In-device overlay metrology using inverse die-to-database alignment

Abstract

A charged particle beam apparatus includes a charged particle beam source, a charged particle optical system, a charged particle detector, and a controller. The charged particle optical directs a beam of primary charged particles at a sample surface. The controller determines a contour of a target feature disposed at the sample surface based on detected charged particles at the detector. A first layer structure and a second layer structure of the target feature are based on a design pattern that includes a first layer design pattern and a second layer design pattern. The controller determines an alignment position of the first layer structure, via an inverse die-to-database (iD2DB) alignment that includes a search for at least a portion of the first layer design pattern in a search area defined by the determined contour of the target feature, an alignment position of the second layer structure, and a difference thereof.

Description

IN-DEVICE OVERLAY METROLOGY USING INVERSE DIE-TQ-DATABASE ALIGNMENTCROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims priority of International application PCT / CN2024 / 099770 which was filed on 18 June 2024, and which is incorporated herein in its entirety by reference.FIELD

[0002] The description herein relates to systems and methods directed to overlay metrology using image analysis.BACKGROUND

[0003] In manufacturing processes of integrated circuits (ICs), unfinished or finished circuit components are inspected to ensure that they are manufactured according to design and are free of defects. Inspection can be performed using systems such as optical microscopes or charged particle beam microscopes (e.g., a scanning electron microscope (SEM)). As the sizes of IC components continue to shrink, accuracy and speed of defect detection become more and more important for yield and throughput. However, imaging resolution and throughput of inspection tools struggle to keep pace with the ever-decreasing feature size of IC components.SUMMARY

[0004] Embodiments of the present disclosure provide a system and method for reliable and fast overlay metrology using an inverse die-to-database alignment method.

[0005] In some embodiments, a charged particle beam apparatus with inverse die-to-database alignment function is provided. The charged particle beam apparatus can comprise a charged particle beam source, a charged particle optical system, a charged particle detector, and a controller. The charged particle beam source can generate a beam of primary charged particles. The charged particle optical system can direct the beam of primary charged particles at a sample surface. The charged particle detector can detect charged particles scattered from the sample surface. The controller can comprise circuitry to determine a contour of a target feature disposed at the sample surface based on the detected charged particles. A first layer structure and a second layer structure of the target feature can be based on a design pattern comprising a first layer design pattern and a second layer design pattern. The circuitry can also determine an alignment position of the first layer structure via an inverse die-to-database (iD2DB) alignment. The iD2DB alignment comprises a search for at least a portion of the first layer design pattern in a search area defined by the determined contour of the target feature. The circuitry can also determine an alignment position of the second layer structure. The circuitry can also determine a difference between the alignment position of the first layer structure and the alignment position of the second layer structure.

[0006] In some embodiments, a non-transitory computer-readable medium that stores a set of instructions for performing an inverse die-to-database alignment and calculating overlay shift is provided. The instructions are executable by at least one processor of an apparatus to cause the apparatus to perform operations. The operations can comprise determining a contour of a target feature disposed at a sample surface based on an inspection of the sample surface. A first layer structure and a second layer structure of the target feature can be based on a design pattern comprising a first layer design pattern and a second layer design pattern. The operations can also comprise determining an alignment position of the first layer structure via an inverse die-to-database (iD2DB) alignment. The iD2DB alignment can comprise searching for at least a portion of the first layer design pattern in a search area defined by the determined contour of the target feature. The operations can also comprise determining an alignment position of the second layer structure. The operations can also comprise determining a difference between the alignment position of the first layer structure and the alignment position of the second layer structure.

[0007] In some embodiments, a method for performing an inverse die-to-database alignment and calculating overlay shift is provided. The method can comprise inspecting a sample surface. The method can also comprise determining a contour of a target feature disposed at the sample surface based on the inspecting. A first layer structure and a second layer structure of the target feature can be based on a design pattern comprising a first layer design pattern and a second layer design pattern. The method can also comprise determining an alignment position of the first layer structure via an inverse die-to-database (iD2DB) alignment. The iD2DB alignment can comprise searching for at least a portion of the first layer design pattern in a search area defined by the determined contour of the target feature. The method can also comprise determining an alignment position of the second layer structure. The method can also comprise determining a difference between the alignment position of the first layer structure and the alignment position of the second layer structure.

[0008] In some embodiments, a charged particle beam apparatus with inverse die-to-database alignment function is provided. The charged particle beam apparatus can comprise a charged particle beam source, a charged particle optical system, a charged particle detector, and a controller. The charged particle beam source can generate a beam of primary charged particles. The charged particle optical system can direct the beam of primary charged particles at a sample surface. The charged particle detector can detect charged particles scattered from the sample surface to generate an image of a target feature disposed at the sample surface. The controller comprise circuitry to extract a contour of the target feature from the image. A first layer structure and a second layer structure of the target feature can be based on a design pattern comprising a first layer design pattern and a second layer design pattern. The circuitry can also determine an alignment position of the first layer structure via an inverse die-to-database (iD2DB) alignment. The iD2DB alignment can comprise a search for at least a portion of the first layer design pattern in a search area defined by the determined contour of the target feature. The circuitry can also split a contour of the second layer structure from a contourthe first layer structure based on the determined alignment position of the first layer structure. The circuitry can also determine an alignment position of the second layer structure based on the contour of the second layer structure being split from the contour of the first layer structure. The circuitry can also determine a difference between the alignment position of the first layer structure and the alignment position of the second layer structure.

[0009] In some embodiments, a non-transitory computer-readable medium that stores a set of instructions for performing an inverse die-to-database alignment and calculating overlay shift is provided. The instructions are executable by at least one processor of an apparatus to cause the apparatus to perform operations. The operations can comprise extracting, from an image, a contour of a target feature disposed at a sample surface. The image can be based on an inspection of the sample surface. A first layer structure and a second layer structure of the target feature can be based on a design pattern comprising a first layer design pattern and a second layer design pattern. The operations can also comprise determining an alignment position of the first layer structure via an inverse die-to-database (iD2DB) alignment. The iD2DB alignment can comprise searching for at least a portion of the first layer design pattern in a search area defined by the determined contour of the target feature. The operations can also comprise splitting a contour of the second layer structure from a contour of the first layer structure based on the determined alignment position of the first layer structure. The operations can also comprise determining an alignment position of the second layer structure based on the contour of the second layer structure being split from the contour of the first layer structure. The operations can also comprise determining a difference between the alignment position of the first layer structure and the alignment position of the second layer structure.

[0010] In some embodiments, a method for performing an inverse die-to-database alignment and calculating overlay shift is provided. The method can comprise extracting, from an image, a contour of a target feature disposed at a sample surface. The image can be based on an inspection of the sample surface. A first layer structure and a second layer structure of the target feature can be based on a design pattern comprising a first layer design pattern and a second layer design pattern. The method can also comprise determining an alignment position of the first layer structure via an inverse die-to-database (D2DB) alignment. The iD2DB alignment can comprise searching for at least a portion of the first layer design pattern in a search area defined by the determined contour of the target feature. The method can also comprise splitting a contour of the second layer structure from a contour of the first layer structure based on the determined alignment position of the first layer structure. The method can also comprise determining an alignment position of the second layer structure based on the contour of the second layer structure being split from the contour of the first layer structure. The method can also comprise determining a difference between the alignment position of the first layer structure and the alignment position of the second layer structure.BRIEF DESCRIPTION OF FIGURES

[0011] The above and other aspects of the present disclosure will become more apparent from the description of exemplary embodiments, taken in conjunction with the accompanying drawings.

[0012] FIG. 1 is a schematic diagram illustrating an exemplary charged-particle beam inspection system, consistent with embodiments of the present disclosure.

[0013] FIG. 2 is a schematic diagram illustrating an exemplary single -beam tool, consistent with embodiments of the present disclosure that can be a part of the example charged particle beam inspection system of FIG. 1.

[0014] FIG. 3 is a schematic diagram illustrating an exemplary overlay shift, consistent with embodiments of the present disclosure.

[0015] FIGS. 4A and 4B are schematic diagrams illustrating exemplary die-to-database alignments, consistent with embodiments of the present disclosure.

[0016] FIG. 5 is a schematic diagram illustrating an exemplary inverse die-to-database alignment, consistent with embodiments of the present disclosure.

[0017] FIG. 6 is a schematic diagram illustrating a layer split process, consistent with embodiments of the present disclosure.

[0018] FIG. 7 is a flowchart of an exemplary process flow for performing an inverse die-to-database alignment and calculating overlay shift, consistent with embodiments of the present disclosure.

[0019] FIG. 8 is a flowchart of an exemplary method for performing an inverse die-to-database alignment and calculating overlay shift, consistent with embodiments of the present disclosure.DETAILED DESCRIPTION

[0020] Reference will now be made in detail to exemplary embodiments, examples of which are illustrated in the drawings. The following description refers to the accompanying drawings in which the same numbers in different drawings represent the same or similar elements unless otherwise represented. The implementations set forth in the following description of exemplary embodiments do not represent all implementations consistent with the invention. Instead, they are merely examples of apparatuses, systems, and methods consistent with aspects related to subject matter that may be recited in the appended claims. For example, although some embodiments are described in the context of utilizing charged-particle beams (e.g., electron beams), the disclosure is not so limited. Other types of charged particle beams (e.g., proton beams) may be similarly applied. Furthermore, other imaging systems may be used, such as optical imaging, photodetection, x-ray detection, or the like.

[0021] Electronic devices are constructed of circuits formed on a piece of silicon called a substrate. Many circuits may be formed together on the same piece of silicon and are called integrated circuits or ICs. With advancements in technology, the size of these circuits has decreased dramatically so that many more of them can fit on the substrate. For example, an IC chip in a smart phone can be as small as a fingernail and yet may include over 2 billion transistors, the size of each transistor being less than 1 / 1, 000th the width of a human hair.

[0022] Making these ICs with extremely small structures or components is a complex, timeconsuming, and expensive process, often involving hundreds of individual steps. Errors in even one step have the potential to result in defects in the finished IC, rendering it useless. Thus, one goal of the manufacturing process is to avoid such defects to maximize the number of functional ICs made in the process, that is, to improve the overall yield of the process.

[0023] Yield is an important metric that characterizes failure rate in device fabrication, which relates to cost and efficiency. Yield can be defined as a ratio of all the wafers that are produced by a fab to the number of wafers that were introduced to the fab. Or yield can be the number of working chips that survive the device fabrication process performed on a wafer to the number of potential chips that can be fabricated from that wafer in the ideal case of zero failure. As some wafers or chips fail during fabrication, the overall yield is less than 100%. For example, to obtain a 75% yield for a 50-step process (where a step can be indicative of the number of layers formed on a wafer), each individual step should have a yield greater than 99.4%. In contrast, if individual steps have a yield of 95%, the compounding errors at each step result in an overall process yield as low as 7-8%. Every wafer or chip lost during fabrication is a sunk cost and lost time for the fab.

[0024] Integral to the making of these ICs with extremely small structures are highly accurate inspection processes, performed in between one or more fabrication steps, to ascertain whether fabrication steps are performing at expected tolerances. Errors in even one step have the potential to result in defects in the finished IC, rendering it useless. A goal of the manufacturing process is to avoid such defects to maximize the number / yield of functional ICs made in the process.

[0025] Inspection can be carried out using a scanning charged-particle microscope (e.g., a scanning electron microscope (SEM)). A scanning charged-particle microscope can be used to image extremely small structures of ICs, by capturing an image of the structures on the wafer. The image can be used to determine if the structure was formed properly (e.g., having the expected dimensions and being properly located on the wafer). If the structure is defective, then the process can be adjusted, so the defect is less likely to recur.

[0026] The working principle of a SEM is analogous to that of a camera. A camera takes a picture by receiving and recording intensity of light reflected or emitted from people or objects. SEMs capture images by receiving and recording energies or quantities of electrons reflected or emitted from the structures of the wafer. Similar to how a camera uses a light source (e.g., ambient, sunlight, or a flash), SEMs use an electron source to send a beam(s) of electrons to a surface of a wafer that has structures of interest for imaging. The electron beam(s) can be deflected and the wafer can be moved (e.g., on a movement stage) so that a plurality of regions of the wafer can be irradiated by the electrons. When the electrons are reflected or emitted (“exiting”) from the structures (e.g., from the wafer surface, from the structures underneath the wafer surface, or both), a detector of the SEM can receive and record the energies or quantities of those electrons to generate an inspection image of the regions of the wafer that were irradiated by electrons.

[0027] Some SEMs use a single electron beam (referred to as a “single -beam SEM”) to capture a single image. Some SEMs use multiple electron beams (referred to as a “multibeam SEM”) to capture multiple sub-images of the wafer in parallel. The multiple sub-images can be viewed separately or stitched together to generate a complete image. Multibeam parallelization is a technique that increases the number of electrons being received at a detector, thereby increasing efficiency and speed of measurement.

[0028] Speed, or throughput, is also an important metric alongside yield. Throughput is a measurable quantity that characterizes the manufacture speed of a fab (e.g., number of IC units produced per unit time). While throughput is a traditionally important metric, it has become even more regarded in view of recent global chip shortages. As there are multiple steps in the fabrication of a chip device (e.g., for multiple layers), each step can have a characteristic throughput. For an inspection operation among the fabrication steps, throughput can characterize how quickly an inspection process can clear a wafer before moving on to the next wafer. Innovations in the design or functions of inspection tools can increase throughput, or at least resolve problems in another aspect while mitigating adverse impact to throughput.

[0029] But conventional inspection tools have their limitations and temperaments / unique variations, and no one measurement technique is without flaw. For example, examination of an SEM image can be performed by a computer algorithm, as opposed to a manual examination by a person. The algorithm can be programmed to search for certain structural patterns that were fabricated on a wafer, the structural patterns being based on a design pattern (e.g., provided in a computer-aided design (CAD) file). If the quality of the image or fabrication process are below tolerances, or if the overlayed structured layers are numerous and complex, a consequence can be that the algorithm spends an inordinate amount of time searching the image for a given structure (e.g., stuck in a loop), which adversely impacts throughput. Yield can also be affected if the algorithm fails to find the expected structures. An example process that is susceptible to this type of drawback is a die-to-database (D2DB) alignment, in which the real structure (on the die) is aligned to a known design pattern (in a database, such as a CAD file).

[0030] Embodiments of the present disclosure provide a system and method to perform an inverse D2DB alignment to mitigate issues that afflict D2DB alignments. For example, a D2DB alignment can fail to align the die structures to the database pattern if the die structures are too complex (e.g., having multiple structures of different layers overlapped). In contrast, the pattern search portion of an inverse D2DB alignment can be made insensitive to the presence of multiple layers in the SEM image, thereby preventing lengthy and potentially inaccurate search processes. In this manner, inspection tools, can be enhanced to perform faster and more accurate inspections.

[0031] Objects and advantages of the disclosure can be realized by the elements and combinations as set forth in embodiments described herein. However, embodiments of the present disclosure are not necessarily required to achieve such exemplary objects or advantages. Some embodiments can allowimplementation of a features or function without necessarily achieving any expressly stated object or advantage.

[0032] Without limiting the scope of the present disclosure, some embodiments are described in the context of providing detection systems and detection methods in systems that use electron beams (“e- beams”). However, the disclosure is not so limited. Other types of charged particle beams can be similarly applied. Furthermore, systems and methods for detection can be used in other imaging systems, such as optical imaging, photon detection, x-ray detection, ion detection, or the like.

[0033] As used herein, unless specifically stated otherwise, the term “or” encompasses all possible combinations, except where infeasible. For example, if it is stated that a component can comprise A or B, then, unless specifically stated otherwise or infeasible, the component can comprise A, or B, or A and B. As a second example, if it is stated that a component can comprise A, B, or C, then, unless specifically stated otherwise or infeasible, the component can comprise A, or B, or C, or A and B, or A and C, or B and C, or A and B and C.

[0034] Relative dimensions of components in drawings may be exaggerated for clarity. Within the following description of drawings, the same or like reference numbers refer to the same or like components or entities, and only the differences with respect to the individual embodiments are described.

[0035] FIG. 1 illustrates a schematic diagram of an exemplary electron beam inspection (EBI) system 100, consistent with embodiments of the present disclosure. EBI system 100 can be used for imaging. EBI system 100 can comprise a main chamber 101, a load / lock chamber 102, a beam tool 104, and an equipment front end module (EFEM) 106. Beam tool 104 can be located within main chamber 101. EFEM 106 can comprise a first loading port 106a and a second loading port 106b. EFEM 106 can comprise additional loading port(s). First loading port 106a and second loading port 106b can receive wafer front opening unified pods (FOUPs) that contain wafers (e.g., semiconductor wafers or wafers made of other material(s)) or samples to be inspected (wafers and samples may be used interchangeably). The term “lot” can refer to a plurality of wafers that can be loaded for processing as a batch.

[0036] One or more robotic arms (not shown) in EFEM 106 can transport the wafers to load / lock chamber 102. Load / lock chamber 102 can be connected to a load / lock vacuum pump system (not shown) that can evacuate gas molecules in load / lock chamber 102 to reach a first pressure below atmospheric pressure. After reaching the first pressure, one or more robotic arms (not shown) can transport the wafer from load / lock chamber 102 to main chamber 101. Main chamber 101 can be connected to a main chamber vacuum pump system (not shown) that can evacuate gas molecules in main chamber 101 to reach a second pressure below the first pressure. After reaching the second pressure, the wafer can be subject to inspection by beam tool 104. Beam tool 104 can be a singlebeam system or a multibeam system.

[0037] A controller 109 can be electronically connected to beam tool 104. Controller 109 can be a computer configured to execute various controls of EBI system 100. While controller 109 is shown in FIG. 1 as being outside of the structure that includes main chamber 101, load / lock chamber 102, and EFEM 106, in some embodiments, controller 109 can be part of the structure.

[0038] In some embodiments, controller 109 comprises one or more processors (not shown). A processor may be a generic or specific electronic device capable of manipulating or processing information. For example, the processor can comprise a central processing unit (or “CPU”), a graphics processing unit (or “GPU”), an optical processor, a programmable logic controllers, a microcontroller, a microprocessor, a digital signal processor, an intellectual property (IP) core, a Programmable Logic Array (PL A), a Programmable Array Logic (PAL), a Generic Array Logic (GAL), a Complex Programmable Logic Device (CPLD), a Field-Programmable Gate Array (FPGA), a System On Chip (SoC), an Application-Specific Integrated Circuit (ASIC), any type circuitry capable of data processing, or any combination of any number thereof. The processor can be a virtual processor. The virtual processor can include one or more processors distributed across multiple machines or devices coupled via a network.

[0039] In some embodiments, controller 109 further comprises one or more memories (not shown). A memory can be a generic or specific electronic device capable of storing instructions, code, or data accessible by the processor (e.g., via a bus). For example, the memory can comprise a random-access memory (RAM), a read-only memory (ROM), an optical disc, a magnetic disk, a hard drive, a solid- state drive, a flash drive, a security digital (SD) card, a memory stick, a compact flash (CF) card, any type of storage device, or any combination of any number thereof. The instructions, code, or data can include an operating system (OS) and one or more application programs (or “apps”) for specific tasks. The memory can be a virtual memory. The virtual memory can include one or more memories distributed across multiple machines or devices coupled via a network.

[0040] FIG. 2 illustrates a schematic diagram of an exemplary beam tool 104 and an image processing system 199 that can be configured for use with EBI system 100 (FIG. 1), consistent with embodiments of the present disclosure. In some embodiments, beam tool 104 is a single-beam tool that uses one primary electron beam to scan locations on a wafer in series (one location after the other).

[0041] Beam tool 104 can comprise a wafer holder 136 supported by motorized stage 134 to hold a wafer 150 to be inspected. Beam tool 104 can also comprise an electron emitter comprising several elements, such as a cathode 103, an anode 121, and a gun aperture 122. Beam tool 104 can also comprise a beam limit aperture 125, a condenser lens 126, a column aperture 135, an objective lens assembly 132, and a detector 144. Objective lens assembly 132 can be a swing objective retarding immersion lens (SORIL) or a modified version thereof. Objective lens assembly 132 can comprise a pole piece 132a, a control electrode 132b, a deflector unit 132c, and an exciting coil 132d. In a detection or imaging process, an electron beam 161 emanating from the tip of cathode 103 can beaccelerated by anode 121 voltage, can pass through gun aperture 122, beam limit aperture 125, condenser lens 126, and can be focused into a probe spot 170 by objective assembly 132. Electron beam 161 can impinge onto the surface of wafer 150. A deflector can be used to scan probe spot 170 across the surface of wafer 150 (e.g., deflector unit 132c or other deflector(s) in the SORIL lens). Secondary or scattered particles, such as secondary electrons or scattered primary electrons emanated from the wafer surface can be collected by detector 144 to determine intensity of the beam and so that an image of an area of interest on wafer 150 can be reconstructed.

[0042] In some embodiments, image processing system 199 comprises an image acquirer 120, a storage 130, and controller 109. Image acquirer 120 can comprise one or more processors. For example, image acquirer 120 can comprise a computer, server, mainframe host, terminals, personal computer, any kind of mobile computing devices, or the like. Image acquirer 120 can be connected with detector 144 of beam tool 104 through a communication medium, such as an electrical conductor, optical fiber cable, portable storage media, IR, Bluetooth, internet, wireless network, wireless radio, or the like. Image acquirer 120 can receive a signal from detector 144. Image acquirer 120 can construct an image based on the signal from detector 144. Image acquirer 120 can thus acquire images of wafer 150. Image acquirer 120 can also perform various post-processing functions, such as image averaging, generating contours, superimposing indicators on an acquired image, or the like. Image acquirer 120 can perform adjustments of brightness and contrast, etc. of acquired images. Storage 130 can be a storage medium, such as a hard disk, random access memory (RAM), cloud storage, other types of computer readable memory, or the like. Storage 130 can be coupled with image acquirer 120 and can be used for saving scanned raw image data as original images, as well as postprocessed images. Image acquirer 120 and storage 130 can be connected to controller 109. In some embodiments, image acquirer 120, storage 130, and controller 109 can be integrated together as one electronic control unit (e.g., on the same chip).

[0043] In some embodiments, image acquirer 120 acquires one or more images of a sample based on an imaging signal received from detector 144. An imaging signal can correspond to a scanning operation that is executed for the purposes of charged particle imaging. An acquired image can be a single image comprising a plurality of imaging areas that can contain various features of wafer 150. The single image can be stored in storage 130. Imaging can be performed on the basis of imaging frames.

[0044] The condenser and illumination optics of the electron beam tool can comprise, or be supplemented by, electromagnetic quadrupole electron lenses. In the example of FIG. 2, electron beam tool 104 comprises a first quadrupole lens 148 and a second quadrupole lens 158. The quadrupole lenses can be used for controlling the electron beam. First quadrupole lens 148 can be controlled to adjust the beam current and second quadrupole lens 158 can be controlled to adjust the beam spot size and beam shape.

[0045] FIG. 2 shows an example of a charged particle beam apparatus that uses a single primary beam configured to generate secondary electrons by interacting with wafer 150. Detector 144 can be placed proximal to or aligned with optical axis 105. The primary electron beam can be configured to travel along optical axis 105. Accordingly, detector 144 can comprise a hole at its center so that the primary electron beam can pass through to reach wafer 150. In some embodiments, a detector can be placed off-axis relative to the optical axis along which the primary electron beam travels. In such instances a beam separator can be provided to divert secondary electron beams toward a detector placed off-axis.

[0046] The images generated by scanning charged-particle microscope (e.g., an SEM) can be used for defect inspection. A generated image capturing a test device region of a wafer can be compared with a reference image of the same test device region. The reference image can be predetermined (e.g., by simulation) and include no known defect. If a difference between the generated image and the reference image exceeds a tolerance level, a potential defect can be identified. For another example, an SEM is used to scan multiple regions of a wafer, each region including a test device region designed to be constant from region to region. The SEM can generate multiple images capturing those test device regions as fabricated. The multiple images can be compared with each other. If a difference between the multiple images exceeds a tolerance level, a potential defect can be identified.

[0047] FIG. 3 illustrates an image comparison for an exemplary overlay shift 300, consistent with embodiments of the present disclosure. In some embodiments, overlay shift 300 is analyzed with reference to structures identified in two different images — structures resolved via a scanning charged- particle microscope (e.g., an SEM) and corresponding structures from a database (e.g., a GDS or CAD file). For brevity, electron beam systems, such as SEMs, will be used for describing embodiments. However, it is to be appreciated that the scope of the present disclosure also includes non-electron variants in the same vein as the electron-based embodiments (e.g., optical imaging, photon detection, x-ray detection, ion detection, or the like). An SEM image 302 and a design pattern 304 (e.g., a GDS or CAD layout) stored in a database will be used to describe overlay shift 300.

[0048] Design pattern 304 can comprise a layout for a target feature 306. Design pattern 304 can comprise a first layer design pattern 308 and a second layer design pattern 310, both of which are part of target feature 306. To facilitate description, dimensions in arbitrary units are provided via coordinate positions (0,0), (2,0), (0,2), and (2,2) as a non-limiting example. In this example, target feature 306 spans a 2x2 area. Furthermore, while some embodiments disclosed herein are described using two design layers, it is to be appreciated that embodiments described herein can be implemented with any suitable number of layers (e.g., two or more layers). The enhancements brought about by practicing embodiments described herein can increase as the number of layers are increased.

[0049] When design pattern 304 is used for fabricating device structures on a wafer 312, structures of first layer design pattern 308 can be fabricated on wafer 312 via a first fabrication step, which canthen be followed by fabrication of structures of second layer design pattern 310 via a second fabrication step. The second layer can be applied over the first layer, hence the term overlay. As explained above, non-conformities in fabrication processes can cause the fabricated structures of different layers to become misaligned. This type of misalignment is referred to as overlay error or overlay shift. Overlay shifts can cause critical structures to become inoperable (e.g., a shift to a transistor gate causes an electrical short, an electrical contact is missed resulting in an open circuit, or the like). In other words, defects that reduce yield can form.

[0050] In the context of metrology, the term “sample” can be used to refer to an object to be inspected or measured (e.g., a wafer or a region of a wafer can be a sample). Hence, an inspected surface can be referred to as a sample surface.

[0051] To identify defects, inspection can be performed to generate images such as SEM image 302 for measuring overlay shifts, as well as other types of defects. Furthermore, aberrations in lithographic processes need not be wafer-wide (e.g., can be localized). Hence, different localized regions of a wafer can be subject to different amounts of overlay shift. Two regions of interest are emphasized as non-limiting examples. SEM image 302 can be taken from in-device region 314 or out-of-device region 316. In-device regions can refer to regions of the wafer where device structures are fabricated (e.g., CPU transistors). A die (or a field) 318 can represent a complete device (e.g., one complete CPU). One large wafer is capable of hosting hundreds or thousands of devices, if not more. In contrast, out-of-device regions can refer to regions of the wafer that do not receive device structures. A scribe lane can be an out-of-device region. Scribe lanes are regions that are scribed during the final steps of fabrication to cut apart the multiple devices on the wafer into individual devices.

[0052] For purposes of monitoring the accuracy of fabrication processes, different inspection tools can rely on metrology targets (such as target feature 306) being placed throughout the wafer. Some inspection tools are optimized for out-of-device inspection and some are optimized for in-device inspection. A desirable aspect of in-device target placement is the ability to accurately measure overlay shifts closer to critical structures. Overlay shifts can vary from point to point on a wafer, and in-device inspection can help in monitoring regional variations. A drawback is that the limited wafer real estate in die 318 becomes occupied by metrology targets instead of a device structure (e.g., a CPU would have fewer transistors owing to the space sacrificed for the metrology targets). Another complication is that an in-device metrology target can be surrounded in a sea of device structures, thereby having the potential to confuse search algorithms that are programmed to identify the metrology target in an image. In contrast, out-of-device metrology targets often find themselves further away from the device structures, potentially yielding less accurate information about the overlay inside die 318, but allow for maximizing space usage in die 318.

[0053] SEM image 302 can represent a region of wafer 312 that has been inspected by an SEM. SEM image 302 can comprise image representations of a first structure 322 and a second structure 324 on wafer 312. The negative space (e.g., the featureless regions) can correspond to bare regions of thewafer 312. When presented with SEM image 302 for the first time, a search / identification algorithm is initially uninformed as to the significance of a first structure 322 and a second structure 324 (e.g., it is not yet assumed that the structures should be associated with a target feature 320, nor is there yet an assumption that the structures belong in different layers). However, after the search algorithm is able to identify the structures, associations can be made. For example, first structure 322 is referred to as first layer structure 308' based on being associated with first layer design pattern 308. Similarly, second structure 324 can be referred to as second structure pattern 310' based on being associated with second layer design pattern 310. Target feature 320 can be referred to as target feature 306' based on first layer structure 308' and second layer structure 310'.

[0054] In some embodiments, an image processor compares SEM image 302 and design pattern 304 to compute an overlay shift. The image processor can be a controller, a computing device, or the like. The computation can involve aligning the structures in the SEM image to the patterns in the design layout. When first layer structure 308' is aligned with first layer design pattern 308, the image processor can determine that an overlay shift of (0,-1) has occurred based on the known database position of second layer design pattern 310 relative to first layer design pattern 308.

[0055] FIGS. 4A and 4B illustrate die-to-database (D2DB) alignments 400 and 400', consistent with embodiments of the present disclosure. In some embodiments, D2DB alignments 400 and 400' are each performed using an image from a scanning charged-particle microscope (e.g., an SEM) and a design pattern in a database (e.g., a CAD file).

[0056] In some embodiments, an SEM image 402 is acquired using an SEM. SEM image 402, in its raw form, comprises contrasts in intensity that delineate structures 404, 406, 408, 410, and 412. An image processor can analyze SEM image 402 to extract a contour map 418 of structures 404, 406, 408, 410, and 412. Contour extraction can also be referred to as contour drawing, defining contours, or the like. A contour can be an outline representing or bounding the shape or form of a structure. The image processor can identify the boundaries and breadth of structures 404, 406, 408, 410, and 412 based on abrupt changes in signal intensity (contrast) in SEM image 402.

[0057] The image processor can be cognizant that a design pattern 420 has been used for the fabrication of structures 404, 406, 408, 410, and 412 (e.g., image processor can receive, or be directed to, design pattern 420 in / from a database). Using contour map 418, or even just SEM image 402 in raw form, the image processor can search for the likeness of structures 404, 406, 408, 410, and 412 and relative positions thereof within design pattern 420. If a match is found, the image processor can then align the identified structures 404, 406, 408, 410, and 412 from SEM image 402 to the corresponding patterns in design pattern 420 to establish a relative position of structures 404, 406, 408, 410, and 412. This process of aligning structures on a die (e.g., as measured by SEM) to a design pattern in a database can be referred to as a D2DB alignment.

[0058] It is to be appreciated that design pattern 420 can be a modified design pattern based on a CAD file that takes into account effects of lithographic fabrication. The modified design pattern canbe generated by simulation or controlled experimentation of real lithographic processes under ideal conditions. Therefore, the modified design pattern can provide a more accurate reference that includes unavoidable deviations from the original design pattern (e.g., smoothing of sharp corners, structure widening, or the like).

[0059] An ideal scenario is that the image processor is able to perform D2DB alignments without error or delay. However, complexity of pattern layering can create issues with D2DB alignments. D2DB alignment 400' in FIG. 4B shows an example that can cause the D2DB alignment process to falter.

[0060] In some embodiments, an SEM image 402' is acquired. SEM image 402' can also comprise contrasts in intensity that delineate structures 404, 406, 408, 410, and 412 that were also present in SEM image 402 in the previously described example. However, SEM image 402' has additional contrasts in intensity that delineate structures 414 and 416. In a state of ignorance, the image processor that is to analyze SEM image 402' has no information about the layering scheme of the imaged structures. Nonetheless, the image processor can still analyze / process SEM image 402' to extract a contour map 418' of structures 404, 406, 408, 410, 412, 414, and 416.

[0061] Similar to FIG. 4A, the image processor can be programmed to for the likeness of structures 404, 406, 408, 410, 412, 414, and 416 and relative positions thereof within design pattern 420. However, design pattern 420 can be a first layer design pattern in which information about structures 414 and 416 are not present. A second layer design pattern (not shown) can include information about structures 414 and 416. The image processor, having no ability to ascertain which intensity blobs belong to an upper layer or a lower layer, can fail to identify the likeness of structures 404, 406, 408, 410, and 412 on account of the presence of the additional structures 414 and 416. This can be because the image processor is programmed to search for those structures present in SEM image 402', which would include searching for structures 414 and 416 in design pattern 420 to no avail. This error can lead to long delays (e.g., stuck in a search loop) or even misidentification and misalignment of the measured structures with respect to the database pattern. Therefore, to enhance D2DB alignments, an inverse D2DB alignment scheme is provided. And while such errors might be mitigated by userintervention (e.g., prompting a user to select which structures to search for), such workarounds may not be conducive for automation and higher throughput.

[0062] FIG. 5 illustrates an inverse D2DB (iD2DB) alignment 500, consistent with embodiments of the present disclosure. In some embodiments, an SEM image 502 is acquired by an SEM. It is to be appreciated that images as described herein need not be limited to the graphical representations in the drawings (e.g., images shown on a computer display). An image can be in the form of an analog signal (e.g., detector signals from an SEM), can be stored as a digital file or in volatile memory, or the like. When describing image analysis, image manipulation, and other image processes, it is to be appreciated that such processes can be executed on any form or representation of the image. Forexample, cropping an image can correspond to truncating the corresponding pixels in a digital image file.

[0063] SEM image 502 also comprises contrasts in intensity that delineate structures 504, 506, 508, 510, and 512, as well as structures 514 and 516. An image processor can analyze SEM image 502 to determine presence and shapes of structures. Optionally, image processor can extract a contour map 518 of the determined structures. In FIGS. 4A and 4B, D2DB alignments used an SEM image as an “unknown” while a database design pattern was used as a known reference, and the image processor would attempt to determine and align the unknown SEM image structures within the search field defined by the known reference that is the design pattern. The iD2DB process illustrated in FIG. 5 inverts the process, hence the term inverse D2DB.

[0064] In some embodiments, the image processor uses design pattern 520 (e.g., database pattern) as an unknown while using SEM image 502 or contour map 518 as the reference and search field. The iD2DB alignment method can be particularly potent when determining layer-specific alignment, as well as inspecting targets disposed inside a device (e.g., in-device region 314 (FIG. 3)). Design pattern 520 can be layer specific so as to include only those shapes for a specific layer (e.g., a first layer design pattern). Notably missing in design pattern 520 are shapes that would correspond to structures 514 and 516 in SEM image 502 or contour map 518. Nonetheless, the image processor can still execute and succeed in the pattern search and subsequent alignment since the search is concerned with finding shapes that correspond to structures 504, 506, 508, 510, and 512 while disregarding extraneous structures, such as structures 514 and 516. Furthermore, the presence of actual device features (e.g., in-device region 314 (FIG. 3)) do not impose a burden on the search since nonmatching structures are disregarded, whereas a standard D2DB alignment would be burdened by including the device structures as objects to be found in the search. The image processor can determine that structures 514 and 516 have no likeness to any of the shapes in design pattern 520 and can therefore be disregarded. In this manner, iD2DB alignment can mitigate the error scenarios described above in reference to D2DB alignments while promoting automation and higher throughput.

[0065] FIG. 6 illustrates a before and after of a layer split process 600, consistent with embodiments of the present disclosure. FIG. 6 will be described in contour space, but it should be appreciated that layer splitting embodiments can also use raw image data (e.g., intensity contrast information) to execute a layer split. A layer splitting process can be executed after an iD2DB alignment (or even a D2DB alignment) has identified structures of a given device layer in an image and the image also includes unidentified extraneous structures. An example of such an image is an aligned contour map 602.

[0066] In some embodiments, aligned contour map 602 can include aligned structures 604, which are represented by crosshatch filling. Aligned structures are designated as “aligned” by virtue of having been identified and positioned via an iD2DB / D2DB alignment. In other words, an image processor can be made cognizant that structures 604 correspond to a given layer of a design pattern. Alsopresent in aligned contour map 602 are remainder structures 606, which are represented by white filling. Remainder structures 606 can be those structures that were found to not match a shape in the given layer during the iD2DB / D2DB alignment.

[0067] For the present example, the given layer of structures 604 will be referred to as a first layer, which can be one of a top layer or a bottom layer. A second layer can be the other one of the top layer or bottom layer. Layer splitting can comprise defining regions in aligned contour map 602 that are to be removed, occluded, or otherwise disregarded in a future iD2DB / D2DB alignment performed for the second layer. The defined regions are shown as regions 608, which are represented by dotted lines.

[0068] The boundaries of regions 608 can conform to a tolerance determined from statistical analysis of distances between contour points in the measured image (e.g., SEM image) and contour points from its counterpart database pattern (e.g., a design pattern). The image processor can determine whether any structures found in aligned contour map 602 are disposed at distances that exceed the determined tolerances. The image processor can determine remainder structures 606 exceed the tolerance.

[0069] In some embodiments, the image processor can determine image information that removes, or otherwise flags for disregard, structures associated with the first layer. Split contour map 610 illustrates an example of image information that retains remainder structures 606 while occluding structures 604 of the first layer. In other words, the first layer has been split off or isolated from the occluded first layer, which results in split contour map 610. A layer split can also be referred to as a contour split when operating in contour space.

[0070] With the split layer information now available, a subsequent iD2DB / D2DB alignment can be performed on split contour map 610. The subsequent iD2DB / D2DB alignment can determine the positions of structures in the second layer, thereby allowing for a comparison of overlay and subsequent determination of overlay error.

[0071] FIG. 7 illustrates a process flow 700 for determining an overlay shift, consistent with embodiments of the present disclosure. The process flow can be executed using devices and functions described in reference to FIGS. 1-6, such as controller 109 of FIG. 1.

[0072] In some embodiments, at step 702, an image of a sample surface is acquired. The image can be an SEM image (e.g., SEM image 302 of FIG. 3, SEM image 402 of FIG. 4A, SEM image 402' of FIG. 4B, SEM image 502 of FIG. 5). The SEM image can be acquired in real time via a measurement using an SEM or post-measurement by accessing an image file. The SEM image can comprise information of structures of a target feature on the sample surface (e.g., target feature 306' and structures 322 or 324 of FIG. 3, structures 404, 406, 408, 410, 412, 414, or 416 of FIGS. 4A and 4B, structures 504, 506, 508, 510, 512, 514, or 516 of FIG. 5, structures 604 or 606 of FIG. 6). The target feature can be present on the sample surface (e.g., a wafer surface) as a result of lithographic fabrication processes. The fabrication processes rely on a design pattern (e.g., a GDS or CAD file in a database) to place layers of structures of the target feature according to the predeterminedarrangement as specified by the design pattern (e.g., design pattern 304 of FIG. 3, design pattern 420 of FIGS. 4A and 4B, design pattern 520 of FIG. 5). The design pattern can be a modified design pattern that takes into account the effects of lithographic processes. Hence, the target feature can comprise a first layer structure that corresponds to a first layer of a design pattern and a second layer structure that corresponds to the second layer of the design pattern.

[0073] At step 704, contours can be extracted from intensity information in the SEM image to generate a contour map (e.g., contour map 602 of FIG. 6). The extracted contours can correspond to the boundaries of the various structures in two or more layers of the target feature. The contour extraction of step 704 can be optional as downstream steps can be made to work with intensity contrast information to determine structures present in the SEM image.

[0074] At step 706, iD2DB alignment can be performed using the SEM image or the contour map (e.g., iD2DB alignment 500 of FIG. 5). Since it is an object of the iD2DB alignment to discriminate between layers of the design pattern, the iD2DB alignment can be executed by using at least a portion of just the first layer design pattern for the shapes to be searched for while the SEM image (or contour map) can be used as the search field. Even if the SEM image includes structures of multiple layers in the search field, the iD2DB alignment can disregard structures in the SEM that do not match with the first layer design pattern. In this manner, an alignment position of the first layer structure can be determined relative to the design pattern. The disregarded structures that did not have a match with the first layer design pattern can be considered as remainder structures (e.g., as belonging to a layer that is not the first layer).

[0075] At step 708, layer splitting can be performed by occluding or disregarding the first layer structures as identified in the iD2DB alignment so as to isolate the remainder structures (e.g., layer split process 600 of FIG. 6).

[0076] At step 710, iD2DB / D2DB alignment can be performed using the remainder structures (e.g., D2DB alignments 400 in FIG. 4A or iD2DB alignment 500 in FIG. 5). Since it is an object of the iD2DB / D2DB alignment to discriminate between layers of the design pattern, the iD2DB / D2DB alignment is executed by using at least a portion of just the second layer design pattern for the shapes to be searched for while an image with just the remainder structures is used as the search field. In this manner, an alignment position of the first layer structure can be determined relative to the design pattern.

[0077] While the alignment at step 706 is performed on a first layer and the alignment at step 710 is performed on a second layer, it should be appreciated that the process can be performed with the layers reversed (e.g., alignment at step 706 performed on the second layer and alignment at step 710 performed on the first layer).

[0078] In some embodiments, at step 712, a difference is calculated. The design pattern comprises information about the desired arrangement of the first layer design pattern and the second layer design pattern. In other words, the expected relative distances between patterns of different layers arepredetermined by the GDS or CAD information. The calculated difference is between the alignment position of the first layer structure and the alignment position of the second layer structure.

[0079] At step 714, the calculated difference can be output as an overlay shift (the error in overlaying the first and second layers). The alignment positions of the first layer design pattern and the second layer design pattern can be defined relative to a common frame (that is, the ideal alignment represented in the design pattern can be said to have zero overlay shift). Hence, a non-zero difference between the alignment position of the first layer structure and the alignment position of the second layer structure can be considered as overlay shift.

[0080] FIG. 8 illustrates a method 800 for determining performing iD2DB and obtaining an overlay shift, consistent with embodiments of the present disclosure. The method can be executed using devices and functions described in reference to FIGS. 1-7, such as controller 109 of FIG. 1. Method 800 represents an alternative description of the process flow in FIG. 7.

[0081] In some embodiments, at step 802, a shape of a target feature is determined (e.g., target feature 306' and structures 322 or 324 of FIG. 3, structures 404, 406, 408, 410, 412, 414, or 416 of FIGS. 4A and 4B, structures 504, 506, 508, 510, 512, 514, or 516 of FIG. 5, structures 604 or 606 of FIG. 6). The target feature can be disposed at a sample surface. The determination can be based on detection of charged particles scattered from the sample surface. A first layer structure and a second layer structure of the target feature are based on a design pattern comprising a first layer design pattern and a second layer design pattern (e.g., design pattern 304 of FIG. 3, design pattern 420 of FIGS. 4A and 4B, design pattern 520 of FIG. 5).

[0082] At step 804, an alignment position of the first layer structure is determined via an iD2DB alignment (e.g., iD2DB 500 of FIG. 5). The iD2DB alignment can comprise searching for at least a portion of the first layer design pattern in a search area defined by the determined shape of the target feature.

[0083] At step 806, an alignment position of the second layer structure is determined.

[0084] At step 808, a difference is determined. The difference is between the alignment position of the first layer structure and the alignment position of the second layer structure.

[0085] Some embodiments were described with reference to images that can be taken using the single beam tool in FIG. 2. However, embodiments of the present disclosure are not so limiting. Images acquired via other charged particle apparatuses (e.g., multibeam SEM) can also be used with embodiments described herein.

[0086] Some embodiments were described with reference to non-overlapping structures of different layers (e.g., as shown in FIGS 4A, 4B, 5, and 6). However, embodiment of the present disclosure are not so limiting. Embodiments described herein are also applicable to overlapping structures of different layers, which can be accounted for in iD2DB alignments and layer splitting operations.

[0087] A non-transitory computer-readable medium may be provided that stores instructions for a processor of a controller (e.g., controller 109 in FIG. 1) for performing iD2DB alignments anddetermining overlay shifts according to the exemplary flowcharts of FIGS. 7 and 8 above, consistent with embodiments in the present disclosure. For example, the instructions stored in the non-transitory computer-readable medium may be executed by the circuitry of the controller for performing process flow 700 and method 800 in part or entirely. Common forms of non-transitory media include, for example, a floppy disk, a flexible disk, hard disk, solid-state drive, magnetic tape, or any other magnetic data storage medium, a Compact Disc Read-Only Memory (CD-ROM), any other optical data storage medium, any physical medium with patterns of holes, a Random Access Memory (RAM), a Programmable Read-Only Memory (PROM), and Erasable Programmable Read-Only Memory (EPROM), a FLASH-EPROM or any other flash memory, Non-Volatile Random Access Memory (NVRAM), a cache, a register, any other memory chip or cartridge, and networked versions of the same.

[0088] Some embodiments may further be described using the following clauses:1. A charged particle beam apparatus, comprising: a charged particle beam source configured to generate a beam of primary charged particles; a charged particle optical system configured to direct the beam of primary charged particles at a sample surface; a charged particle detector configured to detect charged particles scattered from the sample surface; and a controller comprising circuitry configured to: determine a contour of a target feature disposed at the sample surface based on the detected charged particles, wherein a first layer structure and a second layer structure of the target feature are based on a design pattern comprising a first layer design pattern and a second layer design pattern; determine an alignment position of the first layer structure via an inverse die-to-database (iD2DB) alignment, wherein the iD2DB alignment comprises a search for at least a portion of the first layer design pattern in a search area defined by the determined contour of the target feature; determine an alignment position of the second layer structure; and determine a difference between the alignment position of the first layer structure and the alignment position of the second layer structure.2. The charged particle beam apparatus of clause 1, wherein: the controller comprises circuitry further configured to, prior to determining the alignment position of the second layer, occlude the first layer structure based on the determined alignment position of the first layer structure; and the alignment position of the second layer structure is determined based on the second layer structure being isolated from the occluded first layer structure.3. The charged particle beam apparatus of clause 2, wherein the controller comprises circuitry further configured to determine an occlusion boundary based on a deviation of a contour of the first layer structure from the first layer design pattern.4. The charged particle beam apparatus of clause 3, wherein the controller comprises circuitry further configured to retain the second layer structure as the first layer structure is occluded based on determining that the second layer structure is disposed outside of the occlusion boundary.5. The charged particle beam apparatus of any one of clauses 1 to 4, wherein the controller comprises circuitry further configured to acquire a scanning electron microscope image to determine the contour of the target feature.6. The charged particle beam apparatus of any one of clauses 1 to 5, wherein the controller comprises circuitry further configured to perform a die-to-database (D2DB) alignment to determine the alignment position of the second layer structure.7. The charged particle beam apparatus of any one of clauses 1 to 6, wherein the controller comprises circuitry further configured to compile training data comprising the determined difference to train a machine learning model to identify overlay shift at another sample surface.8. The charged particle beam apparatus of any one of clauses 1 to 7, wherein the controller comprises circuitry further configured to determine an in-device overlay shift based on the determined difference.9. A non-transitory computer-readable medium that stores a set of instructions that is executable by at least one processor of an apparatus to cause the apparatus to perform operations comprising: determining a contour of a target feature disposed at a sample surface based on an inspection of the sample surface, wherein a first layer structure and a second layer structure of the target feature are based on a design pattern comprising a first layer design pattern and a second layer design pattern; determining an alignment position of the first layer structure via an inverse die-to-database (iD2DB) alignment, comprising: searching for at least a portion of the first layer design pattern in a search area defined by the determined contour of the target feature; determining an alignment position of the second layer structure; and determining a difference between the alignment position of the first layer structure and the alignment position of the second layer structure.10. The non-transitory computer-readable medium of clause 9, wherein: the operations further comprise, prior to determining the alignment position of the second layer, occluding the first layer structure based on the determined alignment position of the first layer structure; and determining the alignment position of the second layer structure is based on the second layer structure being isolated from the occluded first layer structure.11. The non-transitory computer-readable medium of clause 10, wherein the operations further comprise determining an occlusion boundary based on a deviation of a contour of the first layer structure from the first layer design pattern.12. The non-transitory computer-readable medium of clause 11, wherein the operations further comprise retaining the second layer structure as the first layer structure is occluded based on determining that the second layer structure is disposed outside of the occlusion boundary.13. The non-transitory computer-readable medium of any one of clauses 9 to 12, wherein the operations further comprise acquiring a scanning electron microscope image to determine the contour of the target feature.14. The non-transitory computer-readable medium of any one of clauses 9 to 13, wherein determining the alignment position of the second layer structure comprises performing a die-to- database (D2DB) alignment.15. The non-transitory computer-readable medium of any one of clauses 9 to 14, wherein the operations further comprise compiling training data comprising the determined difference to train a machine learning model to identify overlay shift at another sample surface.16. The non-transitory computer-readable medium of any one of clauses 9 to 15, wherein the operations further comprise determining an in-device overlay shift based on the determined difference.17. A method comprising: inspecting a sample surface; determining a contour of a target feature disposed at the sample surface based on the inspecting, wherein a first layer structure and a second layer structure of the target feature are based on a design pattern comprising a first layer design pattern and a second layer design pattern; determining an alignment position of the first layer structure via an inverse die-to-database (iD2DB) alignment, comprising: searching for at least a portion of the first layer design pattern in a search area defined by the determined contour of the target feature; determining an alignment position of the second layer structure; and determining a difference between the alignment position of the first layer structure and the alignment position of the second layer structure.18. The method of clause 17, wherein: the method further comprises, prior to determining the alignment position of the second layer, occluding the first layer structure based on the determined alignment position of the first layer structure; and determining the alignment position of the second layer structure is based on the second layer structure being isolated from the occluded first layer structure.19. The method of clause 18, further comprising determining an occlusion boundary based on a deviation of a contour of the first layer structure from the first layer design pattern.20. The method of clause 19, further comprising retaining the second layer structure as the first layer structure is occluded based on determining that the second layer structure is disposed outside of the occlusion boundary.21. The method of any one of clauses 17 to 20, further comprising acquiring a scanning electron microscope image to determine the contour of the target feature.22. The method of any one of clauses 17 to 21, wherein determining the alignment position of the second layer structure comprises performing a die-to-database (D2DB) alignment.23. The method of any one of clauses 17 to 22, further comprising compiling training data comprising the determined difference to train a machine learning model to identify overlay shift at another sample surface.24. The method of any one of clauses 17 to 23, further comprising determining an in-device overlay shift based on the determined difference.25. A charged particle beam apparatus, comprising: a charged particle beam source configured to generate a beam of primary charged particles; a charged particle optical system configured to direct the beam of primary charged particles at a sample surface; a charged particle detector configured to detect charged particles scattered from the sample surface to generate an image of a target feature disposed at the sample surface; and a controller comprising circuitry configured to: extract a contour of the target feature from the image, wherein a first layer structure and a second layer structure of the target feature are based on a design pattern comprising a first layer design pattern and a second layer design pattern; determine an alignment position of the first layer structure via an inverse die-to-database (iD2DB) alignment, wherein the iD2DB alignment comprises a search for at least a portion of the first layer design pattern in a search area defined by the determined contour of the target feature; split a contour of the second layer structure from a contour the first layer structure based on the determined alignment position of the first layer structure; determine an alignment position of the second layer structure based on the contour of the second layer structure being split from the contour of the first layer structure; and determine a difference between the alignment position of the first layer structure and the alignment position of the second layer structure.26. The charged particle beam apparatus of clause 25, wherein the controller comprises circuitry further configured to occlude the first layer structure to split the contour of the second layer structure from the contour of the first layer structure.27. The charged particle beam apparatus of clause 26, wherein the controller comprises circuitry further configured to determine an occlusion boundary based on a deviation of the contour of the first layer structure from the first layer design pattern.28. The charged particle beam apparatus of clause 27, wherein the controller comprises circuitry further configured to retain the contour of the second layer structure as the contour of the first layer structure is occluded based on determining that the contour of the second layer structure is disposed outside of the occlusion boundary.29. The charged particle beam apparatus of any one of clauses 25 to 28, wherein the image is a scanning electron microscope image.30. The charged particle beam apparatus of any one of clauses 25 to 29, wherein the controller comprises circuitry further configured to perform a die-to-database (D2DB) alignment to determine the alignment position of the second layer structure.31. The charged particle beam apparatus of any one of cl clauses aims 25 to 30, wherein the controller comprises circuitry further configured to compile training data comprising the determined difference to train a machine learning model to identify overlay shift at another sample surface.32. The charged particle beam apparatus of any one of clauses 25 to 31 , wherein the controller comprises circuitry further configured to determine an in-device overlay shift based on the determined difference.33. A non-transitory computer-readable medium that stores a set of instructions that is executable by at least one processor of an apparatus to cause the apparatus to perform operations comprising: extracting, from an image, a contour of a target feature disposed at a sample surface, wherein the image is based on an inspection of the sample surface, and wherein a first layer structure and a second layer structure of the target feature are based on a design pattern comprising a first layer design pattern and a second layer design pattern; determining an alignment position of the first layer structure via an inverse die-to-database (iD2DB) alignment, comprising: searching for at least a portion of the first layer design pattern in a search area defined by the determined contour of the target feature; splitting a contour of the second layer structure from a contour of the first layer structure based on the determined alignment position of the first layer structure; determining an alignment position of the second layer structure based on the contour of the second layer structure being split from the contour of the first layer structure; and determining a difference between the alignment position of the first layer structure and the alignment position of the second layer structure.34. The non-transitory computer-readable medium of clause 33, wherein splitting the contour of the second layer structure from the contour of the first layer structure comprises occluding the first layer structure to split the second layer structure from the first layer structure.35. The non-transitory computer-readable medium of clause 34, wherein splitting the contour of the second layer structure from the contour of the first layer structure further comprises determiningan occlusion boundary based on a deviation of the contour of the first layer structure from the first layer design pattern.36. The non-transitory computer-readable medium of clause 35, wherein splitting the contour of the second layer structure from the contour of the first layer structure further comprises retaining the contour of the second layer structure as the contour of the first layer structure is occluded based on determining that the contour of the second layer structure is disposed outside of the occlusion boundary.37. The non-transitory computer-readable medium of any one of clauses 33 to 36, wherein the image is a scanning electron microscope image.38. The non-transitory computer-readable medium of any one of clauses 33 to 37, wherein determining the alignment position of the second layer structure comprises performing a die-to- database (D2DB) alignment.39. The non-transitory computer-readable medium of any one of clauses 33 to 38, wherein the operations further comprise compiling training data comprising the determined difference to train a machine learning model to identify overlay shift at another sample surface.40. The non-transitory computer-readable medium of any one of clauses 33 to 39, wherein the operations further comprise determining an in-device overlay shift based on the determined difference.41. A method comprising : extracting, from an image, a contour of a target feature disposed at a sample surface, wherein the image is based on an inspection of the sample surface, and wherein a first layer structure and a second layer structure of the target feature are based on a design pattern comprising a first layer design pattern and a second layer design pattern; determining an alignment position of the first layer structure via an inverse die-to-database (D2DB) alignment, comprising: searching for at least a portion of the first layer design pattern in a search area defined by the determined contour of the target feature; splitting a contour of the second layer structure from a contour of the first layer structure based on the determined alignment position of the first layer structure; determining an alignment position of the second layer structure based on the contour of the second layer structure being split from the contour of the first layer structure; and determining a difference between the alignment position of the first layer structure and the alignment position of the second layer structure.42. The method of clause 41, wherein splitting the contour of the second layer structure from the contour of the first layer structure comprises occluding the first layer structure to split the second layer structure from the first layer structure.43. The method of clause 42, wherein splitting the contour of the second layer structure from the contour of the first layer structure further comprises determining an occlusion boundary based on a deviation of the contour of the first layer structure from the first layer design pattern.44. The method of clause 43, wherein splitting the contour of the second layer structure from the contour of the first layer structure further comprises retaining the contour of the second layer structure as the contour of the first layer structure is occluded based on determining that the contour of the second layer structure is disposed outside of the occlusion boundary.45. The method of any one of clauses 41 to 44, wherein the image is a scanning electron microscope image. 46. The method of any one of clauses 41 to 45, wherein determining the alignment position of the second layer structure comprises performing a die-to-database (D2DB) alignment.47. The method of any one of clauses 41 to 46, further comprising compiling training data comprising the determined difference to train a machine learning model to identify overlay shift at another sample surface. 48. The method of any one of clauses 41 to 47, further comprising determining an in-device overlay shift based on the determined difference.

[0089] It will be appreciated that the embodiments of the present disclosure are not limited to the exact construction that has been described above and illustrated in the accompanying drawings and that various modifications and changes may be made without departing from the scope thereof.

Claims

CLAIMS1. A charged particle beam apparatus, comprising: a charged particle beam source configured to generate a beam of primary charged particles; a charged particle optical system configured to direct the beam of primary charged particles at a sample surface; a charged particle detector configured to detect charged particles scattered from the sample surface; and a controller comprising circuitry configured to: determine a contour of a target feature disposed at the sample surface based on the detected charged particles, wherein a first layer structure and a second layer structure of the target feature are based on a design pattern comprising a first layer design pattern and a second layer design pattern; determine an alignment position of the first layer structure via an inverse die-to- database (iD2DB) alignment, wherein the iD2DB alignment comprises a search for at least a portion of the first layer design pattern in a search area defined by the determined contour of the target feature; determine an alignment position of the second layer structure; and determine a difference between the alignment position of the first layer structure and the alignment position of the second layer structure.

2. The charged particle beam apparatus of claim 1, wherein: the controller comprises circuitry further configured to, prior to determining the alignment position of the second layer, occlude the first layer structure based on the determined alignment position of the first layer structure; and the alignment position of the second layer structure is determined based on the second layer structure being isolated from the occluded first layer structure.

3. The charged particle beam apparatus of claim 2, wherein the controller comprises circuitry further configured to determine an occlusion boundary based on a deviation of a contour of the first layer structure from the first layer design pattern.

4. The charged particle beam apparatus of claim 3, wherein the controller comprises circuitry further configured to retain the second layer structure as the first layer structure is occluded based on determining that the second layer structure is disposed outside of the occlusion boundary.

5. The charged particle beam apparatus of claim 1, wherein the controller comprises circuitry further configured to acquire a scanning electron microscope image to determine the contour of the target feature.

6. The charged particle beam apparatus of claim 1 , wherein the controller comprises circuitry further configured to perform a die-to-database (D2DB) alignment to determine the alignment position of the second layer structure.

7. The charged particle beam apparatus of claim 1, wherein the controller comprises circuitry further configured to compile training data comprising the determined difference to train a machine learning model to identify overlay shift at another sample surface.

8. The charged particle beam apparatus of claim 1, wherein the controller comprises circuitry further configured to determine an in-device overlay shift based on the determined difference.

9. A non-transitory computer-readable medium that stores a set of instructions that is executable by at least one processor of an apparatus to cause the apparatus to perform operations comprising: determining a contour of a target feature disposed at a sample surface based on an inspection of the sample surface, wherein a first layer structure and a second layer structure of the target feature are based on a design pattern comprising a first layer design pattern and a second layer design pattern; determining an alignment position of the first layer structure via an inverse die-to-database (iD2DB) alignment, comprising: searching for at least a portion of the first layer design pattern in a search area defined by the determined contour of the target feature; determining an alignment position of the second layer structure; and determining a difference between the alignment position of the first layer structure and the alignment position of the second layer structure.

10. The non-transitory computer-readable medium of claim 9, wherein: the operations further comprise, prior to determining the alignment position of the second layer, occluding the first layer structure based on the determined alignment position of the first layer structure; and determining the alignment position of the second layer structure is based on the second layer structure being isolated from the occluded first layer structure.

11. The non-transitory computer-readable medium of claim 10, wherein the operations further comprise determining an occlusion boundary based on a deviation of a contour of the first layer structure from the first layer design pattern.

12. The non-transitory computer-readable medium of claim 11, wherein the operations further comprise retaining the second layer structure as the first layer structure is occluded based on determining that the second layer structure is disposed outside of the occlusion boundary.

13. The non-transitory computer-readable medium of claim 9, wherein the operations further comprise acquiring a scanning electron microscope image to determine the contour of the target feature.

14. The non-transitory computer-readable medium of claim 9, wherein determining the alignment position of the second layer structure comprises performing a die-to-database (D2DB) alignment.

15. A method comprising: inspecting a sample surface; determining a contour of a target feature disposed at the sample surface based on the inspecting, wherein a first layer structure and a second layer structure of the target feature are based on a design pattern comprising a first layer design pattern and a second layer design pattern; determining an alignment position of the first layer structure via an inverse die-to-database (iD2DB) alignment, comprising: searching for at least a portion of the first layer design pattern in a search area defined by the determined contour of the target feature; determining an alignment position of the second layer structure; and determining a difference between the alignment position of the first layer structure and the alignment position of the second layer structure.

Images