other method described herein. Attached Figure Description

[0016] Those skilled in the art will appreciate further advantages of the invention after receiving the following detailed description of preferred embodiments and referring to the accompanying drawings, in which:

[0017] Figure 1 and 2 This is a schematic diagram of a side view illustrating an embodiment of a system configured as described herein;

[0018] Figure 3 This is a schematic diagram of an embodiment illustrating the relationship between the characteristics of different modes of a visualization imaging subsystem that can be generated by the embodiments described herein and the properties of information determined for a sample from images generated using different modes.

[0019] Figure 4 This is a flowchart illustrating an embodiment of steps that can be executed to select a mode for a mode selection process; and

[0020] Figure 5 This is a block diagram illustrating an embodiment of a non-transitory computer-readable medium storing program instructions for causing a computer system to perform the computer-implemented methods described herein.

[0021] While various modifications and alternatives to the invention are possible, specific embodiments thereof are shown by way of example in the drawings and described in detail herein. The drawings may not be drawn to scale. However, it should be understood that the drawings and their detailed description are not intended to limit the invention to the specific forms disclosed, but rather, the invention is intended to cover all modifications, equivalents, and alternatives falling within the spirit and scope of the invention as defined by the appended claims. Detailed Implementation

[0022] Turning to the figures, it should be noted that the figures are not drawn to scale. Specifically, some components in the figures are enlarged significantly to emphasize their characteristics. It should also be noted that the figures are not drawn to the same scale. Components shown in more than one figure that can be similarly configured have been indicated using the same reference numerals. Unless otherwise stated herein, any of the components described and shown may include any commercially available components.

[0023] Generally, the embodiments described herein are configured to select a mode for a mode selection process. Some embodiments are configured as a recommendation system for autonomous mode selection of imaging tools, such as inspectors and other tools described herein. Although some embodiments may be described herein with respect to inspectors, embodiments may be configured for any of the other imaging subsystems described herein.

[0024] The "recipe authorization" on an optical inspector means that the recipe has been set to provide optimal performance for a specific sample. Selecting the "correct" optical mode is arguably the most critical part of achieving this goal. The process of selecting the optical mode on an optical inspector tool has traditionally been entirely manual. The user must find the "correct" optical mode from many (e.g., 10,000 or more) possible choices. Therefore, the success of this process depends on the technical skill and experience of the human expert working on the sample and the tool.

[0025] User-based pattern selection is not the desired method for achieving authorization, not only because it is extremely expensive (in terms of workload and manpower), but also because the results are subjective and inconsistent. However, despite these obvious drawbacks of current pattern selection methods, attempts to automate this process, aside from basic brainstorming activities, have not been successful. A major obstacle to automating pattern selection is that algorithms are not considered feasible solutions to the problem. Instead, the focus tends to be on optical simulations that can be used to suggest the "correct" optical pattern.

[0026] The embodiments described herein provide a “recommendation system” that offers a deterministic way of selecting the “correct” optical pattern for a given sample without any human intervention. The idea of ​​using a recommendation system for pattern selection in inspectors (and other tools described herein) is considered original.

[0027] In some embodiments, the sample is a wafer. The wafer may comprise any wafer known in the semiconductor field. Although some embodiments may be described herein with respect to one or more wafers, the embodiments are not limited to the samples that can be used. For example, the embodiments described herein can be used for samples such as photomasks, tablets, personal computer (PC) boards, and other semiconductor samples.

[0028] Figure 1 This document illustrates an embodiment of a system configured to select a mode for a mode selection process. System 10 includes an imaging subsystem 100 configured to generate images of samples using different modes of the imaging subsystem. The imaging subsystem may be configured as one of the types of imaging subsystems described herein, such as inspection, metrology, or defect re-examination subsystems, and may include and / or be coupled to computer subsystem 36 and / or one or more computer systems 102.

[0029] The terms "imaging system" and "imaging subsystem" are used interchangeably herein and generally refer to hardware configured to generate images of a sample. Generally, the imaging subsystem described herein includes at least a power source and a detector. The power source is configured to generate energy directed to the sample. The detector is configured to detect energy from the sample and generate an output in response to the detected energy.

[0030] In a light-based imaging subsystem, the energy directed to the sample contains light, and the energy detected from the sample also contains light. For example, as... Figure 1 As shown, the imaging subsystem includes an illumination subsystem configured to guide light to sample 14. The illumination subsystem includes at least one light source, such as light source 16. The illumination subsystem is configured to guide light to the sample at one or more incident angles (which may include one or more tilt angles and / or one or more normal angles). For example, as... Figure 1 As shown, light from light source 16 is guided through optical element 18 and then through lens 20 to sample 14 at an angle of incidence. The angle of incidence can include any suitable angle of incidence, which can vary depending on, for example, the characteristics of the sample and the defects to be detected on the sample, the characteristics of the sample to be measured, etc.

[0031] The illumination subsystem can be configured to guide light to the sample at different incident angles. For example, the imaging subsystem can be configured to change one or more parameters of one or more elements of the illumination subsystem, so that the light can be directed to the sample at different incident angles. Figure 1 The different incident angles shown are guided to the sample. In one example, the imaging subsystem can be configured to move the light source 16, optical element 18, and lens 20 such that light is guided to the sample at different oblique incident angles or normal (or near-normal) incident angles. The illumination subsystem can have any other suitable configuration known in the art for guiding light to the sample sequentially or simultaneously at one or more incident angles.

[0032] The illumination subsystem can also be configured to direct light with different characteristics to the sample. For example, optical element 18 can be configured as a spectral filter and the properties of the spectral filter can be changed in various ways (e.g., by replacing one spectral filter with another) so that light of different wavelengths can be directed to the sample at different times.

[0033] Light source 16 may comprise a broadband plasma (BBP) source. In this manner, the light generated by the source and directed to the sample may comprise broadband light. However, the source may comprise any other suitable source, such as any suitable laser known in the art configured to produce light of any suitable wavelength. Additionally, the laser may be configured to produce monochromatic or near-monochromatic light. In this manner, the laser may be a narrowband laser. The source may also comprise a multicolor source that produces light of multiple discrete wavelengths or bands.

[0034] Light from optical element 18 can be focused onto sample 14 by lens 20. Although lens 20 is... Figure 1 While shown as a single refractive optical element, lens 20 may actually comprise several refractive and / or reflective optical elements that combine to focus light from the optical element onto the sample. Figure 1 The illumination subsystem shown and described herein may include any other suitable optical elements (not shown). Examples of such optical elements include (but are not limited to) polarization components, spectral filters, spatial filters, reflective optics, apodizers, beam splitters, apertures, and the like, which may include any such suitable optical elements known in the art. Additionally, the system may be configured to change one or more elements of the illumination subsystem based on the type of illumination used for imaging.

[0035] The imaging subsystem may also include a scanning subsystem configured to change the location on the sample where light is directed to and detected, and to cause the light to scan above the sample. For example, the imaging subsystem may include a stage 22 on which the sample 14 is placed during imaging. The scanning subsystem may include any suitable mechanical and / or robotic assembly (which includes the stage 22) configured to move the sample such that light can be directed to and detected from different locations on the sample. Alternatively or additionally, the imaging subsystem may be configured such that one or more optical elements of the imaging subsystem perform some scans of light above the sample, such that light can be directed to and detected from different locations on the sample. The light may scan above the sample in any suitable manner (e.g., in a serpentine path or a spiral path).

[0036] The imaging subsystem includes one or more detection channels. At least one of the detection channels includes a detector configured to detect light from the sample as the system illuminates the sample and to generate an output in response to the detected light. Figure 1 The imaging subsystem shown includes two detection channels: one formed by collector 24, element 26, and detector 28, and the other formed by collector 30, element 32, and detector 34. The two detection channels are configured to collect and detect light at different collection angles. In some examples, the two detection channels are configured to detect scattered light, and the detection channel is configured to detect light scattered from the sample at different angles. However, one or more of the detection channels may be configured to detect another type of light from the sample (e.g., reflected light).

[0037] exist Figure 1In this embodiment, the two detection channels are shown positioned within the plane of the paper, and the illumination subsystem is also shown positioned within the plane of the paper. Therefore, in this embodiment, the two detection channels are positioned (e.g., centered) within the incident plane. However, one or more of the detection channels may be positioned outside the incident plane. For example, the detection channel formed by collector 30, element 32, and detector 34 may be configured to collect and detect light scattered outside the incident plane. Therefore, this detection channel may generally be referred to as a "side" channel, and this side channel may be centered in a plane substantially perpendicular to the incident plane.

[0038] although Figure 1 An embodiment of an imaging subsystem comprising two detection channels is shown, but the imaging subsystem may comprise a different number of detection channels (e.g., only one detection channel or two or more detection channels). The detection channel formed by collector 30, element 32, and detector 34 may form a side channel as described above, and the imaging subsystem may comprise an additional detection channel (not shown) formed as another side channel positioned on the opposite side of the incident surface. Thus, the imaging subsystem may comprise a detection channel comprising collector 24, element 26, and detector 28 and centered in the incident surface and configured to collect and detect light at a scattering angle normal to or near normal to the sample surface. Therefore, this detection channel may generally be referred to as a “top” channel, and the imaging subsystem may also comprise two or more side channels configured as described above. Thus, the imaging subsystem may comprise at least three channels (i.e., one top channel and two side channels), and each of the at least three channels is configured to collect light at a scattering angle different from that of each of the other collectors.

[0039] As further described above, one or more of the detection channels can be configured to detect scattered light. Therefore, Figure 1 The imaging subsystem shown herein can be configured for dark-field (DF) imaging. However, the imaging subsystem may also, or alternatively, include a detection channel configured for bright-field (BF) imaging. Therefore, the imaging subsystem described herein can be configured for DF-only, BF-only, or both DF and BF imaging. Although each of the collectors... Figure 1 It is shown as a single refractive optical element, but each of the collectors may contain refractive optical elements and / or reflective optical elements.

[0040] One or more detection channels may contain any suitable detector known in the art, such as a photomultiplier tube (PMT), charge-coupled device (CCD), or time-delay integration (TDI) camera. The detector may also contain non-imaging detectors or imaging detectors. If the detector is a non-imaging detector, each of the detectors may be configured to detect certain properties of light (e.g., intensity), but may not be configured to detect such properties that vary depending on position within the imaging plane. Thus, the output generated by each detector in each detection channel may be a signal or data, but not an image signal or image data. In such examples, a computer system may be configured to generate an image of the sample from the non-imaging output of the detector. However, in other examples, the detector may be configured as an imaging detector, configured to generate an imaging signal or image data. Therefore, the imaging subsystem may be configured to generate images in several ways.

[0041] Computer subsystem 36 may be coupled to the detector of the imaging subsystem in any suitable manner (e.g., via one or more transmission media, which may include “wired” and / or “wireless” transmission media), such that the computer subsystem can receive the output generated by the detector. Computer subsystem 36 may be configured to use the detector output to perform several functions, as further described herein. Computer subsystem 36 may be further configured as described herein.

[0042] Computer subsystem 36 (and other computer subsystems described herein) may also be referred to herein as a computer system. Each of the computer subsystems or systems described herein may take various forms, including personal computer systems, graphics computers, mainframe computer systems, workstations, network devices, Internet devices, or other devices. Generally, the term "computer system" may be broadly defined to encompass any device having one or more processors that execute instructions from memory media. A computer subsystem or system may also include any suitable processor known in the art, such as a parallel processor. Additionally, a computer subsystem or system may include a computer platform with high-speed processing and software, as a standalone or networked tool.

[0043] If the system comprises more than one computer system, the different computer systems can be coupled to each other, allowing images, data, information, instructions, etc., to be transmitted between the computer systems. For example, computer subsystem 36 can be coupled to computer system 102 via any suitable transmission medium (which may include any suitable wired and / or wireless transmission medium known in the art), such as... Figure 1 The two or more components in such computer systems may also be effectively coupled by a shared computer-readable storage medium (not shown).

[0044] In the electron beam imaging subsystem, the energy guided to the sample contains electrons, and the energy detected from the sample also contains electrons. Figure 2 In one embodiment shown, the imaging subsystem includes an electron column 122, and the system includes a computer subsystem 124 coupled to the imaging subsystem. The computer subsystem 124 can be configured as described above. Additionally, this imaging subsystem can be configured as described above and... Figure 1 It is coupled to one or more computer systems in the same way as shown in the diagram.

[0045] For example Figure 2 As shown, the electron column includes an electron beam source 126 configured to generate electrons focused onto the sample 128 by one or more elements 130. The electron beam source may include, for example, a cathode source or an emitter tip, and the one or more elements 130 may include, for example, a gun lens, an anode, a beam-limiting aperture, a gate valve, a beam current selection aperture, an objective lens, and a scanning subsystem, all of which may include any such suitable elements known in the art.

[0046] Electrons returning from the sample (e.g., secondary electrons) can be focused onto detector 134 by one or more elements 132. One or more elements 132 may include, for example, a scanning subsystem, which may be the same scanning subsystem included in element 130.

[0047] The electron column may contain any other suitable element known in the art. Additionally, the electron column may be further configured as described below: U.S. Patent No. 8,664,594, issued April 4, 2014, by Jiang et al.; U.S. Patent No. 8,692,204, issued April 8, 2014, by Kojima et al.; U.S. Patent No. 8,698,093, issued April 15, 2014, by Gubbens et al.; and U.S. Patent No. 8,716,662, issued May 6, 2014, by MacDonald et al., which are incorporated herein by reference as if fully described herein.

[0048] Despite the electron column Figure 2 The diagram illustrates a configuration where electrons are guided to the sample at an angle of incidence and scattered from the sample at another angle. However, the electron beam can be guided to and scattered from the sample at any suitable angle. Furthermore, the electron beam imaging subsystem can be configured to use multiple modes to produce the output of the sample further described herein (e.g., with different illumination angles, collection angles, etc.). These multiple modes of the electron beam imaging subsystem can differ in any output production parameters of the imaging subsystem.

[0049] Computer subsystem 124 may be coupled to detector 134, as described above. The detector detects electrons returning from the surface of the sample, thereby forming an electron beam image of the sample (or other output of the sample). The electron beam image may contain anything suitable for an electron beam image. Computer subsystem 124 may be configured to perform any of the steps described herein. Figure 2 The imaging subsystem shown in the document can be further configured as described herein.

[0050] This article provides Figure 1 and 2 This document broadly describes the configuration of imaging subsystems that may be included in the system embodiments described herein. Obviously, the imaging subsystem configurations described herein can be modified to optimize the performance of the imaging subsystem, as typically exhibited when designing commercial imaging systems. Additionally, the systems described herein can be implemented using existing imaging subsystems (e.g., by adding the functionality described herein to an existing inspection system), such as tools purchased from KLA Corporation in Milpitas, California. For some such systems, the methods described herein may be provided as optional features of the imaging subsystem (e.g., and other functions of the imaging system). Alternatively, the imaging systems described herein may be designed “from scratch” to provide entirely new systems.

[0051] Although the imaging subsystem is described above as a light or electron beam imaging subsystem, it can also be an ion beam imaging subsystem. This imaging subsystem can be as follows: Figure 2 The configuration shown is simply that the electron beam source can be replaced with any suitable ion beam source known in the field. Additionally, the imaging subsystem may include any other suitable ion beam system, such as those found in commercially available focused ion beam (FIB) systems, helium ion microscopy (HIM) systems, and secondary ion mass spectrometry (SIMS) systems.

[0052] The imaging subsystem is configured to produce outputs (e.g., images) of the sample using multiple modes. Generally, a "mode" is defined by the values ​​of parameters of the imaging subsystem used to produce the image of the sample (or the output of the image of the sample). Therefore, a mode can differ in the value of at least one of the parameters of the imaging subsystem (rather than the location on the sample where the output is produced). For example, a mode can differ in any or more modifiable parameters of the imaging subsystem (e.g., illumination polarization, angle, wavelength, etc.; detection polarization, angle, wavelength, etc.). The imaging subsystem can be configured to scan the sample using different modes in the same or different scans, depending, for example, on the ability to scan the sample simultaneously using multiple modes.

[0053] In a similar manner, the electron beam subsystem is configured to produce images using two or more modes (which can be defined by the values ​​of the parameters of the electron beam subsystem used to generate the image of the sample). Therefore, the modes can differ in the value of at least one of the electron beam parameters of the electron beam subsystem. For example, different modes can use different illumination incident angles.

[0054] The imaging subsystem described herein can be configured as an inspection system, a metrology system, and / or a defect re-inspection system. For example, Figure 1 and 2 The embodiments of the imaging subsystem shown can be modified with one or more parameters to provide different imaging capabilities depending on the application in which the embodiments will be used. In one such example, the imaging subsystem can be configured to have a higher resolution when used for metrology than when used for inspection. In other words, Figure 1 and 2 The embodiments of the imaging subsystems shown herein are described in several ways that will be understood by those skilled in the art to produce some general and various configurations of the imaging subsystems with different imaging capabilities suitable for virtually any application.

[0055] In this manner, the imaging subsystem can be configured to generate output suitable for detecting or re-inspecting defects on a sample in either an inspection system or a defect re-inspection system, and suitable for measuring one or more characteristics of a sample in a metrology system. In an inspection system, Figure 1 The computer subsystem 36 shown can be configured to detect defects on sample 14 by applying a defect detection method or algorithm to the output generated by one or more of the detectors. In the defect re-inspection system, Figure 2 The computer subsystem 124 shown can be configured to re-detect defects on sample 128 by applying a defect re-detection method to the output generated by detector 134, and may use the output generated by the detector to determine additional information about the re-detected defects. In a metrology system, Figure 1 The computer subsystem 36 shown herein can be configured to determine one or more characteristics of sample 14 using the output generated by detectors 28 and / or 34. The system can be further configured to detect or re-detect defects on the sample, determine characteristics of the sample, determine other information about the sample, etc., as further described herein.

[0056] As mentioned above, the imaging subsystem is configured to scan energy (e.g., light, electrons, etc.) over a physical version of the sample, thereby producing an output of the physical version of the sample. In this way, the imaging subsystem can be configured as a "real" subsystem rather than a "virtual" subsystem. However, storage media (not shown) and Figure 1The computer system 102 shown herein can be configured as a “virtual” system. Specifically, the storage media and computer system can be configured as a “virtual” imaging system, as described in U.S. Patent No. 8,126,255, issued February 28, 2012, by Bhaskar et al., jointly assigned, and U.S. Patent No. 9,222,895, issued December 29, 2015, by Duffy et al., which are incorporated herein by reference as if their entirety were described. The embodiments described herein can be further configured as described in those patents.

[0057] The system includes a computer subsystem, which may contain any configuration of the aforementioned computer subsystem or any of the systems described herein. The computer subsystem is configured to determine at least one value of a characteristic of a sample and a feature of that information from images generated from an initial subset of images produced using different modes. In one embodiment configured for inspection, the computer subsystem may be configured to determine information about the sample by detecting defects on the sample using images generated by the imaging subsystem, wherein the characteristic of the information may be the signal-to-noise ratio (SNR) of the detected defects, and the value of the characteristic may be a determined SNR value. Although some embodiments may be described herein with respect to SNR values, the embodiments described herein may be configured to use any other value that can be determined from images generated by any of the imaging subsystems described herein for any of the processes described herein.

[0058] In this embodiment, the computer subsystem can be configured to perform defect detection on samples individually using images generated from each of the different modes in the initial subset. In other words, the computer subsystem can be configured to perform a first defect detection using images generated from a first mode in the initial subset, a second defect detection using images generated from a second mode in the initial subset, and so on. Therefore, the number of defect detections performed can be equal to the number of different modes in the initial subset, and the input to each defect detection can contain only the image generated by the mode performing the defect detection.

[0059] In one such example, the computer subsystem may individually input the images generated by each of the patterns in the initial subset into a defect detection method or algorithm that can be executed in any suitable manner. The defect detection method or algorithm used for each of the different patterns may be the same. Furthermore, the defect detection method or algorithm may include any defect detection method or algorithm known in the art. For example, the defect detection algorithm may be the MDAT algorithm available on some inspection tools purchased from KLA. The defect detection algorithm may also include deep learning (DL) type defect detection algorithms.

[0060] The image input to the determination step may include any of the outputs of the imaging subsystem further described herein, such as images, image data, signals, image signals, etc. The image input to the determination step may include raw detector output, meaning the detector output has not been processed in any way before being used for defect detection. However, the image input to the determination step may include patterned output that has been processed in some way, as is typical in normal inspection (or other) processes. This processing may include, for example, high-pass filtering, image alignment, and the like. In this way, the image input to the determination step may be identical, as if it were patterned image processing for inspection (or other) processes.

[0061] Regardless of the mode, the input to the determination step may not only include processed or unprocessed images generated by the mode for the sample. For example, the input to the determination step for any mode may include additional information, such as one or more reference images corresponding to one or more sample images, one or more difference images generated by subtracting the reference sample images from the test sample image, design information of the sample, and the like. In this way, multiple input channels may exist for the determination step of any mode performed by the computer subsystem for any mode.

[0062] The inputs for the determination step can vary depending on the type of process to be performed using the ultimately selected mode. For example, the inputs for the determination step might only include mode outputs for die-to-die inspection, but the inputs for the determination step might include mode outputs and references for die-to-database inspection. Clearly, these are just two examples of the defect detection types that can be configured for the determination step, and which inputs are used for the determination step will ultimately depend on the sample process settings.

[0063] The regions of the sample for which the images are generated and input to the determining steps may be the same for each mode, but this is not necessarily the case, for example, when a mode fails to produce any output for a portion of the sample. Furthermore, the mode images used in any of the steps described herein may or may not include all images generated by any mode of the sample. In some cases, it may be advantageous to generate images for as many portions of the sample as possible using any mode (which may be determined by known or anticipated regions to be examined during sample inspection). However, to save time and resources, fewer than all of these images may be used for any of the steps performed herein for any mode. For example, instead of generating and using mode images for the entire sample in the embodiments described herein, the generated and used mode images may be mode images generated for only half of the sample, mode images generated for several limited regions on the sample (e.g., a central region and an edge region on the sample, or a limited number of bare areas), etc. In other words, the amount of mode images used for any of the steps described herein can be wisely chosen to reduce the amount of time and resources used without compromising the accuracy of the results.

[0064] In one embodiment, the initial subset of different modes comprises one or more best known patterns (BKMs) of the process performed on the sample using images generated by the imaging subsystem. For example, a variety of modes can be used to initialize the optical selection process. Additionally, as... Figure 4 As shown in step 400, the computer subsystem can select an initial subset of different patterns. In this way, the computer subsystem can select one or more BKMs as an initial subset of patterns for a given process. The computer subsystem can identify or determine BKMs in any suitable manner. For example, BKMs can be identified by searching for process recipes of the same type and similar types of samples in a storage medium (e.g., a wafer fab database or storage medium accessible to the computer subsystem) and identifying patterns in the process recipes as an initial subset. In this way, the initial subset of patterns can vary depending on the process (e.g., array inspection versus logic inspection, inspection versus metrology, etc.) and sample type (etched wafer versus photolithographic wafer, wafer versus photomask, wafer with a front-end layer formed versus wafer with a back-end layer formed, etc.).

[0065] BKM can be set as described herein or can be set using different systems or methods. For example, BKM can be a manually set recipe. However, the initial subset of patterns can be arbitrarily selected or selected for each pattern selection process. For example, when BKM or other set recipes are not available to the computer subsystem, the computer subsystem can arbitrarily select one or more patterns for the initial subset and / or may take a default subset of patterns as the initial subset.

[0066] This document describes various “subsets” of patterns, such as the initial subset, additional subsets, top-level subsets, etc. Any “subset” described herein may contain one or more patterns. Furthermore, all “subsets” described herein will contain fewer patterns than all patterns available on the tool. In this way, unless otherwise stated herein, the term “subset” of a pattern, as used herein, is defined as one or more patterns.

[0067] Once an initial subset of patterns is selected or identified, the imaging subsystem generates sample images using the selected patterns, such as... Figure 4 As shown in step 402. The imaging subsystem may generate an image in any example of step 402, as further described herein. The computer subsystem may then determine at least one value of information and characteristics of the information from the image, as shown in step 404, which may be performed as further described herein. As shown in step 406, the computer subsystem may then determine whether the current pattern under consideration is an initial subset of patterns. If the current pattern is an initial subset, then the computer subsystem may continue to search and evaluate the unexplored pattern space, as shown in steps 408 and 410, which may be performed as further described herein. If the current pattern is not an initial subset, then the computer subsystem may compare the results generated with different subsets of patterns to determine in step 412 whether the process has converged, which may be performed as further described herein.

[0068] The computer subsystem is also configured to predict the probability of determining a better value for a feature from images generated using different patterns other than the initial subset, based on at least one value determined by the feature and the relationship between the associated values ​​of features of different patterns and information. Figure 4 As illustrated in step 408. For example, the embodiments described herein utilize the fundamental principle that most optical modes are not entirely independent. In other words, different modes can be defined by at least some of the same hardware parameters. In this way, a not entirely independent mode means a mode that is not completely separated in the parameter space. Specifically, some optical modes can be considered as a combination of a smaller set of unique optical modes available on the tool.

[0069] In one such example, a mode may use a wide wavelength band, which may contain smaller wavelength bands used by other modes on the tool, and can be considered as unique narrow bands. In this way, a mode may use the same wavelength band as a combination of two or more other modes. For example, other mode parameters of the aperture configuration may have similar relationships to each other. Additionally, some mode parameters, such as focus shift, may be considered as functions of different values ​​of those parameters available on the tool. In this case, if the initial subset of different modes contains several values ​​of focus shift, then the values ​​of characteristics determined from the images produced using those modes can be used to interpolate and predict the response of characteristic values ​​to a wide range of focus shifts.

[0070] The embodiments described herein use this information to construct relations that can be used to predict the responses of optical modes given the responses of some initiating optical modes. Furthermore, the embodiments described herein can construct relations that can predict the responses of the entire optical mode space given the responses of some unique optical modes. These relations can be embodied in algorithms that can be used to predict probabilities. In this way, the input to the algorithm can be the values ​​of the characteristics of the initiating optical mode (or the current mode) and the information determined from the image produced by the initiating (or current) mode. The output of the algorithm is the probability that other modes will produce images from which better values ​​of their characteristics can be determined. In this way, the computer subsystem can use the relations to analyze the unexplored mode space (i.e., the modes of the imaging subsystem other than the initial subset of modes).

[0071] The "preferred" value can vary depending on the type of value determined for the characteristics of the information. For example, if the information is a defect detected on the sample and the value of the characteristic is the SNR value of the defect detected in the sample image, then the preferred value would be the higher SNR value of the detected defect. In another instance, if the information is a defect detected on the sample and the value of the characteristic is the number of obstructing points also detected in the image generated for the sample, then the preferred value would be the lower number of obstructing points in the defect detection result. The value can be quantitative, but it can also be qualitative. For example, if the information is a defect detected on the sample, then the qualitative value of the information could be an indication of whether the defect can be detected across the entire inspection area of ​​the sample (e.g., whether the number of obstructing points is so large that the inspection may be prematurely terminated or whether the optical mode cannot produce a useful image for a portion of the sample). In this way, the preferred qualitative value would be an indication of the entire inspection area of ​​the sample that the optical mode can be used to inspect.

[0072] In one embodiment, relations are determined through reinforcement learning. Therefore, a key novel feature of the embodiments described herein is the use of machine learning (ML) methods to address the problem of selecting optical patterns for defects. Reinforcement learning is a type of ML that can be used to make decisions to achieve optimal outcomes. Reinforcement learning differs from supervised and unsupervised learning and is generally based on Markov decision processes. Reinforcement learning for determining relations can use any suitable algorithm known in the field. Reinforcement learning can also be performed in a deep learning (DL) configuration using deep neural networks. In any of these embodiments, reinforcement learning can be model-based or model-free.

[0073] In model-based reinforcement learning, the physics of the imaging subsystem can be used to construct a representation or model of the environment, i.e., the relations described herein. Model-based reinforcement learning can be useful in the embodiments described herein because the physics of the imaging subsystem can be relatively well-defined and invariant. Relations can be constructed by taking actions from state to state within the environment, identifying new states and their associated values, and then associating action-state transitions with values. Once relations are constructed, the computer subsystem or another component can model sequences through relations based on the probability of the optimal accumulated value. In this way, values ​​can also be assigned to action sequences. Additionally, different policies within the environment can be learned to achieve the desired objective.

[0074] In some embodiments, the relationship is determined through reinforcement learning, where different patterns are states and associated values ​​are the values ​​of those states. For example, in the case of SNR values, embodiments may use reinforcement learning methods where optical patterns are considered states and associated SNRs are considered the values ​​of those states. This relationship configuration may be suitable for situations where the imaging subsystem is an inspection subsystem and when the information determined from a sample image contains defects detected in the sample image. However, this relationship can also be used in other processes, such as metrology, where high-signal, low-noise images are more suitable for determining metrological information about a sample. However, the states and the values ​​of the states may vary based on the imaging subsystem and the sample. In other words, the relationship may be defined based on the patterns of the imaging subsystem and any values ​​of any characteristics of the information about the sample determined from the image generated from the patterns.

[0075] In another embodiment, the predicted probabilities are not trained using pre-existing data. For example, a key novel feature of the embodiments described herein is the idea of ​​using a physics-added prediction algorithm instead of training on pre-existing data. More specifically, if the embodiments described herein were configured to recommend patterns using a learning algorithm set up through supervised or unsupervised training, the embodiments would have to train and use a trained algorithm to predict patterns for new wafers using a large amount of pre-collected optical and SNR data from different wafers, wafer layers, and tools. In contrast, the embodiments described herein do not use any pre-collected data. On each new wafer, the embodiments run in each iteration using results (e.g., SNR values) from an initial pattern and patterns predicted by a computer subsystem. Relationships (e.g., nodes and connections in a graph) are constructed using knowledge of how the components of the optical pattern are connected in parameter space. In this way, the embodiments use a physics-added prediction algorithm because the relationships (e.g., graph structure) are static and fixed for a given tool model.

[0076] In another embodiment, at least one value of the feature and its associated value are the SNR value of the defect detected in the image. For example, in a reinforcement learning method where optical modes are considered states and associated SNRs are considered values ​​of those states, the computer subsystem can construct transitions connecting various modes based on physics. In this way, the probability prediction step can include predicting the probability of transitions on any axis to further improve the SNR. As further described above, the associated values ​​of states in the relation can vary depending on the imaging subsystem and the samples. Additionally, the values ​​of the features of the information determined by the embodiments described herein should be the same as the associated values ​​of the features in the relation. For example, if the embodiments described herein aim to find the mode with the highest SNR value for defects detected in the image, then both at least one value of the features of the information determined from images generated with subsets of different modes and the associated value of the features in the relation should be the SNR value of the defect detected in the image.

[0077] In an additional embodiment, the relationship is based on the physics of the imaging subsystem, and the system is configured to select a mode for a mode selection process performed on different samples, producing images of the different samples in the processes performed on those samples. Specifically, because the relationships described herein (by a computer subsystem or another system or method) are generated based on the physics of the imaging subsystem (i.e., the modes available on the imaging subsystem and the transitions or connections between them), the same relationships can be used for different samples and different formulations. For example, the patterns (e.g., the patterns described herein) can be the same for different stages of a fabrication process or wafer. In other words, as long as the imaging subsystem and its mode space remain unchanged, the same relationships can be used in the mode selection process to select modes for significantly different formulations (e.g., array pattern inspection versus logic inspection) and for significantly different samples (e.g., different types of wafers). The same relationships can also be used for different examples of the same tool configuration, such as multiple inspection tools of the same brand and model. In this way, the relationships described herein can change with the tool type. Additionally, if the optical parameter space of the tool changes, the relationships can be modified and / or updated. The reusability of relationships (and any graphs generated from relationships) offers a significant advantage in producing different formulations for different samples executed with the same tools.

[0078] In another embodiment, the computer subsystem is configured to determine relations through reinforcement learning. For example, the computer subsystem (and therefore the system described herein) may be configured to generate relations and then use them, as described herein. However, another system or method may be configured to determine relations, and then the embodiments described herein may access or retrieve relations (e.g., in a computer-readable storage medium that has been stored by another system or method) and use the relations, as described herein. In this way, one system or method can generate relations and another system or method can use them.

[0079] In another embodiment, the computer subsystem is configured to generate a graph visualizing relationships. The relationships and the graph visualizing those relationships can be generated using reinforcement learning, as further described herein. In another embodiment, different modes encompass the entire optical mode space for the imaging subsystem. For example, the entire optical mode space can be visualized as a large graph when optical modes are considered states, associated SNR values ​​are considered values ​​of those states, and the computer subsystem constructs transitions connecting the various modes based on the physics of the imaging subsystem. The computer subsystem can be configured to construct the graph by generating states based on the physics and modes of the imaging subsystem. The values ​​of the states can be determined by the computer subsystem (e.g., from images captured using a tool). Therefore, the associated values ​​of the states may not be static or pre-computed values. For example, for each run performed, the computer subsystem can collect values ​​(e.g., SNR values) and populate the graph with those values.

[0080] Figure 3 This illustrates one embodiment of the graph. In graph 300, each of states 302 corresponds to a parameter value of an imaging subsystem of a mode or a definable mode, and transition 304 connects the various modes. This can produce... Figure 3 The graphics shown illustrate the fully connected optical mode space. (See the image below.) Figure 3 As can be seen, the pattern of the entire optical mode space can be quite complex, especially when there are many variable imaging subsystem parameters such as wavelength, aperture, polarization, and focus shift.

[0081] The computer subsystem may be configured to display graphics to a user (e.g., using a display device (not shown)) to facilitate the user's understanding of the relationship between patterns and their associated values. The computer subsystem may be configured to further facilitate the user's understanding of the graphics in various ways (not shown). For example, the state of a subset of patterns currently being evaluated in the pattern space may be indicated in the graphics (e.g., through highlighting, specific colors, etc.). Previously explored and unexplored patterns in the pattern space may also be distinguished from each other through highlighting, different colors, etc. Transitions between patterns may also be similarly indicated in the graphics for previously explored, currently explored, and unexplored patterns.

[0082] In a similar manner, a computer subsystem can be configured to extract... Figure 3 The graphic displayed corresponds to a small portion of the entire mode space, and the extracted portion of the graphic is displayed to the user separately. For example, a portion of the entire mode space showing only the transitions between wavelength parameters of different modes can be extracted by the computer subsystem and displayed to the user. Additionally, unlike a single graphic (e.g....), Figure 3 The graphs shown herein are used for the entire mode space of the imaging subsystem. The computer subsystem can be configured to generate several smaller graphs, each of which can be generated for a mode parameter, such as the illumination wavelength, where different values ​​of the mode parameter represent states and transitions between them. Combinations of multiple such individual mode parameter graphs can cover the entire mode space of the imaging subsystem.

[0083] The computer subsystem can also be configured with graphics and a UI for displaying graphics to the user, such that, for example, information about a pattern or transition can be displayed to the user by clicking on any state or transition. The UI included in the system for such purposes can have any suitable configuration known in the field.

[0084] In one embodiment, the computer subsystem is configured to predict additional probabilities of determining additional better values ​​for features from images generated using additional different modes from an imaging subsystem other than the initial subset, based on additional relationships between at least one determined value of the feature and additional associated values ​​of the feature with additional different modes and information. In this manner, for any imaging subsystem, there may be more than one relationship (each of which may be generated individually as further described herein), and the computer subsystem may perform the probability prediction step individually and independently with each different relationship. The at least one determined value of the feature used with each different relationship may be the same. For example, the SNR value of a defect detected in an image generated with an initial subset of different modes may be used together with each of the different relationships to predict different probabilities. The probabilities predicted with each relationship can be used to select other subsets of different modes, as described herein. In other words, the probabilities predicted with different relationships may be input together into the selection steps described herein, such that subsequent modes can be selected based on all predicted probabilities.

[0085] In one embodiment, different modes and additional different modes are defined by at least one mutually exclusive parameter of the imaging subsystem. A mode defined by at least one “mutually exclusive parameter” of the imaging subsystem can be any mode having one or more parameters that are different such that information determined by one of the modes cannot be used to predict the probability described herein for another mode. Such modes can be identified based on the physics of the imaging subsystem, and then different relationships can be generated and used as described herein for such modes.

[0086] In another embodiment, different modes and additional different modes are defined by at least one orthogonal and non-overlapping parameter of the imaging subsystem. For example, a computer subsystem may be configured to generate the aforementioned patterns for mode components intended to produce orthogonal information. In one example, mutually exclusive illumination polarizations can generate orthogonal information. Therefore, different relationships can be generated for such modes and can be used to individually evaluate their corresponding modes.

[0087] The concept of interdependence of modes forms the basis of how embodiments evaluate optical modes. Reinforcement learning is only one way to effectively utilize this concept. For example, another method that can be used by the embodiments described herein to predict probabilities is to predict an image generated from an unexplored optical mode in view of images of several modes (e.g., an initial subset of modes or several unique modes). A computer subsystem can use such predicted images to compute a predicted SNR, which, for example, can advantageously eliminate the need to run optical modes to obtain an SNR. Image prediction can be performed by physics-based addition of images from independent mode components or by using a deep generation algorithm (which can generate images of unexplored modes using pre-existing data or SEM / design images).

[0088] The computer subsystem is further configured to select, based on predicted probabilities, an additional subset of different modes, which are then used by the imaging and computer subsystems to perform generation and determination steps, respectively. Figure 4 As shown in step 410. Therefore, the computer subsystem selects a potentially good pattern or a set of potentially good patterns based on predicted probabilities. In this way, the recommendation system embodiments described herein can find a minimal set of optical patterns that can be run on the tool to reach the top pattern. In one embodiment, selecting an additional subset of different patterns includes selecting one or more of the different patterns, excluding the initial subset, where the highest predicted probability value is found. For example, the pattern with the highest probability can be selected and run (i.e., the sample can be scanned with the selected pattern to generate an image with the selected pattern), and a new SNR score can be determined for the new pattern. In this way, the embodiments described herein provide a deterministic way of selecting evaluation patterns during the recipe setting phase. Additionally, as further described herein, the embodiments provide a deterministic way of identifying when the pattern selection process has converged (meaning the pattern evaluation results have stopped changing).

[0089] Performing these deterministic steps in an automated manner enables the embodiments described herein to converge after evaluating a relatively small number of all available patterns. For example, experiments performed by the inventors using the embodiments described herein for two different types of wafers were able to predict the top pattern of the inspection process for each of the wafer types within 40 reviewed patterns without any user intervention. Because the top pattern can be identified by reviewing only 40 patterns (rather than the thousands available on the inspection tool) without any user intervention, the embodiments described herein can identify the top pattern substantially faster than currently used methods or systems, which do not deterministically select evaluation patterns and / or where every pattern on the tool must be run to generate an image and determine information from the image.

[0090] In some embodiments, the imaging and computing subsystems are configured to perform generation and determination steps for additional subsets of different modes, respectively. For example, the imaging subsystem may be configured to generate images of samples using additional subsets of modes, and the computing subsystem may then determine information and at least one value of a characteristic of the information from the images generated using additional subsets of different modes. These steps may be performed as further described herein. For example, the imaging subsystem may repeat the process for additional subsets of different modes. Figure 4 The computer subsystem may repeat step 404 for images generated using an additional subset of the pattern, as shown in step 402. When steps 402 and 404 are repeated for an additional subset of the pattern, in step 406, the computer subsystem determines that the current pattern is not an initial subset of the pattern and then proceeds to step 412, which is further described herein.

[0091] In one embodiment, the computer subsystem is configured to update the predicted probabilities based on at least one value of a characteristic determined with an additional subset of different modes. For example, using a new SNR score, the computer subsystem can update the probabilities of all possible shifts to further improve the SNR. In other words, a newly selected subset of different modes can be run on the imaging subsystem, the value of the characteristic of the information determined with the new mode can be used to update the predicted probabilities, and then a new mode can be selected again based on the updated probabilities. In this way, different subsets of modes can be iteratively selected and analyzed based on the results of previously selected modes. For example, this iterative process can be performed when the value of the determined characteristic in step 412 does not converge. Alternatively, the predicted probabilities can be updated by inputting the latest mode results into the relation and generating new values ​​for the predicted probabilities. The updated predicted probabilities can then be input to step 410, where the next subset of the mode can be selected.

[0092] In another embodiment, the computer subsystem is configured to compare information from samples determined from additional and initial subsets of images generated using different modes and, based on the result of the comparison step, determine whether selecting additional subsets of different modes converges to a top-level subset of the different modes of the process performed on the samples. For example, once the computer subsystem determines in step 406 that the currently considered mode is not an initial subset of modes, the computer subsystem can determine in step 412 whether the value of the characteristic determined with the currently considered mode has converged. A key novel feature of the embodiments described herein is the reliance on convergence criteria to stop exploring new modes. For example, a major challenge faced by users in mode selection is not knowing when to stop. The embodiments described herein advantageously address this challenge using the convergence criteria and steps described herein.

[0093] In one embodiment, the computer subsystem is configured to determine that the selection step has converged to the top subset of different modes when the comparison results indicate that there is no improvement in the information of the samples determined from the images generated with additional and initial subsets of different modes. For example, the next or additional subset of different modes with the highest probabilities can be run, and the search can continue until the computer subsystem does not see any further improvement (e.g., an increase in SNR, a decrease in the number of detected perturbation points, etc.). At this point, the search can be referred to as convergence.

[0094] As shown in step 414, when the computer subsystem determines in step 412 that the values ​​of the properties determined from the image generated with the considered pattern have converged, the computer subsystem identifies the pattern to be used in the process. This step can be performed as further described herein. Alternatively, when convergence is detected, the process can be stopped, and any or all results generated prior to this can be output by the computer subsystem, as further described herein, allowing another method or system to select the final pattern to be used in the process performed on the sample. In other words, one system or method can be configured for pattern recommendation and pattern analysis, and another system or method can be configured for final pattern selection.

[0095] If the computer subsystem determines in step 412 that the value of the feature determined from the image generated with the current mode has not yet converged, then the computer subsystem may repeat step 408 to predict the probability of determining a better value of the feature from images of other modes in the still-to-be-explored mode space. Another subset of modes may be selected in step 410 based on the new predicted probabilities, and steps 402, 404, 406, and 412 may be repeated for this new subset of modes. In this way, the imaging and computer subsystem may repeat the mode review and mode space exploration steps until convergence of the mode results is detected.

[0096] Before or after the computer subsystem determines that the values ​​of the characteristics for pattern determination have ceased to change or show no further improvement, the computer subsystem may perform additional analysis on all patterns examined so far. For example, this analysis may be performed to eliminate the need to further consider patterns that may include those already considered (via image generation and information determination of samples from images) and / or those not yet considered in this manner. For example, based on some results of the steps described herein (e.g., the information determination step), some unconsidered patterns may be eliminated in other steps, such as the probability prediction step. In one such example, if one of the patterns already considered by image generation and information determination shows that it is completely unsuitable for the sample and the pattern selection process, then other patterns in the currently unexplored space that are substantially similar in one or more imaging parameters may be completely eliminated. Of course, such patterns can be identified by the probability prediction step, but they can also be eliminated based on their similarity to the considered patterns.

[0097] The analysis may also include ranking patterns that have been considered and / or are in the unexplored space. For example, at any point during pattern analysis and selection, the computer subsystem may rank patterns from which images have been generated based on values ​​of characteristics that determine information about the samples using their corresponding images. Considering the example of SNR values, the computer subsystem may rank the run patterns from best to worst SNR values. This ranking may be performed through the pattern selection process to track which patterns(s) are best. Ranking may also be performed on the unexplored space based on, for example, predicted probabilities. In one such example, once the predicted probability step is performed, the computer subsystem may rank the unexplored patterns from best to worst predicted probability values, making it possible to select (in step 410) and run (in steps 402 and 404) unexplored patterns, for example, from patterns more likely to be good to patterns less likely to be good.

[0098] The embodiments described herein can be used to select patterns for single-mode and multi-mode processes in several different ways. For example, the relationships described herein can be generated only for a single mode. Top-level single modes can be identified and selected for use in a sample process as described herein. Relationships can also be generated for one or more combinations of modes, or alternatively. For example, each state in the graph generated for the relationship can correspond to a combination of modes. This graph can also include states of single modes. In this way, the probabilities described herein can be predicted in the same steps and using the same relationships for both single modes and combinations of multiple modes, and the modes selected for analysis can then include single-mode configurations and / or multi-mode tool configurations.

[0099] Depending on the approach, the embodiments described herein can be configured to analyze and select only a single mode, and a top-level single mode can be identified for use in the process. Any identified top-level single mode can be combined with one or more other modes to produce different mode combinations. The different mode combinations can then be analyzed as described herein or in any other suitable manner known in the art. This analysis can be performed to determine whether an image generated using the identified top-level single mode can be combined with an image generated using one or more other modes to synergistically improve the process results. In this manner, the embodiments described herein can first select a single optimal mode and then determine whether one or more other modes can be combined with it to further improve its performance.

[0100] Once a mode is selected for a process, the computer subsystem or another method or system may select one or more other parameters for the process. Such parameters may include any image processing parameters, any parameters of the method or system used to determine information from the image generated using the selected mode, any parameters of the method or system used to determine additional information about the sample from the determined information, any parameters of the method or system used to produce the results of the process, etc. To give just one example, the computer subsystem may be configured to determine a threshold for a defect detection method to be applied to the image generated by the selected mode to detect defects in the image. Such parameters may be selected in any suitable manner known in the art.

[0101] In one embodiment, the steps of generating, determining, predicting, and selecting are performed without user input. For example, the embodiments described herein eliminate all user dependence when selecting an optical mode to be evaluated for a given sample layer and reliably predict the optimal optical mode with a minimum number of mode evaluations. Specifically, the embodiments described herein provide an ML-based process that can fully automate mode selection. Therefore, the embodiments described herein differ from currently used methods, where steps such as analyzing an unexplored mode space, selecting a subset of potentially good modes, and determining whether a suitable mode has been found are performed manually by the user.

[0102] Compared to currently used methods and systems, the embodiments described herein offer numerous significant improvements and advantages in pattern selection. One advantage is that the embodiments can grant recipes at every layer. Another advantage is that optical pattern selection is a difficult skill to train. Therefore, the embodiments can significantly reduce application engineering training. The embodiments described herein will also advantageously save users of both research and development and ramp-stage tools significant tooling time, as the number of patterns evaluated will be more controlled and deterministic. Improved user satisfaction is another advantage of the embodiments described herein, as users do not rely on expert application engineers for recipe setup. Another advantage of the embodiments described herein is that the selected top-level pattern does not change based on the user, as the process is deterministic.

[0103] An additional advantage of the embodiments described herein is that the recommendation system can neutralize some possible forms of user bias (e.g., bias towards defect discovery patterns) because it is designed to explore the pattern space until convergence. For example, defect discovery may be performed before optical selection, where multiple patterns are used to attempt to discover the types of defects that may be present on a sample. Then, during the currently performed optical selection, the user may be biased (or reject) certain patterns that capture (or miss) defects during the discovery process. However, the patterns used for defect discovery may not operate under optimal conditions, and without using the embodiments described herein, pattern selection may be performed without considering the entire optical pattern space and may result in the selection of suboptimal patterns for the inspection process. The embodiments described herein completely eliminate any such user bias towards the pattern selection process.

[0104] In some embodiments, the computer subsystem is configured to store information about a selected pattern for subsequent steps in the pattern selection process and / or for processes such as sample inspection. The computer subsystem may be configured to store information in a recipe or by generating a recipe in which a process will use the selected pattern. As used herein, the term "recipe" is generally defined as a set of instructions that can be used by a tool to perform a process on a sample. In this way, generating a recipe may include information about how the process will be performed, which can then be used to generate instructions for performing the process. The information about the selected pattern stored by the computer subsystem may include any information that can be used to identify and / or use the selected pattern (e.g., filename and its storage location, and the file may contain information about the pattern, such as the pattern name, pattern parameter values, etc.).

[0105] A computer subsystem may be configured to store information of a selected pattern in any suitable computer-readable storage medium. The information may be stored along with any results described herein and may be stored in any manner known in the art. The storage medium may include any storage medium described herein or any other suitable storage medium known in the art. After the information is stored, it may be accessed in the storage medium and used by any of the methods or system embodiments described herein, formatted for display to a user, used by another software module, method, or system, etc. For example, the embodiments described herein may generate the inspection recipe described above. The inspection recipe may then be stored and used by a system or method (or another system or method) to inspect samples or other samples to generate information about the samples or other samples (e.g., defect information).

[0106] The computer subsystem and / or inspection system may be configured to perform an inspection process on a sample and / or other samples of the same type using the results of one or more steps described herein. This inspection process may produce results for any defects detected on the sample, such as information about the bounding box of the detected defects (e.g., location), inspection scores, information about defect classification (e.g., category labels or IDs), or any such suitable information known in the field. The results of the defects may be generated by the computer subsystem and / or inspection system in any suitable manner. The results of the defects may have any suitable form or format, such as standard document types. The computer subsystem and / or inspection system may generate and store the results, such that the results may be used by the computer subsystem and / or another system or method to perform one or more functions on the sample or another sample of the same type. For example, the information may be used by the computer subsystem or another system or method to sample defects for defect re-inspection or other analysis, determine the root cause of defects, etc.

[0107] Similarly, the computer subsystems and / or other imaging subsystems (e.g., metrology or defect re-inspection subsystems) described herein can be configured to use the results of one or more steps described herein to perform other processes on samples and / or other samples of the same type. Such metrology or defect re-inspection processes can produce any suitable results known in the art, and the results of such processes can be generated and / or stored by the computer subsystems and / or imaging subsystems described herein. Those results can then be used in the same manner as described herein.

[0108] Functions that can be performed using this information also include (but are not limited to) modifying, in a feedback or feedforward manner, the processes, such as manufacturing processes or steps, performed on or to be performed on the inspected sample or another sample. For example, a computer subsystem may be configured to determine one or more modifications to the processes performed on a sample inspected as described herein and / or to the processes to be performed on the sample based on detected defects. The modifications to the processes may include any suitable changes to one or more parameters of the processes. The computer subsystem preferably determines those modifications such that defects can be reduced or prevented on other samples to which a corrective process is performed, defects on the sample can be corrected or eliminated in another process performed on the sample, defects can be compensated for in another process performed on the sample, etc. The computer subsystem may determine such modifications in any suitable manner known in the art. Such modifications may also be determined using the results of other processes described herein.

[0109] Those changes can then be sent to a semiconductor fabrication system (not shown) or a computer subsystem and a storage medium (not shown) accessible by the semiconductor fabrication system. The semiconductor fabrication system may or may not be part of the system embodiments described herein. For example, the computer subsystem and / or inspection system described herein may be coupled to the semiconductor fabrication system via one or more common elements (e.g., housing, power supply, sample handling device, or mechanism, etc.). The semiconductor fabrication system may include any semiconductor fabrication system known in the art, such as lithography tools, etching tools, chemical mechanical polishing (CMP) tools, deposition tools, and the like.

[0110] Therefore, as described herein, the embodiments can be used to set up new processes or formulations. The embodiments can also be used to modify existing processes or formulations, whether they are processes or formulations for a sample or processes or formulations developed for one sample and applicable to another.

[0111] The above embodiments can be combined together to form a single embodiment. In other words, unless otherwise stated herein, none of the embodiments are mutually exclusive with any other embodiments.

[0112] Another embodiment relates to a computer-implemented method for selecting a mode for a mode selection process. The method includes generating images of a sample using different modes of an imaging subsystem, for example, as shown in step 402. The method also includes determining at least one value of information about the sample and a characteristic of the information from images generated using an initial subset of different modes, for example, as shown in step 404. Additionally, the method includes predicting a probability of determining a better value of the characteristic from images generated using different modes other than the initial subset, based on the relationship between the determined at least one value of the characteristic and the associated values ​​of different modes and the characteristic of the information, for example, as shown in step 408. The method further includes selecting an additional subset of different modes, based on the predicted probability, for which the imaging subsystem and computer subsystem respectively perform the generation and determination steps, for example, as shown in step 410. The determination, prediction, and selection steps are performed by the computer subsystem.

[0113] Each of the steps in the method may be performed as further described herein. The method may also include any other steps that may be performed by the imaging subsystem and / or computer subsystem described herein. Furthermore, the above method may be performed by any of the system embodiments described herein.

[0114] Additional embodiments relate to a non-transitory computer-readable medium storing program instructions executable on a computer system to perform a computer-implemented method for selecting a mode for a mode selection process. Figure 5 An example of this embodiment is shown below. Specifically, as... Figure 5As shown herein, the non-transitory computer-readable medium 500 contains program instructions 502 that can be executed on a computer system 504. A computer-implemented method may include any step of any method described herein.

[0115] Program instructions 502 for implementing the method (such as those described herein) may be stored on a computer-readable medium 500. The computer-readable medium may be a storage medium, such as a disk or optical disc, magnetic tape, or any other suitable non-transitory computer-readable medium known in the art.

[0116] Program instructions can be implemented in any of a variety of ways, including procedural, component-based, and / or object-oriented technologies. For example, program instructions can be implemented using ActiveX controls, C++ objects, JavaBeans, Microsoft Foundation Classes (“MFC”), SSE (Streaming SIMD Extensions), Python, Tensorflow, or other technologies or methods as needed.

[0117] Computer system 504 may be configured according to any of the embodiments described herein.

[0118] Based on this description, those skilled in the art will understand further modifications and alternative embodiments of various aspects of the invention. For example, methods and systems for selecting a mode for a mode selection process are provided. Therefore, this description is merely illustrative and intended to teach those skilled in the art the general manner of implementing the invention. It should be understood that the forms of the invention shown and described herein should be considered as the currently preferred embodiments. Elements and materials may be substituted for those described herein, components and processes may be reversed, and certain properties of the invention may be utilized independently, all of which will be understood by those skilled in the art after benefiting from this description of the invention. Changes may be made to the elements described herein without departing from the spirit and scope of the invention as set forth in the appended claims.

Claims

1. A system configured to select a mode for a mode selection process, comprising: An imaging subsystem configured to generate images of samples using different modes of the imaging subsystem; and Computer subsystem, configured for: Determine information about the sample and at least one value of the characteristics of the information from the images generated from the initial subsets of the different patterns; Based on the relationship between the determined at least one value of the characteristic and the associated values ​​of the different patterns and the characteristic of the information, predict the probability of determining a better value of the characteristic from the image generated using the different patterns other than the initial subset; and Based on the predicted probabilities, an additional subset of the different modes is selected, which is then used by the imaging and computer subsystems to perform the generation and determination steps.

2. The system according to claim 1, wherein the relationship is determined by reinforcement learning.

3. The system of claim 1, wherein the computer subsystem is further configured to determine the relationship by reinforcement learning.

4. The system of claim 1, wherein the relationship is determined by reinforcement learning of the different patterns being states and the associated values ​​being values ​​of the states.

5. The system of claim 1, wherein the at least one value of the characteristic and the associated value of the characteristic are signal-to-noise ratio values ​​of defects detected in the image.

6. The system of claim 1, wherein the relationship is based on the physics of the imaging subsystem, and wherein the system is further configured to select a mode for a mode selection process performed on different samples, the imaging subsystem generating images of the different samples in a process performed on the different samples.

7. The system of claim 1, wherein the computer subsystem is further configured to generate a graph visualizing the relationship.

8. The system of claim 1, wherein the different modes comprise the entire optical mode space for the imaging subsystem.

9. The system of claim 1, wherein the computer subsystem is further configured to predict additional probabilities of determining additional preferred values ​​of the characteristic from images generated using the additional different modes other than the initial subset, based on the determined at least one value of the characteristic and additional relationships between additional different modes of the imaging subsystem and additional associated values ​​of the characteristic of the information.

10. The system of claim 9, wherein the different modes and the additional different modes are defined by at least one mutually exclusive parameter of the imaging subsystem.

11. The system of claim 9, wherein the different modes and the additional different modes are defined by at least one orthogonal and non-overlapping parameter of the imaging subsystem.

12. The system of claim 1, wherein the imaging and computer subsystem is further configured to perform the generation and determination steps on the additional subsets of the different modes respectively, and wherein the computer subsystem is further configured to compare the information of the samples determined from the images generated with the additional and initial subsets of the different modes, and to determine, based on the result of the comparison, whether the selection of the additional subsets of the different modes converges to the top subset of the different modes of the process performed on the samples.

13. The system of claim 12, wherein the computer subsystem is further configured to determine that the selection has converged to the top subset of the different modes when the result of the comparison indicates that the information of the samples determined from the images generated with the additional and initial subsets of the different modes has not improved.

14. The system of claim 1, wherein the imaging and computer subsystem is further configured to perform the generation and determination steps for the additional subsets of the different modes respectively, and wherein the computer subsystem is further configured to update the predicted probability based on the at least one value of the characteristic determined by the additional subsets of the different modes.

15. The system of claim 1, wherein the initial subset of the different modes comprises one or more best-known modes of the process performed on the sample using the image generated by the imaging subsystem.

16. The system of claim 1, wherein selecting the additional subset of the different modes comprises selecting one or more of the different modes other than the initial subset in which the highest value of the predicted probability is obtained.

17. The system of claim 1, wherein the generating, determining, predicting, and selecting steps are performed without user input.

18. The system of claim 1, wherein the probability prediction is not trained using pre-existing data.

19. A non-transitory computer-readable medium storing program instructions executable on a computer system to perform a computer-implemented method for selecting a mode for a mode selection process, wherein the computer-implemented method includes: Images of the samples are generated using different modes of the imaging subsystem; Determine information about the sample and at least one value of the characteristics of the information from the images generated from the initial subsets of the different patterns; Based on the relationship between the determined at least one value of the characteristic and the associated values ​​of the different patterns and the characteristic of the information, predict the probability of determining a better value of the characteristic from the image generated using the different patterns other than the initial subset; and Based on the predicted probabilities, an additional subset of the different modes is selected, which is then used by the imaging and computer subsystems to perform the generation and determination steps.

20. A computer-implemented method for selecting a pattern for a pattern selection process, comprising: Images of the samples are generated using different modes of the imaging subsystem; Determine information about the sample and at least one value of the characteristics of the information from the images generated from the initial subsets of the different patterns; Based on the relationship between the determined at least one value of the characteristic and the associated values ​​of the different patterns and the characteristic of the information, predict the probability of determining a better value of the characteristic from the image generated using the different patterns other than the initial subset; and Based on the predicted probabilities, an additional subset of the different modes is selected, which is then used by the imaging subsystem and the computer subsystem to perform the generation and determination steps, respectively, wherein the determination, prediction, and selection are performed by the computer subsystem.

Images