Interface detection in reciprocal space

Abstract

The invention relates to interface detection in a reciprocal space. Methods and apparatus determine locations of boundaries of a given material in a sample based on classification of reciprocal spatial images at respective locations. The diffraction image at a given location is classified to identify material at that location. Imaging and classification at multiple locations can quickly and reliably find the boundary with submicron accuracy. Binary search provides acceleration. Classification is performed through a neural network. The technology is suitable for distinguishing monocrystalline silicon, polycrystalline silicon, amorphous silicon, high-Z materials (tungsten) or low-Z materials (carbon) and the like. Techniques are integrated into an automated workflow that precisely locates further imaging, probing or milling operations based on the localized boundaries. Examples and variations are disclosed.

Description

Background Art

[0001] In the semiconductor industry, interface detection is a valuable component of sample analysis. The object of interest is often located on or near the boundary between two materials. Examples include blanket films or semiconductor structures built on top of bulk silicon. Conventional techniques include manual inspection of survey images, or machine vision techniques for edge detection, for example. The former approach is naturally not suitable for workflow automation, while the latter techniques can be ad hoc and, in some cases, error-prone. First, an imaging modality is required in which the desired interface is visible as an edge or other discontinuity, which may require iterative research efforts as new materials or device types are introduced. Specifically, the analysis procedure may require the detection of multiple interfaces between different material pairs. Finding a universal imaging modality can be challenging for some samples. Then, machine vision techniques can find edges well, but may not be able to identify which edge is the desired edge. The heuristics for selecting the correct edge can vary from sample to sample, requiring a lot of configuration and testing efforts (usually trial and error). Therefore, there is still a need for a robust, universal, and automated technique to locate material boundaries in semiconductors and other samples. Summary of the invention

[0002] Briefly, examples of the disclosed technology use analysis of reciprocal space images to classify materials in corresponding imaged regions. By scanning over a series of imaged regions (e.g., in a stepwise fashion or by a binary search), the location of transitions (boundaries) from one material to another can be quickly and reliably determined with an accuracy of 100 nm or better.

[0003] The reciprocal space image can be obtained directly, for example, by a diffractometer, or indirectly, for example, by taking a physical space image and applying a transformation. Each reciprocal space image can be classified, for example, by a neural network, to identify the material present at the imaged region. The boundary location of a given material can be determined over multiple imaged regions based on the classification.

[0004] The disclosed technology may be implemented as an apparatus comprising: an imager configured to acquire a plurality of images of a corresponding region of a sample; and a controller. The controller may be configured to: process a reciprocal space representation of each image; and in response to the processing, classify a material within the corresponding region in a predetermined plurality of materials. The controller may be configured to determine a coordinate position of a boundary of a first material in the plurality of materials.

[0005] The above and other objects, features and advantages of the present invention will become more apparent from the following detailed description made with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0006] Figure 1is a diagram of an example device by which the disclosed techniques may be implemented.

[0007] Figure 2 is a collection of images illustrating the application of the disclosed technology.

[0008] Figure 3 is a diagram illustrating the use of binary search in an example of the disclosed technology.

[0009] Figure 4 is a flowchart depicting a first example method according to the disclosed technology.

[0010] FIG. 5A to FIG. 5B is a flow chart depicting an example extension of a method according to the disclosed technology.

[0011] Figure 6 is a diagram illustrating an example neural network that can be used to implement the disclosed techniques.

[0012] Figure 7 is a diagram illustrating an example of using multiple scans to locate material boundaries according to the disclosed technology.

[0013] Figure 8 A generalized example of a suitable computing environment in which the described embodiments, techniques, and technologies related to imaging or milling may be implemented is illustrated. DETAILED DESCRIPTION

[0014] Introduction

[0015] Automated analysis of semiconductor and other samples relies on the registration of the analysis tool to sample features, which are typically material boundaries or predetermined offsets from such boundaries. Conventional techniques can be ad hoc techniques for finding material boundaries. An imaging modality that is appropriate for one type of boundary may not be appropriate for another type of boundary. As samples and compositions vary, it may sometimes take a lot of trial and error to find one or more suitable imaging modalities to capture all material boundaries of interest. Machine vision techniques such as those used for edge detection may also be easily fooled on certain samples, rejecting desired edges in favor of artifacts. A machine vision algorithm that works effectively for one sample type may have a high failure rate when changes are made to the sample.

[0016] The disclosed techniques rely on reciprocal space images (such as diffraction patterns) that can directly provide material characterization without relying on a skilled operator or fragile machine vision heuristics. In some examples, tools such as convergent beam electron diffraction (CBED) can operate with spot sizes less than 10nm, allowing very accurate mapping of materials in a sample. Scanning multiple spots along a path can easily identify where a transition from one material to another occurs. The disclosed techniques can detect changes in CBED images that indicate changes in the underlying imaged material, such as a change from a CBED that is unique to single crystal silicon to a CBED that is not unique. Further, a binary search along the path can find transitions in logarithmic time, meaning that the time taken is linearly related to the logarithm of the required tolerance. Still further, multiple scans can be used to identify the orientation of a boundary, track boundaries with complex shapes, or identify multiple boundaries between corresponding material pairs within a single session.

[0017] the term

[0018] Unless explicitly stated otherwise or contradicted by context, the usage and meaning of all cited terms in this section apply to the entire disclosure.The following terms are extended to the relevant word forms.

[0019] "Analysis" refers to operations used to characterize a sample, and may include material removal, other sample preparation, imaging, probe measurements, non-contact measurements, or secondary evaluation of data obtained by any of these techniques. Analytical operations may include imaging (e.g., FIB, optical or electron microscopy), etching (e.g., ion milling), delay, electron backscatter analysis, electron microscopy, mass spectrometry, material analysis, metrology, nanoprobing, spectroscopy, or surface preparation. The equipment or instrument used to perform this operation is called an "analytical instrument," "analyzer," or "tool." Non-limiting examples of tools that can be used in conjunction with the disclosed technology include a milling machine, an etcher, an electron microscope, an electron spectrometer, an optical microscope, or an optical spectrometer. Specifically, some tools of interest herein include a plasma focused ion beam (PFIB), a scanning electron microscope (SEM), or a fluorescence microscope.

[0020] The term "beam" is a directed flow of particles or energy. Common beams of interest in the present disclosure are particle beams, such as electron beams or ion beams (including plasma focused ion beams); or optical beams, such as laser beams for fluorescence excitation. A beam may have a limited extent transverse to its main longitudinal flow direction. The line connecting the centroids of two or more cross sections of a beam is the "axis" of the beam.

[0021] "Binary search" is a one-dimensional search technique in which discontinuities are accurately found in logarithmic time by successively bisecting the interval spanning the discontinuity. To illustrate the boundary between carbon and silicon, if the images at coordinate locations 0μm and 64μm show the materials carbon ("C") and silicon ("Si"), respectively, then the C-Si boundary may initially be assigned to 32±32μm. If the next image taken at 32μm is determined to be C, then the boundary determination may be improved to 48±16μm. Successive images may show 48μm=Si, 40μm=Si, 36μm=C, 38μm=Si, such that five measurements may improve the boundary determination from 32±32μm to 37±1μm. The number of measurements (5) is equal to the logarithm (base 2) of the improvement in accuracy (32).

[0022] A "boundary" is an interface between two regions that differ in composition, structure, or orientation of the corresponding constituent materials. As an example, a boundary may separate a silicon region from a carbon or tungsten region. As another example, a boundary may separate a single crystal silicon region from an amorphous silicon region. As a further example, a boundary may separate two polycrystalline silicon grains having different orientations. As used herein, determining a boundary may be understood as a shorthand for determining the coordinate position of one or more points on the boundary. A boundary may be an interface between two different materials (such as single crystal Si and amorphous Si), or it may be an interface within a single material such as between two polycrystalline Si grains.

[0023] "Classify" and "classification" refer to the action of assigning an item to one of a finite set of predetermined choices. In some examples of interest herein, the classified item may be an imaged area of ​​a sample, and each of the choices may be a corresponding material. In various examples, classification may be performed by trained machine learning software, by other automated software, or by a user (e.g., using interactive software). A software program that performs classification is referred to as a "classifier."

[0024] A "controller" is an electronic device coupled to one or more actuators to effect a change in a physical parameter, or coupled to one or more sensors to monitor a physical parameter. Some controllers may include a microprocessor that may be programmed to execute machine-readable instructions. The description of computing devices herein generally applies to such controllers, which may also include additional electronic circuitry (such as filters and amplifiers). Other controllers may include analog circuitry, such as filters and amplifiers, without any microprocessor.

[0025] A "convergent beam electron diffractometer" ("CBED") is an analyzer configured to perform electron diffraction analysis of a sample using a converging electron beam. A converging beam can provide a narrower beam spot (typically in the range of 1 nm to 100 nm or 0.3 nm to 300 nm in diameter) than a parallel beam (typically at least 500 nm) to improve spatial resolution. A converging beam can form a cone with a half-width of about 0.1-1°, which can appear as a widening of features observed in reciprocal space, for example, a conventional electron diffraction spot appears as a widening disk. In some examples, a CBED can project the electron diffraction pattern onto a fluorescent screen and image it with a CCD camera. Other detectors and readouts can also be used.

[0026] A "coordinate position" numerically specifies the location of an associated entity in physical space, where each of one or more coordinates indicates the distance of the location in the corresponding dimension. Coordinates can be one-dimensional coordinates along a line or curve, two-dimensional coordinates along a plane or curved surface, or three-dimensional coordinates. Coordinates can include a mixture of distance and angular coordinates (e.g., in polar, cylindrical, or spherical coordinate systems), or only distance coordinates (e.g., in a Cartesian coordinate system). The coordinate position of a point on a boundary is a "boundary position."

[0027] An "electron beam" is a directed flow of electrons.

[0028] An "electron microscope" is an analyzer in which a sample is illuminated by a beam of electrons and the resulting particles or electromagnetic radiation are used to form a spatially resolved direct image. A scanning electron microscope (SEM) images a sample surface based on reflections, secondary or backscattered particles or radiation from one or more surfaces of the sample. Because the beam interactions detected by the SEM occur at or near the surface, the SEM can operate on samples of arbitrary thickness. In contrast, a transmission electron microscope (TEM) images a sample volume based on transmitted electrons (including scattered electrons). TEM operates on samples with a thickness of about 10nm to 150nm, which can be mounted on a grid for mechanical support and thermal conduction. TEM can provide magnifications up to and exceeding 50 million (and resolutions below 5pm), while SEM magnifications are typically limited to about 2 million (resolutions as low as about 0.5nm). In the present disclosure, a scanning transmission electron microscope (STEM) that uses a scanning electron beam to perform imaging of transmitted electrons is considered to be both a SEM and a TEM. The electron beam in an electron microscope can be generated in an electron gun and accelerated, focused or turned toward the sample chamber through a series of stages.

[0029] A "feature" is a structure or other change discernible in an image of a sample. In a direct image, a feature may include a structure or material boundary, such as a component of a memory cell, a transistor, a grain boundary, or an interface between a bulk substrate layer and another layer. In reciprocal space, exemplary features may include a moiré pattern, a Kikuchi line, a central spot, or a halo. Reciprocal space features may provide clues about the composition, structure, or orientation of the material in the imaged region of the sample, and changes in these features from one region to the next may be used to determine material boundaries. A "moiré pattern" is a repetitive oscillation of image intensity, sometimes associated with interference received, for example, from two sources or scatterers. A "Kikuchi line" is a bright band in the electron diffraction pattern associated with a lattice plane of the imaged sample. The position and orientation of the Kikuchi line depends on the tilt and azimuthal orientation of the associated lattice plane relative to the incident electron beam. A reciprocal image may also have a "central spot" as a bounded bright area and a surrounding "halo" as an annular region with an observable intensity (per pixel) at least one to two times lower than the peak intensity of the central spot. The central spot is typically circular or nearly circular (e.g., with a ratio of major diameter to minimum diameter in the range of 1 to 2) and is typically centered in the reciprocal image (e.g., aligned with the axis of the incident beam), but this is not required, and in other examples, the central spot can be eccentric or elongated. Features can exhibit "symmetry" in the direct image or the reciprocal image, meaning that a single feature or group of features remains unchanged under either: (1) a finite rotation of less than 360° about an axis (called the "symmetry axis"), or (2) reflection in a plane (called the "symmetry plane"). The symmetry axis and symmetry plane can be considered to extend orthogonally from the image plane.

[0030] The term "imaging" refers to a process for obtaining a two-dimensional representation (referred to as an "image") of parameter values ​​over a region of interest of a sample. In some examples, the imaging parameter may be backscatter or secondary emission from an incident particle beam (e.g., by a SEM or low current focused ion beam), but this is not required, and visible light or other parameters may also be imaged. In the disclosed examples, the image may indicate an object on or in the sample (e.g., a gate, layer, metallization, or via) or a feature of such an object (e.g., an edge).

[0031] An "imager" (sometimes, an "imaging tool") is a device capable of imaging. Some imagers may use the response to an incident light beam (e.g., beam current, secondary emission, backscatter, or fluorescence) as a measurement parameter for forming an image. Imagers that acquire or provide a direct image of a target sample are referred to as "microscopes." Microscopes that use an incident electron beam or ion beam may be referred to as "electron microscopes" (e.g., SEM, STEM, or TEM) or "ion microscopes" (e.g., FIB), respectively. Imagers in which the measurement parameter includes the intensity, wavelength, or another property of the light may be referred to as "optical microscopes." Some optical microscopes use an incident optical beam (e.g., for fluorescence), but this is not required, and other beams or diffuse or ambient lighting may also be used. Other imagers may acquire or provide reciprocal space images of a sample and may include diffractometers, such as CBEDs or X-ray diffractometers. The term "charged particle imager" covers electron microscopes, ion microscopes, and CBEDs. Various detectors may be used to measure particles or radiation emitted from the imaged sample, including secondary electron detectors, secondary ion detectors, photodetectors, or spectrometers. The technique used to obtain the image is called an "imaging modality." Thus, scanning electron microscopy, transmission electron microscopy, and convergent beam electron diffraction are different modalities utilized by SEM, STEM, TEM, and CBED tools, respectively. In some examples, the same beam column can be used for multiple imaging modalities. In further examples, the analyzer can include multiple beam columns, such as corresponding columns for electron beams, ion beams, or laser beams.

[0032] "Material" refers to a characterization of the substance present in a sample region. Characterization may include composition, for example, silicon, silicon dioxide, and carbon are three different materials. Characterization may also include structure, for example, carbon may have a hexagonal or cubic structure in graphite and diamond, respectively. Silicon may have a single crystal, polycrystalline, or amorphous form. "Polycrystalline" silicon ("poly-Si") may be organized into grains (typically ranging from 2nm to 500nm), with a regular crystal structure within each grain and orientation changes (or another discontinuity) between adjacent grains. In the present disclosure, if the composition and structure are the same, the substances in two regions are considered to be the same material even if the orientation is different. Therefore, adjacent polycrystalline silicon grains are considered to be the same material. In addition, a material may have internal boundaries, such as where there are discontinuities in its structure (such as at grain boundaries). "Amorphous" silicon lacks a regular structure: the orientation of Si-Si bonds 1nm or more away may not be relevant. "Single crystal" silicon ("monocrystalline Si") has a uniform crystalline structure over 1 μm to 30 cm in a plane, but may be adjacent to layers comprising amorphous silicon or polycrystalline silicon. Some materials of interest herein may have a composition characterized by an atomic number. The atomic number of silicon is 14. Elements with atomic numbers greater than or equal to 15 (such as tungsten) may be considered "high Z" materials, while elements with atomic numbers less than or equal to 13 (such as carbon) may be considered "low Z" materials. In a further example, the structure of a material may encompass the physical layout of devices or components on or within a sample. For example, a memory array may have a specific regular structure that is different from the organization of an adjacent logic or processor block. Thus, a memory array and a logic or processor block may be considered to be different materials that can be easily distinguished in reciprocal space (e.g., in a transformation of a direct image), with the memory array having a sharp spectral peak due to its regular layout and the logic or processor block having a diffuse spectral density.

[0033] A "neural network" is an artificial network of "units" (or "cells") that have connections that simulate the behavior of biological neurons and can be implemented by a software program on a computer. A common neural network can be organized into layers, and at each layer, an input array of data is operated on to generate an output array. Some of the neural networks described herein are "convolutional neural networks" (CNNs), in which at least one layer involves multiple segments of an input array, which have one or more common cores to generate a corresponding output array. In an example, the disclosed technology can be implemented on a residual CNN (ResNet). A neural network has a structure, such as layers, neuron types, and operations or paths from input to output array at each layer; and also has an array of coefficients or weight data to be applied to each neuron or operation. Examples of the disclosed technology apply neural networks to, for example, classification of materials in an imaged sample area.

[0034] A "parameter" is a quantity that can have a specific "value". While parameters often have numeric values, this is not required, and some parameter values ​​can be logical values, strings, or data structures. Some parameters can be control parameters, such as those set by a controller to affect a physical device or physical operation. Other parameters can be sensed parameters, such as those determined by the physical environment and measured or evaluated (e.g., to form an image or to measure or evaluate from an image) by a controller or measurement device.

[0035] "Physical space" refers to the two-dimensional or three-dimensional area in which a sample exists tangibly. A sample has a length, a width, and a thickness (or alternatively an extent) in physical space. An image in which the extent or spacing of features is proportional to their extent or spacing in physical space is called a "direct image." For illustration, a visible light camera or a microscope can form a direct image of a sample.

[0036] "Reciprocal space" refers to a two-dimensional or three-dimensional range in which coordinate values ​​are proportional to spatial frequencies in physical space. For illustration, consider a spatially periodic feature A with a wavelength that is twice that of another spatially periodic feature B. In reciprocal space, the coordinate position of feature A may be half that of feature B. Images in reciprocal space are called "reciprocal images." Diffraction tools such as CBED can generate reciprocal images directly from a sample. Reciprocal images can also be obtained by transforming a direct image (e.g., by FFT). For example, the transformed image may be sensitive to the texture of the imaged surface, or may allow materials to be identified or differentiated based on spectral content.

[0037] An "area" is a contiguous two-dimensional portion of a sample surface, or a contiguous three-dimensional portion of a sample volume.

[0038] A "region of interest" (ROI) is a portion of a sample that includes a feature or structure that is the subject of subsequent analysis. A ROI can be defined relative to a surface of a sample (e.g., an area of ​​an imaged surface that contains a particular feature, or an area in which a pattern can be milled to form a slice containing this feature) or relative to a volume (e.g., a volume containing a structure to be further analyzed). The term ROI does not refer to any human interest. A region may be a region of interest, however some regions may lack a particular feature or structure and may not be a region of interest.

[0039] A "sample" is a physical object on which imaging, milling, or other analysis or manufacturing processes are performed. Common samples may include biological samples, multilayer electrical or electronic structures, or other material structures. Electrical or electronic samples may include semiconductor devices such as logic gates, memory cells, optoelectronic devices, passive components (including interconnects), transistors, and may be at various stages of manufacture or disassembly. Material samples may include microelectromechanical systems (MEMS) or nanostructures. Biological samples may include cells, proteins, DNA, RNA, viruses, or fragments thereof. The disclosed techniques may be applied during sample preparation, sample characterization, or during manufacture. A sample may include a "substrate" on which cells, structures, or devices are deposited or manufactured prior to thin section preparation or other analysis.

[0040] A "scan" is a pass of the imager over the sample, such as taking a series of images at corresponding locations. The imager can scan imaging positions on a predetermined path. In some examples, consecutive imaging positions can be "sequential", meaning that the coordinate values ​​of the imaging positions change monotonically during the scan. For illustration, for each consecutive image, the coordinates of the imaging positions can be incremented in steps of 200nm. In other examples, a binary search can be advantageously used so that the coordinate position of the boundary is quickly narrowed to a predetermined tolerance. Although the scan is typically a one-dimensional scan performed along a straight or curved path, this is not required, and two-dimensional scans can also be used, for example to map out complex boundary contours. Further, a discrete set of one-dimensional scans can be used in any combination in different directions or offset from each other.

[0041] A "semiconductor device" is a device that includes one or more semiconductor regions (e.g., doped or undoped silicon, germanium, III-V semiconductors, or similar materials). A "fabricated semiconductor device" can be a finished product (e.g., a product that can perform its intended electronic function when an appropriate electrical signal is applied), a portion thereof, or a product at an intermediate stage of fabrication (e.g., a product on which at least one patterning operation has been performed). Semiconductor devices typically include other materials to provide insulation or conductive paths between semiconductor regions. Non-limiting examples of semiconductor devices include transistors, memory cells, and some optoelectronic devices. Memory interface circuits can include both conductive wiring and logic gates.

[0042] "Software" refers to a computer executable program, instructions, or associated data structures. Software may be in an active or static state. In an active state, the software may be loaded into a memory or may be executed by one or more processors. In a static state, the software may be stored on a computer-readable medium, awaiting transmission or execution. "Instructions" (or "program instructions") encode the operations to be performed by the processor (sometimes together with one or more operands, operand addresses, or destination addresses). A set of instructions may be organized to perform a function. "Interactive software" refers to a software program that presents information to a user and receives commands from the user in response to the information presented. "Automated software" refers to a software program that receives information from equipment or from other software programs and selects one or more actions to be performed by equipment or other software programs without user input.

[0043] "Tolerance" is a predetermined limit on the acceptable deviation between the reported value of a parameter and its true value. In some examples, a material boundary may be determined to a 100nm tolerance, meaning that the reported boundary coordinate location is within 100nm of the exact location of the boundary. The tolerance range may be 1nm to 10mm.

[0044] The terms "top", "bottom", "upward", "downward", "above", "below", "horizontal", "vertical", etc. are used for convenience relative to a common configuration in which an exposed major surface of a horizontal sample has an outward normal pointing upward, such as a sample that can approach a processing tool from above. The axis of the processing tool (e.g., the boresight of an imaging tool or the beam axis of a FIB milling machine) can be generally vertical down to the surface, or can be tilted at an angle (sometimes in the range of 40° to 60°) relative to the normal. Those skilled in the art will understand from this disclosure that the choice of the actual orientation can be varied without departing from the scope of the disclosed technology.

[0045] "Training" refers to the process of determining the values ​​(coefficients) to be applied at the neurons of a neural network, and may render the neural network operable to achieve its desired function. Training may be performed using a training data set including images whose desired classification is known. Comparing the actual output of the trained neural network to the desired classification may provide a loss function, which may be back-propagated through the neural network, applying gradient descent or another established technique to adjust the coefficients of each layer. As training proceeds, the neural network output may converge to the desired classification, and the magnitude of the loss function may decrease. When the loss function decreases below a predetermined threshold, the neural network may be considered "trained" and may be deployed to classify new images for which the correct classification may not be known.

[0046] "Transformation" refers to the act of converting a direct image (or direct space data) to a reciprocal image (or reciprocal space data) or vice versa. The transformation may be performed by software (e.g., using a Fast Fourier Transform (FFT) or a wavelet transform), or may be physically performed by an imager. For illustration, a CBED may transform a physical structure into a reciprocal space image.

[0047] A "user" is a person. An exemplary user may control an analytical process or equipment, or may receive presentations or notifications of data from such analytical equipment. A user may interact with equipment, samples, or software via a communications network, a computer input or output interface, or client software.

[0048] Instance Device

[0049] Figure 1 100 is a diagram of an example device by which the disclosed technology can be implemented. The device delivers an imaging beam to a sample to obtain an image, which can be classified by a controller. The imaging beam can be scanned relative to the sample surface to determine a boundary between a first material and another material.

[0050] Imager 140 may include a beam source 142, a detector 144, and a reader 146. Imaging beam source 142 may be configured to direct incident beam 112 to sample 120 along axis 105. Beam 112 may be scanned over sample 120 to obtain a series of images at corresponding locations on sample 120. Sample 120 may be fixed on stage 130. Detector 144 may be configured to detect scattered particles of sample 120. Signals representative of the particles may be measured by reader 146.

[0051] In an example, the imager can be a SEM, STEM, TEM, or CBED, but this is not necessary, and another microscope or diffractometer can also be used. As shown, the light beam 112 is convergent, as may be used in a CBED, but this is not necessary, and a parallel light beam can be used. A divergent light beam can also be used, for example to increase the irradiation area of ​​the sample. As shown, the detector 144 can be a segmented annular detector, such as can be found in a diffractometer, but this is not necessary, and in other examples, one or more discrete sensors can be used, such as an Everhart Thornley detector, a microchannel plate, or a Faraday cup. The reader 146 may include one or more amplifiers and / or ampere meters. The scanning of the light beam 112 above the sample 120 can be performed by manipulating the light beam 112, translating the stage 130, or a combination thereof.

[0052] The controller 150 may be coupled directly or indirectly to the imager beam source 142, the stage 130, the reader 146, or the detector 144. Thus, the controller 150 may control and acquire images from the imager 140 and cause the beam 112 to be scanned relative to the sample 120. The controller 150 may also process the images received from the imager 140 to classify the material within the imaged region of the sample 120, for example, in a predetermined set of materials. In some examples, the imager 140 may provide a reciprocal image directly to the controller 150, while in other examples, the imager 140 may provide a direct image to the controller 150, which may be transformed by the controller 150 into a reciprocal space representation prior to classification. Based on the material classification at a series of imaged regions of the sample 120, the controller 150 may determine the coordinate location of the boundary of the first material in the sample 120.

[0053] A variety of extensions or variations of the device shown may be implemented within the scope of the disclosed technology. In an example, the imager 140 may be a direct space imager (such as a SEM, STEM, TEM, or FIB) or a reciprocal space imager (such as a CBED). The illustrated device may also support a second imaging modality, such as another electron microscopy modality, FIB, or fluorescence imaging. The controller 150 may be configured to acquire another image of the sample 120 using a second imaging modality. The image may be acquired at a second coordinate position having a predetermined spatial relationship relative to the coordinate position of the boundary. In some examples, the beam 112 may be used for two imaging modalities, while in other examples, two imaging modalities may be provided in a dual beam tool. The detector 146 may be a pixelated electron diffraction detector having at least 100 pixels (and up to one billion pixels). The current received at each pixel of the detector 146 may be accumulated as a charge at each pixel, and the pixel charge may be read out using a pixel array readout electronic device 146.

[0054] In some examples, the controller 150 may be configured to identify a boundary of single crystal silicon, for example, an interface between a single crystal silicon bulk layer and one or more other materials, such as amorphous silicon, polycrystalline silicon, tungsten or another high-Z material, or carbon or another low-Z material. In some examples, the controller 150 may be configured to perform a binary classification, such as (1) a given first material or (2) not a given first material, while in other examples, the controller 150 may be configured to classify an image as indicating any one of three or more materials.

[0055] The controller 150 may use a convolutional neural network to perform the classification. The controller 150 may be configured to scan the beam 112 at sufficiently closely spaced locations on the sample 120 to establish the boundary location within a predetermined tolerance. The tolerance may be about 100 nm, or in the range of 10 nm to 50 nm, 50 nm to 200 nm, or 200 nm to 1 μm.

[0056] The controller 150 may have a wired or wireless network interface to connect 152 to a remote computer or control station.

[0057] Example Principle

[0058] Figure 2 201-205 is a collection of images illustrating the operating principles of an example of the disclosed technology. Image 201 is a reproduction of a high angle annular dark field (HAADF) STEM image of a sample, which is a direct image formed by high angle forward scattered electrons. CBED images 202-205 are taken at corresponding regions 212-215 of the sample and have characteristic properties of the material in the sample region.

[0059] Starting with image 202, a bright central spot 221 can be seen that has a speckled but otherwise generally uniform intensity. While no distinct halo is present, some regular Kikuchi lines 223 can be seen flaring outward from the core 221. These features can be about the axis of the incident beam (similar to Figure 1 The axis 105) has the characteristics of a regular crystal structure with symmetry, where strong diffraction from the principal planes enhances the visibility of the Kikuchi lines outside the core, and there are no irregular changes to wash out the spots. Such features can be associated with single crystal silicon, for example, in the bulk layer.

[0060] Turning to image 203, a bright central spot 231 can be seen, with Kikuchi line features 233 (bright), 235 (dark) both within the central spot 231 and extending significantly beyond the central spot. It can be seen that the image has significant asymmetry. Such features may be characteristic of a regular crystal structure, lacking strong symmetry about the axis of the incident beam, which may be produced by tilting of the beam or lattice. Image 203 may also be associated with single crystal silicon.

[0061] Image 204 shows a bright central spot 241 with only modest intensity variation and no significant intensity outside the central spot 241. Such features are indicative of low scattering, and may be associated with low-Z amorphous materials such as carbon.

[0062] Image 205 shows a central spot 251 with low intensity and some speckling, surrounded by a distinct symmetrical diffuse halo 254. The low intensity and speckling indicate mixing of diffraction from grains with different orientations (e.g., polycrystalline material). (The CBED spot size can typically be smaller than the grain size: signal mixing can occur between grains at the sample surface and differently oriented grains buried below.) The halo may be the result of strong scattering from high-Z materials. Image 205 may be associated with tungsten.

[0063] Images 202-205 illustrate how varying materials in regions 212-215 can result in distinctly different features in a reciprocal space representation, such as a diffraction pattern. These images are exemplary only: other features may be used to identify a wide range of materials in addition to or in lieu of the features described. In some examples, CBED may be advantageous due to its small spot size. Materials may be accurately identified from very small areas of the sample surface, and material boundaries may be determined with very high resolution in the range of 10nm to 20nm, 20nm to 50nm, 50nm to 100nm, or 100nm to 200nm. Additionally, artificial neural networks may be trained to perform such material identification with high accuracy, providing better than 90%, 95%, 98%, or 99% correct material identification in varying applications. Automated, accurate boundary identification may enable workflow automation, providing registration of the sample in the field of view of, for example, a STEM analyzer, with which high-resolution TEM imaging, probing, or milling may be performed at precise locations without manual assistance.

[0064] Example application and binary search

[0065] Figure 3 is a diagram 300 illustrating the use of a binary search in an example of the disclosed technology. An image is acquired and processed at a series of regions to narrow down the coordinate location of a boundary within a predetermined tolerance.

[0066] 1. The disclosed technology provides advantages

[0067] Image 310 is a reproduction of a high angle annular dark field (HAADF) STEM image of a sample with fins. Some bright lines in image 310 may be STEM artifacts indicating edges, while other bright lines may indicate a high Z buffer layer. Image 310 shows a relatively simple structure of a semiconductor device, which however has many such bright lines of varying thickness, orientation, and length. In an automated workflow, it may be desirable to find the coordinate position of edge 312. Comparative machine vision techniques may be deceived by other edges in image 310, resulting in low reliability (e.g., below a predetermined threshold, which may be in the range of 50% to 89%), and are therefore not suitable for automated workflows. In contrast, the disclosed technique is not deceived by STEM artifacts, thereby allowing edge 312 to be reliably determined. That is, the disclosed technique may determine the coordinate position of edge 312 within a predetermined tolerance and with a probability greater than a predetermined threshold. The predetermined tolerance may be in the range of 1 nm to 10 μm, 10 nm to 1 μm, 30 nm to 300 nm, or about 100 nm. The predetermined threshold may be in the range of 90% to 99.999%, such as approximately 90%, 95%, 98%, 99% or 99.9%.

[0068] Although image 310 shows a relatively simple structure of a semiconductor device, the disclosed techniques can be applied to fairly complex samples, providing reliable automatic determination of boundary locations.

[0069] 2. Application in automated workflow

[0070] Image 310 depicts a row of gate devices 314 formed above edge 312. In an example application, a high-resolution TEM image or another analysis process can be performed at a predetermined offset from edge 312 based on the coordinate position of edge 312 determined using the disclosed techniques. That is, once edge 312 is found, the layers above edge 312 can be reliably analyzed and characterized. For illustration, if the z-coordinate of edge 312 has been found along path 330, subsequent analysis can be reliably performed at a desired offset along path 330 (e.g., at region 351).

[0071] In another application, two coordinate positions of edge 312 may be determined based on two scans to determine the orientation of edge 312 , thereby allowing reliable positioning of subsequent analysis regions (eg, region 352 ) away from path 330 .

[0072] In further applications, multiple boundary locations may be determined along respective paths and between respective materials, for example to determine the contour of gate 314. Thereafter, one or more regions may be precisely selected relative to the contour of gate 314 for subsequent analysis, such as at region 353.

[0073] 3. Binary search acceleration

[0074] Turning back to edge 312, Figure 3 The right side of exemplifies the coordinate position of edge 312 determined along path 330. Reciprocal space images are obtained at a series of positions along path 330. Representative images 321, 325 are illustrated. Each image 321-325 is located in the z direction according to the corresponding imaging position. For illustration, a lead line 364 points to the area 374 where image 324 was acquired and is marked with the corresponding z coordinate "z4". Similar leads indicate the z positions and imaging areas (in Figure 3 (not marked in the figure).

[0075] Initially, images 321, 322 at two relatively distant positions z1, z2 can be obtained as disclosed herein and these images can be classified. As shown in the figure, images 321, 322 are CBED images and can be directly classified. Different features of images 321, 322 result in their different classifications, thereby enclosing the position of the sought edge 312 within the interval (z1, z2). Then, an image 323 can be obtained at the position z3 = (z1 + z2) / 2 and the image can be classified. Its features may result in a classification that matches the classification of image 321, thereby indicating that the edge 312 can be found within the narrowed interval (z3, z2). A fourth image 324 can be obtained at z4 = (z3 + z2) / 2. The classification of image 324 is different from the classification of image 321, although it may or may not be the same as the classification of image 322. Therefore, the position of the edge 312 can be further narrowed down to (z3, z4).

[0076] So far, the interval enclosing the edge 312 remains greater than a predetermined threshold, for example, T = 100 nm. Therefore, the bisection continues, and finally a reciprocal image 325 can be obtained, and this reciprocal image can also be classified differently from image 321. Now, the interval (z3, z5) containing the coordinate position of the edge 312 has a span Δ = |z5 - z3| < T. Since the span Δ is less than the threshold T, the binary search method can be stopped. The coordinate z = (z3 + z5) / 2 can be returned as the position of the edge 312, within a tolerance of Δ / 2.

[0077] Example Method

[0078] Figure 4 is a flowchart 400 depicting a first exemplary method according to the disclosed technique. In this method, reciprocal space images at a series of sample positions are obtained and these reciprocal space images are classified to identify the boundary position of the first material.

[0079] At process block 410, a series of reciprocal space images can be obtained for corresponding regions of the sample. By way of illustration, the images can be electron diffraction images from consecutive positions on the surface of the sample. At block 420, each image can be classified to identify the materials present in the corresponding region from a predetermined set of materials. Then, at block 430, the boundary position of the first material can be determined based on the classification.

[0080] Many variations and extensions can be implemented within the scope of the disclosed technique. For example, in Figure 4In the figure, boxes 410, 420 are illustrated as overlapping to indicate that classification can begin before image acquisition is completed. Specifically, the image can be classified as it is acquired, and the classification can be used to guide the selection of subsequent imaging areas. The classification at box 420 can be performed independently for each image in the image. The classification can be based on the presence or absence of a moiré pattern, the presence or absence of Kikuchi lines, the intensity of the central spot, the absolute or relative intensity of the halo, or the symmetry of one or more features of the instant reciprocal image. The classification can be based on spectral properties detected in the transformation of the direct image. The sample can be a manufactured semiconductor device. The first material can be single crystal silicon, which is commonly used as a bulk substrate for semiconductor devices.

[0081] In an example, a series of images may be acquired using a binary search on a predetermined path at box 410. Initially, with two imaging locations on opposite sides of the boundary being sought, a third image may be acquired at a location located along the predetermined path and between the two imaging locations. This may be repeated for additional images, with the next image being acquired at a location between the most recent imaging location and another previously imaged location. Since the two previously imaged locations may be on opposite sides of the boundary being sought in any iteration, the next image may narrow the interval containing the desired boundary, and the binary search may be iteratively performed to converge on the desired boundary location within a predetermined tolerance. In some examples, the boundary location may be confirmed by taking two additional images along the predetermined path on opposite sides of the determined boundary location and classifying the additional images to confirm that exactly one of the two images indicates the presence of the first material.

[0082] exist Figure 1 Further extensions and variations are described in the context of or FIG. 5 or elsewhere herein.

[0083] Example method extension

[0084] FIG. 5A to FIG. 5B 5 are flow diagrams 501 - 502 depicting an example extension of the first method.

[0085] Figure 5A The expansion of continues from block 430 and illustrates the use of two scans to determine the orientation of the interface between the two materials. In the example, the orientation of the interface between the two materials can be determined by scanning along a first predetermined path. Figure 4At process block 510, the actions of blocks 410 (acquiring) and 420 (classifying) may be repeated for a second set of images acquired on a second predetermined path different from the first path. Thus, at block 520, a second boundary location of the first material may be determined along the second path. Then at block 530, the two boundary locations determined at blocks 430, 520 may be used to determine the orientation of the interface of the first material. For example, a straight line connecting the first boundary location and the second boundary location (or a normal to the straight line) may indicate the interface orientation.

[0086] In an alternative example, still further scans may be performed to identify further boundary locations, and in the case of a plurality of such boundary locations, the boundary of the first material may be traced for any boundary shape.

[0087] Figure 5B The extension of also continues from block 430 and illustrates the application of the disclosed technology in an automated workflow. As an example, a workflow goal may be to acquire a TEM image at a precise location (e.g., offset by 200 nm) relative to the top surface of a bulk single crystal material. In this workflow, Figure 4 The method can be used to locate the boundary between single crystalline material and polycrystalline material using a first imaging modality (e.g., SEM, STEM, or CBED). Subsequently, the TEM beam can be accurately positioned at a desired offset from the boundary, and a desired TEM image can be acquired. In a variation, the workflow goal can be to perform FIB milling or electrical probing at a precise location relative to the boundary.

[0088] At block 540, another image of the sample may be acquired using a second imaging modality different from the first imaging modality at a coordinate location having a predetermined spatial relationship to the boundary location determined at block 430. In an example, the second imaging modality may be a transmission electron microscope.

[0089] As a further extension, by repeating Figure 4 or Figure 5A The method can be used to determine a plurality of boundary locations or boundaries, and the coordinate position for performing the second imaging modality can be set relative to the plurality of boundary locations or boundaries. In some examples, a common beam column can be used for both imaging modalities.

[0090] Example Neural Network

[0091] Figure 6 600 is a diagram illustrating an example neural network that can be used to implement the disclosed technology. As shown, the neural network 620 is Figure 6 The upper left corner receives the reciprocal space representation of the imaged area and the lower right corner outputs the classification of the material in the imaged area.

[0092] The description begins with an input image. In varying examples, the disclosed techniques may be implemented using an imager that generates a reciprocal space or physical space image. In the former case, the reciprocal image 612 may be directly input to a neural network 620, as shown. In the latter case, the direct image 614 may be transformed at process block 616 (e.g., by FFT) to generate a reciprocal space representation of the imaged region, which may then be input to a neural network 620, as shown.

[0093] Neural network 620 is illustrated as a residual neural network ("ResNet"), which is a deep neural network having multiple convolutional layers with bypass jump connections. ResNet can be configured to reshape a wide and shallow representation of the input data space into a narrow and deep representation of the output data space. As data flows through the neural network, the data space can evolve from a representation of an image to a representation of a classification. However, for ease of description, the data array of the entire neural network is described herein as a pixel array.

[0094] In an example, the input data space can be a reciprocal space image of the sample area, similar to Figure 2 For illustration, the input data size may be a 1024×1024 pixel intensity array, and since the image data is monochrome, there may be only 1 slice of this size. Thus, the input data size may be 1024×1024×1. In a variation, RGB color data with three slices may be used, in which case the input data size may be 1024×1024×3. These image array sizes are exemplary. Slice sizes from 64×64 to 4096×4096 are common, and other smaller, larger, or differently shaped slices may also be used. Specifically, the slice sizes need not be equal, nor need they be powers of 2.

[0095] A. Convolutional block 650

[0096] As shown, neural network 620 includes multiple instances of convolutional blocks 650, labeled 650A-650J, and several other layers. Figure 6 In the upper right corner of FIG, block 650 is shown in parameterized form with J input slices (commonly referred to in the art as a "batch"), K output slices, N×N convolution, and downsampling by a factor of S. Block 650 receives J slices of the data array, copies of which are processed in parallel along two paths.

[0097] The left side of block 650 depicts the convolution path. At block 651, N×N convolution operators are applied across each slice of the input data array, and there may be up to K×J such N×N matrices to cover all combinations of J input slices and K output slices. Additionally, ÷S downsampling may be performed at block 651. For example, ÷2 downsampling may reduce a 1024×1024 data slice to 512×512, which may compensate for the increase in the number of slices as the data flows down the neural network 620. For S=1, no downsampling is performed, and a given slice of the input data retains its size after processing. Then, at block 652, batch normalization may be performed across the K slices output from block 651, followed by a rectified linear unit layer (RelU) at block 653, which may introduce nonlinearity into the transfer function of block 650 or the neural network 620. The output of block 653 may be input to another set of N×N convolutional filters (which may be K×k in number) at block 654, followed by another batch of normalization stages 655. For input data organized as J slices (each slice having R rows and C columns), the output of batch block 655 may be K slices in an (R / S)×(C / S) array.

[0098] The right side of block 650 depicts a skip path that bypasses the convolution path. At block 656, ÷S downsampling may be performed on the input data array slices so that the size of each array slice received from the two paths at block 657 is the same. The data arrays from the two paths may then be added at block 657, and the sum passed through another RelU layer at block 658.

[0099] B. Neural Networks 620

[0100] Returning to the neural network 620, the first block 650A is an instance of a convolution block 650 with J=1 input slices, 7×7 convolutions, ÷2 downsampling, and K=64 output slices. The next block 622 is a pooling layer that can perform ÷2 downsampling. The pooling layer can use a maximum pooling function so that each output pixel has the maximum value in the corresponding four-pixel (2×2) group in the input array. Block 622 is followed by two identical instances 650B-650C of block 650, each of which performs 3×3 convolutions and does not perform downsampling, with 64 input slices and 64 output slices. The next two blocks 650D-650E are instances of block 650 that perform 3×3 convolutions and generate 128 output slices. Block 650D performs ÷2 downsampling to compensate for the increased slice depth, while block 650E maintains the slice size and number of slices. Another pair of blocks 650F-650G is also an instance of block 650, which performs 3×3 convolution and generates 256 output slices. Block 650F may also perform ÷2 downsampling to compensate for the increased slice depth, while block 650G maintains the slice size and number of slices. The last instance of convolution block 650 in neural network 620 is blocks 650H-650J, which performs 3×3 convolution and generates 512 output slices. Block 650H performs ÷2 downsampling to compensate for the increased slice depth, while block 650J maintains the slice size and number of slices. Block 650J is followed by another ÷2 pooling stage 624. This pooling stage may use an average pooling function so that the value of each output pixel is the arithmetic mean of the four input pixels that provide data for it.

[0101] By design, the data shaping from block 650A to block 624 modestly reduces the number of degrees of freedom or total amount of data flowing through the data array of neural network 620. In this illustration, the total amount of data is reduced by a factor of eight over six ÷2 downsampling stages and an increase from 1 slice to 512 slices.

[0102] The output of block 624 can provide data for a fully connected layer at block 626, which can linearly combine its input with corresponding weights. The output of block 626 can be operated by an activation block 628, which can apply a nonlinear transformation to generate a probability vector 632. Specifically, block 628 can limit its output value to a range [0,1] suitable for probability. For a binary classifier, block 628 can be implemented with a sigmoid function, while for three or more classification outputs, a softmax function can be used. The probability vector 632 can be normalized so that the sum of its elements is equal to one.

[0103] C. Example Classifier

[0104] In the first example, a binary classifier for detecting single crystal silicon may output a 2-element probability vector 632, the first element indicating the probability of "yes" (single crystal Si), and the second element indicating the probability of "no" (non-single crystal Si). Another binary classifier may distinguish between single crystal Si and polycrystalline Si, with only two output elements P(single crystal) and P(polycrystalline), where P(single crystal)+P(polycrystalline)=1. A third classifier may also distinguish between single crystal Si and polycrystalline Si, but may support a third result of "neither", such that P(single crystal)+P(polycrystalline)+P(neither)=1. A fourth classifier may provide additional outputs, such as the probability of single crystal Si, polycrystalline Si, amorphous Si, high Z material, low Z material, or "other" in any combination.

[0105] The probability vector 632 may be analyzed to determine which index position has the maximum value (highest probability), which value may be mapped to a classification 634 and output, thereby concluding beam image processing.

[0106] Trained on about 2000 CBED training images similar to Figure 6 to perform binary classification between crystalline silicon and amorphous silicon.

[0107] Example Scan

[0108] Figure 7 700 is a diagram illustrating an example of using multiple scans to locate a material boundary and determine its orientation. The illustrated examples include the case of a boundary extending in a horizontal direction, a vertical direction, or a tilted direction (such as in Figure 7 plane), and tracing the boundaries of complex shapes.

[0109] Figure 7 A cross-sectional view of a sample 710 is shown, wherein a boundary 712 separates regions 714, 716 of different materials A, B that are distinguishable based on a reciprocal space representation of the image data.

[0110] Initially, regions 721-724 form a quadrilateral that defines four paths 725-728 on which scans may be performed. A first scan may be performed along path 725, thereby obtaining a pattern similar to Figure 4A series of images of box 410 of FIG. 410. In some examples, the images can be acquired in a step-by-step manner. For illustration, where regions 721, 722 are separated by a distance of 2 μm and the desired accuracy for determining the boundary position is 100 nm, path 725 can be divided into 10 intervals, each of 200 nm. A series of images can be acquired along path 725 from region 721 to region 722 in the order of 0, 200, 400, 600, ... nm. The reciprocal space images can be classified as they are obtained, or image processing can be postponed until after the scan along path 725 is completed.

[0111] In other examples, a binary search may be performed to obtain higher efficiency. For illustration, reciprocal space images may be obtained at regions 721, 722 and these reciprocal space images may be classified. When it is determined that regions 721, 722 have different materials and therefore the boundary is located between regions 721, 722, a third reciprocal space image may be obtained in the middle between regions 721, 722 to narrow the interval in which the boundary can be found. Figure 3 As described in the context of , the interval may be iteratively further bisected until the coordinate position 761 of the boundary 712 along the path 725 has been determined to a desired accuracy.

[0112] Paths 726-728 may be scanned in a similar manner. For path 726, the classification of the reciprocal images acquired at regions 722, 723 may indicate that the two regions have the same material B, from which it may be determined that boundary 712 does not intersect path 726, and the scan of path 726 may terminate after only these two image points. Similarly, it may be found that path 728 has no boundary crossing, while path 727 does have boundary crossing 762, which may be found in a similar manner to path 725.

[0113] In this manner, a scan along paths 725, 727 may identify a horizontally oriented boundary. Additionally, in the event that boundary 712 has a rotation relative to the horizontal, coordinates 761, 762 may be used to establish the orientation of the boundary.

[0114] The description next turns to the rectangle formed by regions 731-734, which define paths 735-738. Paths 735, 736 are similar to paths 725, 726, and their respective scans result in a coordinate position 763 of boundary 712 intersecting path 735, and no boundary crossing along path 736. It can be found that the scan of path 737 has the same material B at both endpoints 733, 734, and can be terminated immediately, similar to paths 726, 736. Conversely, material classification at endpoints 731, 734 can reveal different materials, and a binary search along path 738 can establish the coordinate position 764 of boundary 712.

[0115] In this way, scanning along paths 735, 738 can identify obliquely oriented boundaries. In addition, coordinates 763, 764 can be used to establish the orientation of boundary 712 that passes through these paths.

[0116] The description next turns to the square formed by regions 741-744, which define paths 745-748. Paths 748, 747 are similar to paths 738, 737, and their respective scans result in a coordinate position 766 of boundary 712 intersecting path 748, and no boundary crossing along path 747. It can be found that the scan of path 745 has the same material A at both endpoints 741, 742, and can be terminated immediately, similar to paths 726, 747. Conversely, material classification at endpoints 742, 743 can reveal different materials, and a binary search along path 746 can establish the coordinate position 765 of boundary 712.

[0117] In this way, a scan along paths 746, 748 can identify vertically oriented boundaries. In addition, coordinates 765, 766 can be used to establish the orientation of boundary 712 that passes through these paths.

[0118] Turning now to the polygon formed by regions 751-754, an initial reciprocal image classification of vertices 751-754 may indicate that all regions 751-754 have the same material A, thereby indicating that boundary 712 does not pass through the polygon. That is, scan 755 can only be terminated after its endpoints 751-752 have been classified, and similarly for scans 756-758.

[0119] Many variations and extensions may be implemented within the scope of the disclosed technology. As shown, the border 712 is not straight. As described, multiple scans may be used as described to track the border along the coordinate positions 761-766 of the border 712 along the corresponding paths 725, 727, 735, 738, 746, 748. On the other hand, Figure 7 The scans in are shown as grouped along the sides of a square (e.g., paths 725-728 form a square), but this is not required. In other examples, any closed polygon may be used. The crossing of a boundary along any two sides of a polygon may indicate the presence and orientation of the boundary as it passes through the polygon. The error in the orientation (angle) reported may be proportional to the roughness of the boundary.

[0120] Additional Embodiments

[0121] The following paragraphs describe additional numbered embodiments of the disclosed technology.

[0122] Embodiment 1 is a device comprising: an imager configured to acquire multiple images of a corresponding area of ​​a sample; and a controller coupled to the imager and configured to: for each of the images: process a reciprocal space representation of the corresponding image; and in response to the processing, classify a material within the corresponding area among a predetermined plurality of materials; and determine a coordinate position of a boundary of a first material among the plurality of materials.

[0123] Embodiment 2 includes the subject matter of embodiment 1, and further specifies that the imager is a reciprocal space imager, and each of the acquired images includes the reciprocal space representation.

[0124] Embodiment 3 includes the subject matter of embodiment 2, and further specifies that the imager is a convergent beam electron diffractometer (CBED).

[0125] Embodiment 4 includes the subject matter of any one of embodiments 1 to 3, and further specifies that the imager is a microscope, and the controller is further configured to, for each of the images, transform the corresponding image into the reciprocal space representation.

[0126] Embodiment 5 includes the subject matter of any one of embodiments 1 to 4, and further specifies that the imager is a transmission electron microscope (TEM) or a scanning electron microscope (SEM).

[0127] Embodiment 6 includes the subject matter according to any one of embodiments 1 to 5, and further specifies that the coordinate position is a first coordinate position, and the controller is further configured to acquire another image at a second coordinate position having a predetermined spatial relationship with the first coordinate position.

[0128] Embodiment 7 includes the subject matter of embodiment 6, and further specifies that the imager is configured to operate in a first imaging modality, and the another image is acquired using a second imaging modality different from the first modality.

[0129] Embodiment 8 includes the subject matter of any one of Embodiments 1 to 7, and further specifies that the first material is single crystal silicon.

[0130] Embodiment 9 includes the subject matter of any one of embodiments 1-8, and further specifies that the controller is configured to classify the material in any one of three or more materials.

[0131] Embodiment 10 includes the subject matter of any one of embodiments 1 to 9, and further specifies that the classification is a binary classification that distinguishes the first material from other materials.

[0132] Embodiment 11 includes the subject matter of any one of Embodiments 1 to 10, and further specifies that the three or more materials include single crystal silicon and two or more of the following materials: polycrystalline silicon, amorphous silicon, high-Z material, or low-Z material.

[0133] Embodiment 12 includes the subject matter of any one of Embodiment 1, and further specifies that the controller comprises a trained neural network configured to classify the material.

[0134] Embodiment 13 includes the subject matter of any one of Embodiment 1, and further specifies that the coordinate positions are determined within a tolerance of 100 nm.

[0135] Embodiment 14 is a method comprising: acquiring a series of images including a reciprocal space representation of a corresponding region of a sample; for each image in the images: classifying each image to identify a corresponding material present in the corresponding region among a plurality of materials; and based on the classification, determining a boundary position of a first material among the plurality of materials.

[0136] Embodiment 15 includes the subject matter described in embodiment 14, and further specifies that the series of images is a first scan on a first predetermined path, the boundary position is a first boundary position, and the method further includes: repeating the acquisition, classification and determination actions for a second scan on a second predetermined path different from the first predetermined path to identify a second boundary position of the first material.

[0137] Embodiment 16 includes the subject matter of Embodiment 15, and further includes determining an orientation of the interface of the first material based on the first boundary location and the second boundary location.

[0138] Embodiment 17 includes the subject matter of any one of Embodiments 15 to 16, and further includes: performing an additional scan; identifying an additional boundary location from the additional scan; tracking the boundary of the first material based on the first boundary location, the second boundary location, and the additional boundary location.

[0139] Embodiment 18 includes the subject matter of any one of Embodiments 14 to 17, and further specifies that the series of images includes three or more images; and wherein the series of images is acquired by performing a binary search on a predetermined path, wherein the third image and any subsequent images of the image are acquired at corresponding areas along the predetermined path between (i) a most recent image of the image and (ii) a corresponding area of ​​another previous image of the image.

[0140] Embodiment 19 includes the subject matter of any one of Embodiments 14 to 18, and further includes: confirming the boundary location by the following steps: acquiring two additional images at corresponding areas along the predetermined path on opposite sides of the boundary location; and classifying the two additional images; and verifying that exactly one of the two additional images indicates the presence of the first material.

[0141] Embodiment 20 includes the subject matter of any one of Embodiments 14 to 19, and further specifies that the classification is based on detecting: the presence or absence of a moiré pattern, or one or more Kikuchi lines; the intensity of the central spot or halo; and the symmetry of one or more features of each image.

[0142] Embodiment 21 includes the subject matter of any one of Embodiments 14 to 20, and further provides for performing the classification independently for each of the images.

[0143] Embodiment 22 includes the subject matter of any one of Embodiments 14 to 21, and further specifies that the sample includes a fabricated semiconductor device.

[0144] General computer environment

[0145] Figure 8 A general example of a suitable computing system 800 is illustrated in which the described examples, techniques, and technologies may be implemented to apply focal stacking to a sample preparation or analysis process. The computing system 800 is not intended to impose any limitations on the scope of use or functionality of the present disclosure, as these innovations may be implemented in a variety of general-purpose or special-purpose computing systems. The computing system 800 may control a CBED imager, a FIB milling machine, a microscope, a stage, an analyzer, or other similar equipment; may perform or control image acquisition, image transformation, scanning on a sample surface, training or application of a neural network, material classification, metrology, or other analysis of an image or other acquired data representing a sample; may control a stage, an ion beam column, or an electron beam column; or may acquire, process, output, or store measurement data.

[0146] refer to Figure 8 , the computing environment 810 includes one or more processing units 822 and a memory 824. Figure 8, this basic configuration 820 is included within the dashed line. The processing unit 822 can execute computer executable instructions, such as for control, metering or other functions as described herein. The processing unit 822 can be a general-purpose central processing unit (CPU), a processor in an application-specific integrated circuit (ASIC), or any other type of processor. In a multi-processing system, multiple processing units execute computer executable instructions to increase processing power. The computing environment 810 may also include a graphics processing unit or a synergistic processing unit 830. The tangible memory 824 can be a volatile memory (e.g., a register, a cache, or a RAM), a non-volatile memory (e.g., a ROM, an EEPROM, or a flash memory) that can be accessed by the processing units 822 and 830, or a combination thereof. The memory 824 stores software 880 that implements one or more innovations described herein in the form of computer executable instructions suitable for execution by the processing units 822 and 830. For example, the software 880 may include software 881 for controlling a CBED, SEM, or other imager, software 882 for controlling a FIB or other milling tool, software 883 for controlling a stage on which a sample is supported, software 884 for implementing a trained neural network or other classifier, software 885 for performing metrology or other analysis on sample data, or other software 886 (including user interfaces, host interfaces, fault detection, or neural network training). The illustrations shown for the software 880 in the storage device 840 may similarly apply to Figure 8 Memory 824 may also store control parameters, calibration data, measurement data, other database data, configuration data, or operational data.

[0147] The computing system 810 may have additional features, such as one or more of a storage device 840, an input device 850, an output device 860, or a communication port 870. An interconnection mechanism (not shown), such as a bus, a controller, or a network, interconnects the components of the computing environment 810. Typically, operating system software (not shown) provides an operating environment for other software 880 executed in the computing environment 810 and coordinates the activities of the components of the computing environment 810.

[0148] The tangible storage device 840 may be removable or non-removable and include disks, magnetic tapes or cassettes, CD-ROMs, DVDs, or any other medium that can be used to store information in a non-transitory manner and that can be accessed within the computing environment 810. The storage device 840 stores instructions (including instructions and / or data) for the software 880 that implements one or more innovations described herein. The storage device 840 may also store image data, measurement data, workflow programs, reference data, calibration data, configuration data, sample data, or other databases or data structures described herein.

[0149] The input device 850 may be a mechanical, touch-sensing, or proximity-sensing input device (such as a keyboard, mouse, pen, touch screen, or trackball), a voice input device, a scanning device, or another device that provides input to the computing environment 810. The output device 860 may be a display, a printer, a speaker, an optical disc writer, or another device that provides output from the computing environment 810. Input or output may also be transmitted to / from a remote device over a network connection via the communication port 870 (e.g., a computer). Figure 1 described in the context of ).

[0150] Communication port 870 enables communication with another computing entity via a communication medium. The communication medium transmits information, such as computer executable instructions, audio or video input or output, or other data in a modulated data signal. A modulated data signal is a signal that has one or more of its characteristics set or changed in such a manner as to encode information in the signal. By way of example and not limitation, the communication medium may use electrical, optical, RF, acoustic or other carriers.

[0151] The data acquisition system may be integrated into the computing environment 810 as an input device 850 or coupled to the communication port 870 and may include an analog-to-digital converter or a connection to an instrument bus. The instrument control system may be integrated into the computing environment 810 as an output device 860 or coupled to the communication port 870 and may include a digital-to-analog converter, a switch, or a connection to an instrument bus.

[0152] In some examples, computer system 800 may also include a computing cloud 890 in which instructions implementing all or part of the disclosed techniques are executed. Any combination of memory 824, storage 840, and computing cloud 890 may be used to store software instructions and data for the disclosed techniques.

[0153] The innovation is described in the general context of computer executable instructions (e.g., those contained in program modules) that can be executed in a computing system on a target real or virtual processor. Typically, a program module or component comprises a routine, program, library, object, class, component, data structure, etc., which performs a specific task or implements a specific data type. In various embodiments, the functionality of program modules can be combined or split as needed. Computer executable instructions for program modules can be executed within a local or distributed computing system.

[0154] The terms "computing system," "computing environment," and "computing device" are used interchangeably herein. Unless the context clearly indicates otherwise, none of the terms implies any limitation on the type of computing system, computing environment, or computing device. In general, a computing system, computing environment, or computing device may be local or distributed, and may include any combination of special-purpose hardware and / or general-purpose hardware and / or virtualized hardware, as well as software that implements the functionality described herein. Virtual processors, virtual hardware, and virtualized devices are ultimately embodied in a hardware processor or another form of physical computer hardware, and therefore include both the software and the underlying hardware associated with virtualization.

[0155] General considerations

[0156] As used in this application and in the claims, the singular forms "a", "an", and "the" include plural forms unless the context clearly indicates otherwise. In addition, the term "including" means "comprising". Furthermore, the term "coupled" does not exclude the presence of intervening elements between the coupled items. Moreover, as used herein, the terms "or" and "and / or" mean any one or more combinations of the phrases.

[0157] The systems, methods, and devices described herein should not be construed as limiting in any way. On the contrary, the present disclosure relates to all novel and non-obvious features and aspects of the various disclosed embodiments, whether individually or in various combinations and sub-combinations formed with each other. The disclosed systems, methods, and devices are not limited to any particular aspect or feature or combination thereof, nor do the disclosed systems, methods, and devices require the presence of any one or more specific advantages or solutions to problems. The technology from any example may be combined with one or more of the described technologies in any other example.

[0158] Although the operations of some of the disclosed methods are described in a specific sequential order for ease of presentation, it should be understood that unless the specific language stated herein requires a specific order, this description covers the rearrangement of the operations. For example, the operations described in sequence may be rearranged or performed simultaneously in some cases. In addition, for simplicity, the accompanying drawings may not show the various ways in which the disclosed systems, methods, and devices can be used in conjunction with other systems, methods, and devices. Additionally, the description sometimes uses terms such as "acquire", "activate", "apply", "average", "bisection", "classify", "configure", "control", "convolution", "determine", "downsample", "filter", "generate", "recognize", "image", "pool", "process", "produce", "scan", "select", "search", "set", "train", "transform", or "verify" to describe the disclosed methods. These terms are high-level abstractions of the actual operations performed. The actual operations corresponding to these terms will vary depending on the specific implementation and can be easily discerned by those skilled in the art.

[0159] In some examples, values, procedures, or devices are referred to as "lowest," "best," "maximum," "optimal," "extreme," etc. It will be appreciated that such descriptions are intended to indicate that a selection may be made among several alternatives or many alternatives, and that such selections are not necessarily better, lesser, or otherwise preferred than other alternatives not considered.

[0160] The theories of operation, scientific principles, or other theoretical descriptions presented herein with reference to the devices or methods of the present disclosure have been provided for the purpose of better understanding for the purpose of explanation and are not intended to be limited in scope. That is, the disclosed systems, methods, and devices are not limited to such theories of operation. The appended claims are not limited to embodiments that function in the manner described by such theories of operation.

[0161] Any of the disclosed methods may be controlled or implemented as computer executable instructions or computer program products stored on one or more computer readable storage media (e.g., tangible, non-transitory computer readable storage media) and executed on a computing device (e.g., any available computing device, including a tablet, smart phone, or other mobile device including computing hardware). A tangible computer readable storage medium is any available tangible medium accessible within a computing environment (e.g., one or more optical media discs such as DVDs or CDs, volatile memory components such as DRAM or SRAM, or non-volatile memory components such as flash memory or a hard drive). By way of example, and with reference to Figure 8, the computer-readable storage medium includes memory 824 and storage device 840. The term "computer-readable medium" or "computer-readable storage medium" does not include signals and carrier waves. In addition, the term "computer-readable medium" or "computer-readable storage medium" does not include a communication port (e.g., 870).

[0162] Any of the computer executable instructions for implementing the disclosed techniques and any data created and used during the implementation of the disclosed embodiments may be stored on one or more computer readable storage media. The computer executable instructions may be, for example, a dedicated software application or part of a software application accessed or downloaded via a web browser or other software application (e.g., a remote computing application). Such software may be executed using one or more network computers, for example, on a single local computer (e.g., any suitable commercially available computer) or in a network environment (e.g., via the Internet, a wide area network, a local area network, a client server network, a cloud computing network, or other such network).

[0163] For the sake of clarity, only some selected aspects of the software-based implementation are described.Other details well-known in the art are omitted.For example, it should be understood that the disclosed technology is not limited to any specific computer language or program.For example, the disclosed technology can be implemented by using Adobe Flash, C, C++, C#, Curl, Dart, Fortran, Java, JavaScript, Julia, Lisp, Matlab, Octave, Perl, Python, Qt, R, Ruby, SAS, SPSS, SQL, WebAssembly, any derivative thereof, or any other suitable programming language, or in some examples, markup languages ​​such as HTML or XML, or any combination of suitable languages, libraries, and data packets to write software.Similarly, the disclosed technology is not limited to the type of any particular computer or hardware.Some details of suitable computers and hardware are well-known and do not need to be elaborated in the present disclosure.

[0164] In addition, any of the software-based embodiments (including, for example, computer-executable instructions for causing a computer to perform any of the disclosed methods) may be uploaded, downloaded, or remotely accessed via appropriate communication means. Such appropriate communication means include, for example, the Internet, the World Wide Web, an intranet, a software application, cables (including fiber optic cables), magnetic communications, electromagnetic communications (including RF, microwave, infrared, and optical communications), electronic communications, or other such communications means.

[0165] In view of the many possible embodiments to which the principles of the disclosed subject matter may be applied, it should be recognized that the illustrated embodiments are only preferred examples and should not be taken as limiting the scope of the claims. Rather, the scope of the claimed subject matter is defined by the appended claims. We therefore claim all that falls within the spirit and scope of these claims.

Claims

1. A device, comprising: an imager configured to acquire a plurality of images of corresponding areas of the sample; and a controller coupled to the imager and configured to: For each of the images: processing a reciprocal space representation of the corresponding image; as well as responsive to the processing, classifying the material within the corresponding region among a predetermined plurality of materials; and The coordinate position of a boundary of a first material of the plurality of materials is determined.

2. The apparatus of claim 1, wherein the imager is a reciprocal space imager and each of the acquired images comprises the reciprocal space representation.

3. The apparatus of claim 2, wherein the imager is a converging beam electron diffractometer (CBED).

4. The apparatus of claim 1, wherein the imager is a microscope and the controller is further configured to, for each of the images, transform the corresponding image into the reciprocal space representation.

5. The apparatus of claim 1, wherein the imager is a transmission electron microscope (TEM). 6 . The apparatus of claim 1 , wherein the coordinate position is a first coordinate position, and the controller is further configured to cause another image to be acquired at a second coordinate position having a predetermined spatial relationship with the first coordinate position.

7. The apparatus of claim 6, wherein the imager is configured to operate in a first imaging modality and the further image is acquired using a second imaging modality different from the first modality.

8. The apparatus of claim 1, wherein the first material is single crystal silicon.

9. The apparatus of claim 1, wherein the classification is a binary classification that distinguishes the first material from other materials.

10. The apparatus of claim 1, wherein the controller is configured to classify the material in any one of three or more materials.

11. The apparatus of claim 10, wherein the three or more materials include single crystal silicon and two or more of the following materials: polycrystalline silicon, amorphous silicon, a high-Z material, or a low-Z material.

12. The apparatus of claim 1, wherein the controller comprises a trained neural network configured to classify the material.

13. The apparatus of claim 1, wherein the coordinate position is determined within a tolerance of 100 nm.

14. A method comprising: acquiring a sequence of images comprising a reciprocal space representation of a corresponding region of the sample; For each of the images: classifying each of the images to identify a corresponding material among a plurality of materials present within the corresponding region; and Based on the classification, a boundary location of a first material of the plurality of materials is determined.

15. The method according to claim 14, wherein the series of images is a first scan on a first predetermined path, the boundary position is a first boundary position, and the method further comprises: The acquiring, classifying, and determining acts are repeated for a second scan on a second predetermined path different from the first predetermined path to identify a second boundary location of the first material.

16. The method according to claim 15, further comprising: An orientation of an interface of the first material is determined based on the first boundary location and the second boundary location.

17. The method according to claim 15, further comprising: Perform additional scans; identifying additional boundary locations from the additional scans; A boundary of the first material is tracked based on the first boundary location, the second boundary location, and the additional boundary location.

18. The method of claim 14, wherein the series of images comprises three or more images; and The series of images is acquired by performing a binary search on a predetermined path, wherein a third image and any subsequent images of the image are acquired at corresponding areas along the predetermined path between (i) a most recent image of the image and (ii) a corresponding area of ​​another previous image of the image.

19. The method according to claim 18, further comprising: Confirm the boundary location by following these steps: acquiring two additional images at corresponding areas along the predetermined path on opposite sides of the boundary location; as well as classifying the two additional images; and Verifying that exactly one of the two additional images indicates the presence of the first material.

20. The method of claim 14, wherein the classification is based on detecting: the presence or absence of a moiré pattern, or one or more Kikuchi lines; the intensity of the central spot or halo; or The symmetry of one or more features of each of the images.

21. The method of claim 14, wherein the classifying is performed independently for each of the images.

22. The method of claim 14, wherein the sample comprises a fabricated semiconductor device.

Images