Focus stacking applications for sample preparation

Abstract

To provide methods and apparatus for applying focus stacking to sample preparation, improving the accuracy of analytic tasks, facilitating the automation, and improving the throughput.SOLUTION: The focus stacking is applied to a set of sample images having different focus depths so as to produce a composite image in which features at different depths are in focus and, optionally, a depth map. A sample location is selected from the composite image and a localized material removal, measurement or imaging operation is performed based on the sample location. A depth value from the depth map is used to set a working depth of a tool for performing the localized operation. Applications include lamella preparation for cryogenic TEM analysis of biological samples. Other applications, techniques and variations are disclosed.SELECTED DRAWING: Figure 8

Description

[Technical field]

[0001] The present invention relates to focus synthesis applications for sample preparation. [Background technology]

[0002] Transmission electron microscope Microscopy (TEM) requires thin samples (typically 10-150 nm thick) known as lamellae. Lamella preparation can involve, for example, imaging to find the features to be analyzed by TEM, and milling to prepare a lamella incorporating the features. Imaging-based workflows can also be employed in other sample preparation applications.

[0003] Some tools, such as focused ion beam (FIB) or fluorescence microscopes, may have a depth of focus that is less than the depth variation of the sample region being imaged, which can pose challenges in using these tools for imaging. In some workflows, this problem can be circumvented by performing surface imaging using a scanning electron microscope (SEM), which has a large depth of focus, and features identified in the SEM image can be used to guide the subsequent FIB milling. However, correlating an SEM view of a surface with a FIB view of the same surface can be difficult and error-prone, especially for surfaces with significant topographical variations ("bumpy surfaces") when the view axes of the SEM and FIB are different.

[0004] In other workflows, fluorescence imaging can be used to find features inside the sample, but it cannot replace SEM imaging. Using fluorescence imaging to guide lamella preparation can require scrutinizing a series of fluorescence images taken at different focal depths to identify features for TEM imaging. This can be a tedious process, made more tedious by the inability to easily visualize relevant features at different depths.

[0005] Thus, there remains a need for improved techniques for imaging to guide sample preparation, in a range of applications not limited to the production of TEM lamellae. Summary of the Invention

[0006] Briefly, examples of the disclosed technology use focus stacking to integrate and present sample features at different depths, all in focus in a single composite image. Sample locations can be selected from the composite image interactively by a user or by automated software. Based on the selected location, local actions can be taken on the sample.

[0007] In some embodiments, focus stacking can also generate a depth map that can directly indicate the depth value at a selected location or at another location that has a predetermined spatial relationship to the selected location. The combination of the lateral coordinate of a composite image pixel and a depth value from the depth map can precisely specify the three-dimensional location of a given feature. The working depth of a tool can be appropriately set depending on the situation to perform precise localized operations.

[0008] In particular, the disclosed techniques can be applied in environments with topographical variations on the sample surface, grazing incidence beams for imaging or milling, or beams with limited focal depth (e.g., comparable to or smaller than the sample size). The disclosed techniques can address the challenge of achieving nanometer precision when working with samples and tools with micron-scale variations. In one application, the disclosed techniques can be applied to mill TEM lamellae from samples, increasing throughput for a variety of reasons further described herein. However, the disclosed techniques are broadly applicable to improve workflow automation or to improve the accuracy or yield of imaging-based analytical operations. Such operations can include material removal, measurement, or further imaging.

[0009] The above and other objects, features and advantages of the present invention will become more apparent from the following detailed description which proceeds with reference to the accompanying drawings. [Brief description of the drawings]

[0010] [Figure 1] FIG. 1 is a diagram of an exemplary apparatus in which the disclosed techniques can be implemented. [Diagram 2] FIG. 1 illustrates a sample mounting configuration for some embodiments of the disclosed technology. [Diagram 3] FIG. 1 is a diagram of an exemplary beam and sample configuration illustrating the disclosed technique. [Figure 4] FIG. 1 is a diagram of pattern milling in an example of the disclosed technology. [Diagram 5] 1 is a set of images illustrating a first exemplary application of the disclosed technology. [Figure 6] 1 is a flowchart depicting an exemplary method in accordance with the disclosed technology. [Figure 7] 1 is a flow chart of a focus stacking method that may be used with the disclosed technology. [Figure 8]1A-1C are diagrams illustrating generation of a focused image and a depth map in an example of the disclosed technology. [Figure 9] 11 is a set of images illustrating focus stacking in a second exemplary application of the disclosed technology. [Figure 10A] 10B, a set of images illustrating focus stacking in a third exemplary application of the disclosed technology is provided. [Figure 10B] 10A provides a set of images illustrating focus stacking in a third exemplary application of the disclosed techniques. [Figure 11A] 11B and 11C provide a set of images illustrating a fourth exemplary application of the disclosed technology. [Figure 11B] 11A and 11C provide a set of images illustrating a fourth exemplary application of the disclosed technology. [Figure 11C] 11A and 11B provide a set of images illustrating a fourth exemplary application of the disclosed technology. [Figure 12] 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 described. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] Introduction Transmission electron microscope TEM (Transmission Electric Microscopy) is increasingly being used for the analysis of biological samples, as well as material science samples and semiconductor devices. Cryo-electron microscopy in particular allows for structure determination without the need for crystalline structures.

[0012] Throughput can be an important consideration in many applications. Often, multiple sample sites can be prepared on a single substrate. Sample sites can be prepared as thin lamellae, for example, using a FIB tool. Positioning the substrate at an oblique angle can bring more sample sites within the limited field of view of the FIB column. To illustrate, orienting the substrate surface at a 10° grazing angle (i.e., 80° incidence angle) to the FIB beam axis can increase the sample area in the FIB field of view by approximately 1 / sin(10°)=5.75 compared to normal incidence. Grazing incidence FIB offers another advantage in that sample material above and below the lamella can be milled away, allowing the lamella to be used while still attached to the original substrate, for example, without having to be separated and reattached to a separate lamella holder.

[0013] Therefore, it may be desirable to identify features and sample sites as viewed along the FIB beam axis. FIB tools used for milling (e.g., at high beam currents) can also be used for imaging (e.g., at low beam currents). However, the focal depth of the FIB beam may be limited to approximately 20-50 μm, which may be smaller than the dimensions of a single 100 μm sample site, let alone a typical 3 mm diameter grid plate.

[0014] To avoid issues with depth of focus, some workflows use normal incidence SEM imaging of the sample. Surface features can be easily identified from the SEM image, but correlating feature positions from the SEM image to the FIB viewpoint can be difficult for reasons including geometric correction for topographical variations (e.g., "bumpy surfaces"), contrast or edge diffraction variations between the SEM and FIB, and time-dependent variations in beam steering. The time taken for image correlation between the SEM and FIB, and the propensity for errors, can directly reduce the achievable throughput.

[0015] The disclosed technique applies focus stacking to allow the FIB image to be used directly for feature identification, even at grazing incidence. Thus, the same tool can be used for imaging and subsequent milling to prepare the lamella. Imaging can also be performed during the milling operation to track and compensate for beam drift. Furthermore, focus stacking can provide a depth map as an auxiliary output, so that the working depth at or near the identified feature can be precisely known and the FIB milling can be set accordingly. Since variations in beam focal depth are related to variations in beam spot size, precise control of the FIB mill can directly provide tight tolerances on the resulting lamella thickness, which may be the critical dimension in TEM applications.

[0016] In other applications, features can be identified within the interior volume of a sample, for example, by fluorescence imaging of fluorophores attached to proteins or other biomolecules. Like FIB, fluorescence microscopy may also have a limited focal depth, and complete imaging of a sample volume may require a stack of 10-100 fluorescent images at different focal depths. Some workflows allow manual scanning between these images, for example using a slider on an interactive display to select or change an image slice, but cannot show features at different depths simultaneously.

[0017] The disclosed technique applies focus stacking to generate a composite image that includes in-focus features from a wide range of depths. The ability to simultaneously visualize features at different depths can provide an informed assessment of features with respect to their environment or with respect to the larger object of which they are a part. In particular, lateral distances in the composite image, or depth values ​​in a companion depth map, can provide clues regarding the orientation of the larger object or feature. With this information, a user or an automated program can make an informed selection of those features that are suitably oriented for TEM imaging or lamella preparation. Furthermore, the orientation information can guide the selection of lamella orientation such that the selected features have a desired orientation for TEM analysis within the lamella. The selected lamella orientation can then guide the placement of the milling pattern to form the lamella. The three-dimensional feature and lamella placement information can be converted from fluorescent beam coordinates (or sample coordinates) to FIB coordinates for milling using trigonometry or matrix operations.

[0018] Focus stacking and depth map generation can be performed by automated software and can be easily integrated with preceding (e.g., image acquisition) and subsequent (e.g., feature selection or pattern milling) steps in an automated or semi-automated workflow. Focus stacking and depth map generation are fast procedures, typically taking 1-10 ms or 10-100 ms on a workstation or instrument controller, depending on the number and size of the individual images.

[0019] term The usage and meaning of all terms cited in this section apply throughout this disclosure unless expressly indicated otherwise or contrary to context. The following terms are expanded to the related word form. For convenience, the terms in this section are organized into the following subsections: (A) General, (B) Geometry, (C) Imaging, (D) Samples, and (E) Tools and Techniques for Sample Analysis.

[0020] A.General A "controller" is an electronic device coupled to one or more actuators for effecting a change in a physical parameter or coupled to one or more sensors for monitoring a physical parameter. Some controllers may include a microprocessor that can be programmed to execute machine-readable instructions. The description of a computing device herein is generally applicable to such controllers. Such controllers may include additional electronic circuitry, such as filters and amplifiers. Some controllers may include analog circuitry, such as filters and amplifiers, without the use of a microprocessor.

[0021] A "parameter" is a quantity that can have a particular "value." Often, parameters have a numerical value, but this is not a requirement, and some parameter values ​​can be logical values, strings, or data structures. Some parameters can be control parameters, e.g., set by a controller to affect a physical device or physical operation. Other parameters can be sensed parameters, e.g., determined by the physical environment and evaluated by a controller or a measurement device (e.g., to form an image or from an image).

[0022] "Software" refers to computer executable programs, instructions, or associated data structures. Software may be active or stationary. In an active state, the software may be loaded into memory or undergo execution by one or more processors. In a stationary state, the software may be stored on a computer readable medium and await transmission or execution. "Instructions" (or "program instructions") encode (sometimes together with one or more operands, operand addresses, or destination addresses) an operation to be performed by a processor. A group 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 presented information. "Automated software" refers to a software program that receives information from an instrument or other software program and selects one or more actions to be performed by the instrument or other software program without user input.

[0023] A "user" is a person. An exemplary user may control an analytical procedure or instrument, or may receive data presentations or notifications from such an analytical instrument. A user may interact with an instrument, sample, or software with the aid of a communication network, a computer input or output interface, or client software.

[0024] B. Geometry "Depth" refers to a coordinate in a direction along the beam, with depth values ​​increasing with distance from the beam source. Thus, for example, in an electron microscope where the sample is viewed from directly above and a multi-beam system with a FIB oriented at about 50° from each electron beam, each beam has its own depth coordinate. "Working depth" is the depth at or near which the beam is expected to interact with the sample. Some tools have beam controls to set the working depth, which may correspond to the depth at which the beam is focused or has a minimum cross-sectional area, e.g., "beam waist". "Focal depth" is the depth at which the beam has an optimal focus, which may be the depth at which the lateral extent of the beam is smallest, and is often the working depth. "Focal depth" is the size of the range at which the beam is focused according to a given criterion. To illustrate, a focal depth of 30 μm at a focal depth of 100 μm can be understood as a beam with an acceptable focus of 85 μm to 115 μm.

[0025] "Height" refers to a coordinate in a direction along a normal from a major surface of the sample. The sample can often be placed so that a major surface is in contact with a stage, with height values ​​increasing from the contact surface through the sample toward and beyond the other major surface.

[0026] "Height variation" refers to the difference between the maximum and minimum height of a feature, structure, or surface of a sample. The height variation may be across the entire sample, across the surface of the sample, across structures or features present in or on the sample, across an imaged area of ​​the sample, or across some other defined area of ​​interest. Some samples of interest herein may have surface height variations in the range of 1-2, 2-5, 5-10, or 10-20 μm over a surface lateral extent of 2, 5, 10, 20, 50, or 100 μm.

[0027] A "major surface" of a sample is a surface of the sample whose area is not substantially greater than any other surface of the sample. For ease of explanation, the sample is considered to have a top major surface and a bottom major surface, the bottom surface being supported on a stage and the top surface being exposed to one or more tools or beams. In an embodiment, a beam may be incident on the sample from below. Desirably, the major surface need not follow the detailed topographical variations of the sample surface, but may be a similar surface such as a smoothed surface derived from the actual surface, the surface of the sample's substrate, the backside of the sample, or the surface of a stage in contact with the sample, any of which may be free of topographical variations.

[0028] A "region" is a contiguous two-dimensional portion of a sample surface or a contiguous three-dimensional portion of a sample volume.

[0029] A "region of interest" (ROI) is a portion of a sample that contains a feature or structure that is the subject of subsequent analysis. An ROI can be defined relative to a surface of the sample (e.g., an area of ​​an imaged surface that contains a particular feature, or an area into which a pattern can be milled to form a lamella containing such a feature), or relative to a volume (e.g., a volume that contains a structure to be further analyzed). The term ROI does not refer to any human interest.

[0030] A "surface normal" is an outward-facing normal to a surface of a sample. The surface normal may apply to a major surface of the sample, although this is not a requirement, and the surface normal may also exist on a cut surface or a topographical feature of the sample.

[0031] The term "lateral" can be used with respect to a sample or a beam. In the former case, the term "lateral" refers to a direction in or parallel to the main surface of the object. In the latter case, lateral refers to a direction in a plane perpendicular to the direction of beam propagation. If not clear from the context, the term may be appropriately modified to indicate whether a sample or a beam is being referred to. For example, a "surface lateral extent" is the extent along the surface of the sample, and an "image lateral extent" is the extent in the image plane.

[0032] Terms such as "top", "bottom", "up", "down", "upper", "lower", "horizontal", "vertical", and the like are used for convenience with respect to the common configuration in which the exposed major surface of a horizontal sample has an outward normal pointing upward, e.g., the sample is accessible from above to a process tool. The axis of the process tool (e.g., the viewing axis of an imaging tool, or the beam axis of a FIB mill) may be approximately perpendicular downward to the surface, or may be tilted at an angle (sometimes ranging from 40° to 60°) relative to the normal. Those skilled in the art will appreciate from this disclosure that the actual orientation choice may be varied without departing from the scope of the disclosed technology.

[0033] C. Imaging The term "imaging" refers to a process for obtaining a two-dimensional representation (called an "image") of parameter values ​​across a region of interest of a sample. In some embodiments, the imaged parameter may be backscattered or secondary emission from an incident particle beam (e.g., with an SEM or low current focused ion beam), although this is not a requirement and visible light, or other parameters, may also be imaged. In disclosed embodiments, the image may show objects (e.g., cells, proteins, fluorophores, nanostructures, or fiducials) on or within the sample, or features (e.g., edges) of such objects.

[0034] A "synthetic image" is an image formed by combining information from multiple input images (e.g., by focus stacking). The synthetic image may provide more information than any one of the input images, or it may provide less erroneous information than the input images.

[0035] A "depth map" is a data structure, image, or software function that indicates depth values ​​as a function of lateral position of an imaging beam. A depth map can provide a depth value for each location (e.g., each pixel) of a focal zone of a composite image obtained by focus stacking. The depth values ​​of the depth map can be converted, for example, using trigonometry, to height values ​​of corresponding locations on the imaged surface of the sample, and the resulting height values ​​can be organized as a data structure, image, or software function that provides a "height map" of height values ​​as a function of lateral position on or within the sample.

[0036] "Focus stacking" refers to an image processing technique for combining individual images of a sample to obtain a composite image. Each individual image can have a respective focal depth and a respective zone of the image that is in focus. The composite image can have a larger zone of focus than any of the individual images. Some focus stacking techniques can also generate a depth map. Focus stacking can be performed using multi-scale transforms such as the Laplacian pyramid technique. Focus stacking is a form of image fusion. The term "focus stacking" can also be used as a verb to perform such techniques.

[0037] An "image pyramid" is a sequence of images obtained by multiple iterations of blurring (e.g., low-pass filtering) and downsampling. If at each iteration, a low-pass image is stored, the pyramid can be called a "Gaussian pyramid." If, except for the last iteration, a high-pass complement to the low-pass image is stored, then the pyramid can be called a "Laplacian pyramid." Laplacian pyramids may be preferred for applications such as focus stacking because they involve higher compression ratios than Gaussian pyramids and provide direct information present at each level of the pyramid (e.g., at each successive length scale).

[0038] An "imaging device" (sometimes referred to as an "imaging tool") is a device capable of imaging. In the disclosed embodiments, an SEM or FIB can be used as the imaging device. Some imaging devices can use a response to an incident beam (e.g., secondary emission, backscattering, or fluorescence) as the measured parameter from which an image is formed. Imaging devices that use an incident electron beam or ion beam can be referred to as an "electron microscope" (e.g., SEM) or an "ion microscope" (e.g., FIB), respectively. Imaging devices in which the measured parameter includes the intensity, wavelength, or another property of light can be referred to as an "optical microscope." Some optical microscopes use an incident light beam (e.g., for fluorescence), but this is not a requirement and other beams, or diffuse or ambient illumination, can also be used. Various detectors can be used to measure particles or radiation emanating from the imaged sample, including secondary electron detectors, secondary ion detectors, photodetectors, or spectrometers.

[0039] A "multi-scale transformation" of a multi-dimensional dataset is a decomposition of the dataset according to the scale length (range) of features represented in the dataset. For example, fewer pixels with coarse granularity may be sufficient to represent coarse-scale features, while more pixels with fine granularity may be needed for fine-scale features. Multi-scale decomposition can allow features at different scales to be processed efficiently. Examples of multi-scale transformations include wavelet transforms, Fourier transforms, and various pyramidal decompositions.

[0040] The "viewing axis" of an image is the direction in which the sample is viewed in the image. When imaging with a beam incident on the sample, the viewing axis is the axis of the incident beam.

[0041] A "zone" is a contiguous portion of an image.

[0042] D. Sample A "sample" is a physical object on which imaging, milling, or other analytical or manufacturing procedures are performed. Typical samples can incorporate biological samples, multi-layered electrical or electronic structures, or other material structures. Biological can include cells, proteins, DNA, RNA, viruses, or fragments thereof. Electrical or electronic samples can include logic gates, memory cells, optoelectronic devices, passive components (including interconnects), semiconductor devices such as transistors, and can be in various stages of fabrication or decomposition. Material samples can include microelectromechanical systems (MEMS) or nanostructures. The disclosed techniques can be applied during sample preparation, during sample characterization, or during fabrication. For example, the disclosed techniques can be applied to a sample to form one or more additional samples, for example, in the form of a lamella for TEM analysis. Some biological samples can be stabilized at cryogenic temperatures (variously below 120K, below 80K, below 20K, or below 10K), for example, by plunge freezing. A sample may include a "substrate" on which a cell, structure, or device is deposited or fabricated prior to lamella preparation or other analysis.

[0043] A "cut surface" is the surface of the sample exposed by etching.

[0044] A "feature" is a structure or other variation in the material composition of a sample that is discernible in an image of the sample. Exemplary features in a biological sample can include cells, binding sites (e.g., Covid virus spikes), mitochondria or other organelles, or fluorophores or other labels applied to biomolecules. Exemplary features in a semiconductor device sample can include components of a memory cell, transistor, or metallization pattern. Other features can include fiducial structures incorporated into the sample.

[0045] A "lamella" is a thin sample imaged or intended to be imaged in a transmission electron microscope (TEM). A lamella can have a thickness in the range of 10-20 nm, 20-30 nm, 8-50 nm, or 5-200 nm.

[0046] A "nanostructure" is a structure having a range of 1 nm to 1 μm in at least two orthogonal directions.

[0047] A "reference structure" (sometimes called a "fiducial") is a structure fabricated on a sample for use in determining the position of another structure or device within the sample. A reference structure is visible in an image of the structure and can be used as a reference for positioning subsequent movements or for measuring coordinates, distances, or angles within the image.

[0048] A "semiconductor device" is a device that incorporates one or more semiconductor regions (e.g., doped or undoped silicon, germanium, III-V semiconductors, or similar materials) configured to perform one or more electronic functions. A semiconductor device may be a finished product, a portion thereof, or a product at an intermediate stage of manufacture. Semiconductor devices generally 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 circuitry may include both conductive wiring and logic gates.

[0049] E. Tools and Techniques for Sample Analysis "Analysis" refers to operations used to characterize a sample and can 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 can include imaging (e.g., FIB, optical, or electron microscopy), etching (e.g., ion milling), stripping, electron backscattering analysis, electron microscopy, mass spectrometry, material analysis, metrology, nano-probing, spectroscopy, or surface preparation. Equipment or instrumentation for performing such operations is referred to as "analytical equipment," "analytical device," or "tool." Non-limiting examples of tools that can be used in conjunction with the disclosed technology include mills, etchers, electron microscopes, electron spectrometers, optical microscopes, or optical spectrometers. In particular, some tools of interest herein incorporate a plasma focused ion beam (PFIB), a scanning electron microscope (SEM), or a fluorescence microscope.

[0050] A "beam" is a directed stream of particles or energy. Typical beams of interest in this disclosure are particle beams, such as electron beams or ion beams (including plasma focused ion beams), or light beams, such as laser beams for fluorescence excitation. A beam may have a finite extent transverse to the primary longitudinal direction of the flow. The line joining the centers of gravity of two or more cross sections of the beam is the "axis" of the beam.

[0051] An "electron beam" is a directed stream of electrons.

[0052] An "electron microscope" is a type of analytical instrument that irradiates a sample with an electron beam and uses the resulting particles or electromagnetic radiation to form a spatially resolved image. A scanning electron microscope (SEM) images a sample surface based on reflected, secondary, or backscattered particles or radiation from one or more surfaces of the sample. Since the beam interactions detected by the SEM occur at or near this surface, the SEM can operate with samples of any thickness. In contrast, a transmission electron microscope (TEM) images the sample volume based on transmitted electrons (including scattered electrons). TEMs operate with samples of about 10-150 nm thickness. The sample can be mounted on a grid for mechanical support and thermal conductivity, which in turn can be held in a sample holder. TEMs can provide resolutions of less than 50 pm, while SEM resolution is often limited to about 0.5 nm. In terms of magnification, a TEM can provide magnifications of 50 million times or more, while SEM magnification is typically limited to about 2 million times. In this disclosure, a scanning transmission electron microscope (STEM), which performs transmitted electron imaging, is considered to be both an SEM and a TEM. The electron beam in an electron microscope is generated in an electron gun and can be accelerated, focused, or steered through a series of stages toward a sample chamber. The disclosed techniques can be used to form one or more lamellae for subsequent TEM analysis.

[0053] "Etching" generally refers to a subtractive process for removing material from a sample. Etching can expose one or more "cuts" of the sample by removing material to one side of the cut. Non-limiting examples of etching techniques include laser etching, plasma etching, ion beam milling, and gas-assisted milling. Although etching removes material, the etching process can include additional preliminary or intermediate operations. For example, photoresist can be deposited on the sample and patterned to obtain a patterned etch in a subsequent operation.

[0054] A "focused ion beam" ("FIB") is an ion beam whose focus can be controlled to direct the beam to a spot on the surface, or which can be swept over a sample in a sweeping pattern. FIBs can be used for analysis, deposition, or removal of material at the incident spot. Some FIBs are used for milling. Typically, FIBs contain positive elemental ions such as Xe+ or Ga+, but this is not a requirement. Ion beam species such as Ga+ can be supplied, for example, from a liquid metal ion source (LMIS), while other ion beam species such as Xe+ can be generated in a plasma. A FIB generated in a plasma source can be referred to as a plasma focused ion beam (PFIB), and references herein to a FIB include PFIBs. Focused spot sizes ranging from 2 to 100 nm in diameter are achievable, and the depth of focus is typically 20 to 50 μm.

[0055] A "localized operation" is an analytical operation that is performed over a finite region that is proximate to or encompasses an identified feature of the sample. The region may have a predetermined spatial relationship to the identified feature and a lateral extent that is smaller than the corresponding lateral extent of the sample. The analytical operation may be a sample preparation or material removal operation, such as milling, etching, or polishing, or may be a diagnostic operation, such as imaging or spectroscopy.

[0056] "Milling" is a path-wise material removal process guided by the movement (sometimes called a "sweep") of a milling tool relative to the sample being milled. The milling operation can expose one or more "cutting surfaces" of the sample by removing material to one side of the cutting surface. In an embodiment, milling can be performed using a focused ion beam (FIB). Milling is a form of "etching", which more generally refers to a subtractive process for removing material from a sample. Diffusion chemical etching may not be considered milling. However, ion beam assisted chemical etching can be site specific at the location of the ion beam and can be considered milling. A "mill" (or "milling tool") is an apparatus capable of milling.

[0057] A "particle beam" is a beam that includes a directed stream of particles. A "particle" is a discrete, undivided unit of a flow of matter. Particles of common interest in this disclosure include, but are not limited to, charged particles such as electrons or ions (such as Ga, Xe, or protons). Electrons in an electron beam are considered a different species than ions in an ion beam, although some ions include bound electrons.

[0058] "Throughput" refers to the number of samples that can be successfully processed per unit of total time, including sample preparation, sample handling, instrument setup, and individual processing of the samples. Eliminating steps such as correlation between the SEM and FIB viewpoints can reduce the time for individual sample processing, and by improving the accuracy of lamella positioning, the yield can also be improved, thereby advantageously improving throughput in two aspects. Furthermore, synthetic focus stacking images can facilitate rapid feature selection (by interactive or automated software) and the accompanying depth map can provide accurate placement of lamella milling patterns with a concomitant improvement in throughput. As a general consideration, instrument setup and certain sample operations are performed in common for an entire microscope grid or substrate, and therefore providing more sample sites within the field of view of the analysis beam on a single grid or substrate (e.g., with a grazing incidence FIB) can also result in improved throughput.

[0059] Exemplary Analytical Device 1 is a diagram 100 of an exemplary apparatus capable of implementing the disclosed techniques. The apparatus can be used to perform imaging and milling operations on samples for lamella preparation or other applications. The described apparatus can be an analytical apparatus.

[0060] As will be described, the walls 110 define a chamber 152 in which analysis can be performed on the sample 120. The electron beam column 106 can operate as an SEM, the ion beam column 140 can operate as an FIB, and optionally, a light beam 150 can excite fluorescence in the sample 120. The sample 120 can be mounted on a stage 124 that can vary the position of the sample 120 in multiple degrees of freedom (translation or rotation) using respective actuators (not shown). The sample stage 124 can be a micromanipulator.

[0061] The vertically mounted SEM 106 comprises an electron source 102, a condenser lens coil 104, an objective lens coil 108, and may include additional beam controls such as a scanning coil and an aperture (not shown). If the scanning coil is not powered, the electron beam axis 130 may be used as the polar axis of the entire analytical apparatus described. Local coordinates may also be defined for individual subsystems or relative to the sample 120.

[0062] The FIB column 140 can be mounted at an angle, for example, such that the ion beam axis 132 is at an angle of 52° (or in the range of 40°-60°) from the electron beam axis 130. The FIB tool 140 can include an ion source 134, electrostatic or electromagnetic beam control 136, 138, and can include additional beam control elements not shown, such as apertures or deflectors for scanning.

[0063] The SEM 106 or FIB 140 can be controlled to set a desired working depth and further scan laterally. In general, the working depth can be set such that the instantaneous ion beam or electron beam is focused on the surface of the sample 120.

[0064] The sample stage 124 can be configured to position the major surface of the sample 120 at a shallow angle of incidence of about 154θ=10° (or in the range 5°-30°) for the oblique view described herein. The detector 128 can collect signals that can form an image of the sample 120. Explained, the detector 128 can receive secondary electrons or backscattered electrons or ions as the electron beam 130 or ion beam 132 is scanned across the sample 120. Exemplary secondary particle detectors include Everhart Thornley detectors and microchannel plates. In a variation, the secondary particle detector 128 can be annular about the beam axis 130 or 132. As another explanation, the detector 128 can be a camera used to record fluorescence from the sample 120. In some embodiments, multiple detectors 128 can be provided (e.g., a detector 128 for fluorescence and a detector 128 for receiving electrons or ions). Other tools can also be provided in the described apparatus, generally represented as a probe 126. In some applications, a gas delivery tool can be included.

[0065] The controller 114 may control or monitor signals from the detector 128 or the tool 126, as shown. The same or a different controller may similarly control or monitor signals from the SEM 106, the FIB 140, or the stage 124. The controller 114 may be equipped with a memory 112 for storing programs executed by the controller 114, and further for storing data input to or output from such programs.

[0066] The pump 116 pumps the chamber 152 at about 10 -9 ~10 -4 The pressure in the vacuum chamber may be maintained at 100 MPa (200 MPa) and any exhaust gases may be exhausted as shown by arrow 118. A combination of turbomolecular, mechanical, or ion pumps may be used.

[0067] The apparatus of Figure 1 may have various additional auxiliary components not shown, such as support structures, low or high voltage power supplies, control electronics, or instrumentation. This apparatus is shown for illustrative purposes only. The disclosed techniques may be readily applied to other apparatuses having more, fewer, or different components.

[0068] Exemplary Mounting 2 is a diagram 200 illustrating a mounting configuration for a biological cell sample. In this configuration, the cells are provided with mechanical support that minimizes interference with any analysis beams that may be applied to the cells.

[0069] The grid 210 provides overall mechanical support and can be handled manually or with robotic laboratory tools. The grid 210 can be placed directly on a stage such as 124 in FIG. 1. As described, the grid 210 has an overall diameter of about 3 mm and a 250 mesh pattern of gold wires forming about 100 μm square openings 212 in a solid ring 214. Other grid patterns and sizes can be used. Other materials such as copper, gold, nickel, rhodium, or composites thereof can also be used. The wire pattern can block about 20-50% of the area inside the ring 214, leaving a substantial remaining area unblocked and available for TEM imaging.

[0070] The grid 210 can be covered with a thin sheet 220 to provide cell support over the openings 212. As will be described, the sheet 220 is a continuous carbon film about 20 nm thick, strong enough to support a cell sample across the 100 μm grid openings 212, but thin enough to allow TEM imaging through the sheet 220. Other thicknesses (e.g., 10-100 nm) and materials can also be used. Perforated films, sometimes referred to as "holey" films, can also be used.

[0071] The cells 230 can be deposited on the film 220, either intentionally placed within each grid opening 212 or dispersed on the film 220. The cells 230 have variable shapes, typically with dimensions of about 2-5 μm. Even without intentional positioning, many cells can be suitably positioned for TEM imaging over the grid openings 212, as beam blockage by the grid wires is relatively low. The areal density of the cells 230 can be selected based on cell size and application requirements. To improve throughput, multiple TEM samples can be prepared on a single grid opening 212.

[0072] Exemplary Beam and Sample Configurations Figure 3 is a diagram 300 of an example beam and sample configuration illustrating the applicability of the disclosed techniques. Figure 3 shows how factors such as a grazing incidence ion beam, topographical features of the sample surface, and ion beam focal depth can combine to adversely affect lamella milling precision, thereby adversely affecting yield and throughput.

[0073] As shown, a cell 330 is supported on a film substrate 320. A FIB is directed towards the cell 330 as indicated by arrow 351. The FIB has a beam axis 354 and a transverse reference plane 356 can serve as a datum for a depth value (e.g., Z=0). The cell 330 forms a topographical feature on the substrate 320 such that the true depth 361 can be significantly less than the expected depth 371 in the absence of height variations.

[0074] The disclosed technique allows for accurate determination of depth 361 through focus stacking of a series of images over a range of working depths Z.

[0075] With precise depth 361, milling can be precisely configured as shown by dotted beam profile 363, for example with an offset equal to the beam waist radius, to precisely mill cell 330 starting at point A 368 or any other desired starting location.

[0076] For example, in existing approaches using dynamic focusing, the depth may be estimated based on a known angle of incidence 358 relative to the surface normal 327 of the substrate 320. The resulting beam pattern 373, having a waist at point B 378, may be significantly in error in actual depth 361.

[0077] 3 is illustrative only: an actual lamella milling pattern may begin away from this point rather than at point A 338. Nevertheless, the principles are still applicable and precise knowledge of depth 361 with the disclosed techniques can directly improve milling precision, lamella yield, and overall throughput.

[0078] Although the illustration in Figure 3 is not to scale and in some respects exaggerated, it should be noted that the tolerance requirements for lamellae with thicknesses of 10-200 nm are very tight and even an error of 1 µm in depth for a FIB with a focal depth of 20 µm can lead to an error of 10-100 nm in the lateral position of the milled edge, which can compromise successful fabrication of the lamellae.

[0079] Also shown in Figure 3 is a set of planes 381 at respective working depths Z. In an example of the disclosed technology, a series of images may be acquired at each of the planes 381 and used as input to a focus stacking procedure as described herein. The planes 381 are shown offset from the substrate 320 and cells 330 for clarity of illustration. In practice, the images in these planes may be extended to include the substrate 320 and cells 330 in the field of view.

[0080] Exemplary Pattern Milling 4 is a photograph 400 of an artist's model illustrating pattern milling of cells according to the disclosed technique. Photograph 400 shows cells 430 immobilized on a perforated film support 420. Photograph 400 has a viewing axis tilted approximately 20° from the surface normal of film 420.

[0081] As can be seen in Fig. 4, lamella 457 has been prepared containing a slice of cell nucleus 432. The dark areas 453, 455 are FIB milled patterns. The FIB beam axis is represented by arrow 451. Pattern 453 is milled at the location shown and below lamella 457. Pattern 455 is milled above lamella 457. Both patterns 453, 455 penetrate the cell 430 and the support film 420.

[0082] First Exemplary Application Figure 5 is a set of images 501-505 showing a first exemplary application of the disclosed techniques in a cryo-lamella preparation workflow. First, cells can be deposited on a support structure (e.g., similar to that of Figure 2) and cryofixed, for example, by plunge freezing.

[0083] SEM image 501 is taken from above using an electron microscope in a configuration similar to that shown in Figure 1. Numerous cells 512 are discernible, and the background pattern 514 reflects the pattern of perforations in the support film similar to 220 in Figure 2.

[0084] FIB image 502 shows a substantially similar scene to image 501, but from an oblique viewing axis similar to the FIB and sample configuration of Figure 1. Cells 522, pore patterns 524, and impurities 526 are discernable.

[0085] The FIB image 503 is a copy of the image 502, on which a feature 531 (e.g. the centroid of a cell) has been selected for lamella preparation and two mill patterns 533, 535 have been positioned. Since the same FIB can be used for imaging and milling, correlation between the respective viewpoints of the different tools can be avoided and the placement of the mill patterns 533, 535 can be efficiently performed.

[0086] FIB image 504 shows the same scene as images 502-503 after partial milling of patterns 543, 545. Pattern 543 has been milled cleanly through the support film, but pattern 545 still has material beyond the milled depth. The bright outlines of patterns 543, 545 are imaging artifacts that indicate the location of the edges of milled patterns 543, 545.

[0087] Finally, FIB image 505 shows the same scene as previous images 502-504 after completion of milling and polishing. A thin lamella 557 (approximately 10 nm thick) remains between patterns 553, 555. Lamella 557 and patterns 553, 555 can be similar to lamella 457 and milled patterns 453, 455 described in the context of FIG.

[0088] Exemplary Methods 6 is a flowchart 600 depicting an exemplary method in accordance with the disclosed technology in which local actions are performed on a sample based on output from a focus stacking technique applied to images of the sample at various focal depths.

[0089] The first method begins at process block 610, where multiple images of the sample can be acquired at each focal depth. In an embodiment, the images can be acquired using a FIB or a fluorescent microscope, although this is not a requirement and other imaging modalities can be used. The images can be acquired over a predefined set of working depths, and can be acquired sequentially (e.g., 0, 20 μm, 40 μm, 60 μm, ... for any starting depth) or in an irregular depth scan (e.g., 0, 60 μm, 80 μm, 20 μm, 100 μm, ...). In some embodiments, an initial image can be acquired using an autofocus mode of the imaging device, and successive images are acquired with subsequent adjustments of the working depth from the autofocus configuration. To mitigate beam drift or other temporal variations, multiple images can be acquired at each working depth setting and combined, for example, by registration and averaging of one or more criteria.

[0090] At block 620, focus stacking may be performed on the images acquired at block 610 to generate a composite image 630 that may have a range of focus depths. To illustrate, if the individual images are acquired with working depths ranging from z1 to z2, the composite image 630 may have features that are in focus at all depths [z1,z2]. The bracket notation [z1,z2] represents a closed interval that includes both endpoints z1 and z2. Illustrative examples of focus stacking are described further herein.

[0091] The first method may proceed to process block 640 where an indication of location selection may be received. In various embodiments, the location selection in block 640 may be performed interactively or by automated software. In the former case, a user may use a mouse or similar pointing device to select a point or zone of the composite image 630 on the display, thereby selecting a corresponding location on the physical sample. In the latter case, the automated software may apply machine vision techniques to identify edges, shapes, or objects on the composite image. The selected location may correspond to an identifiable feature in the composite image, which may be a representation of an object (such as a cell, fluorophore, structure, or pattern shape), or an attribute of such an object (e.g., an edge, a centroid, an extremum in image intensity, or a vertex of a bounding polygon).

[0092] The first method can follow arrow 645 from block 640 to block 670 and perform a local operation based on the selected location. A wide range of operations can be performed at the selected location, over a region encompassing the selected location, at a location with a predefined offset from the selected location, or over a region with a predefined spatial relationship to the corresponding location. The predefined offset can be defined with reference to beam coordinates, e.g., a lateral offset from a line of sight passing through the selected location, or with reference to sample coordinates, e.g., a lateral offset from a surface normal intersecting the selected location. Non-limiting examples of such operations can include, for example, milled indentations, crosses, or other markers, e.g., pattern milling for lamella formation, other etching operations, electron or optical spectroscopy, or other metrology operations. The local operation can be part of a workflow for manufacturing a lamella incorporating the selected location. The local operation can also include subsequent TEM analysis, or other microscopy, on the lamella prepared at the selected location. The localized operation may be performed by any suitable tool, non-limiting examples of which may include a mill, an etcher, an electron microscope, an electron spectrometer, an optical microscope, or an optical spectrometer.

[0093] A second method may also be described with reference to FIG. 6. An extension to the first method is shown in FIG. 6 with dashed lines. The second method may follow the first method from blocks 610 to 640. However, in the second method, the focus compounding in block 620 may also generate a depth map 650 that accompanies the composite image 630. In particular, the depth map 650 may provide a depth value for any pixel of the composite image 630. In some embodiments, the depth map 650 may be represented as an image such that every first pixel in the composite image 630 has a corresponding second pixel in the depth map 650 whose value (pixel intensity) represents the depth to the sample surface at the location of that first pixel. The second pixels of the depth map image 650 may have a 1:1 relationship to the first pixels of the composite image 630, although this is not a requirement and the depth map 650 may be downsampled to a smaller representation having fewer pixels than the composite image 630. Additionally, other representations of the depth map 630 can be used, such as a contour map or a look-up table. Because the granularity of the depth values ​​can be coarser than the granularity of the pixels in the composite image 630, run-length encoding or another compression scheme can advantageously reduce the size of the depth map 650. In some embodiments, the depth map 650 can store depth values ​​relative to an observation plane that is perpendicular to the viewing axis of the imaging beam (as described with respect to depths 361, 371 in FIG. 3), while in other embodiments, the depth values ​​in the map 650 can be relative to a reference plane that is generally parallel to the "top" major surface of the sample. In the latter case, a perfectly smooth surface would have uniform depth values ​​across the entire field of view, with variations in depth reflecting topographical variations in the sample surface.

[0094] Thus, the second method may use the selected location from block 640 and look up a corresponding depth value from the depth map 650. In block 660, this depth value may be used to set a working depth of a tool to perform a local operation 670. The second method then proceeds to block 670 where a local operation may be performed similar to the first method.

[0095] In some examples, the working depth may be equal to the depth value at the selected location, although this is not a requirement and one or more offsets may be applied. To illustrate, a FIB with a focal depth of 40 μm set to a working depth of Z μm may provide milling with a tolerance of Z-20 μm to Z+20 μm. Thus, if the depth value to the sample surface is 80 μm, milling may be started with the working depth set to 100 μm, thereby requiring fewer adjustments of the FIB as milling progresses. As another illustration, for example, as illustrated in FIG. 5, may be performed at a lateral displacement of 10 μm from the selected location. Thus, the depth value may be read from the depth map 650 at an offset from the selected location, rather than at the selected location itself. Furthermore, the four corners of the pattern mill may have depths of 60, 80, 90, 70 μm, which may be averaged together to set the working depth to an average of (60+80+90+70) / 4=75 μm. Multiple such offsets may be applied together. The offset, use of a diagonal reference plane, or other scaling may be implemented by applying a linear transformation to the depth values ​​stored in depth map 650.

[0096] Numerous variations and extensions can be made within the scope of the disclosed technology, some of which are described above in the context of the first and second methods, some of which are described immediately below, and others of which are described elsewhere in this specification.

[0097] In an embodiment, the automated selection of locations in the composite image can be performed by software trained by machine learning, for example, using a training dataset of images and selected feature locations.

[0098] In embodiments, the local operation may be performed using one or more beams, such as an ion beam, a plasma focused ion beam, an electron beam, or a light beam. Such beams may be directed at the sample along a beam axis having an angle of incidence in any of the following ranges: 70°-85°, 60°-87°, 0°-15°, 15°-40°, or 40°-60°. In various embodiments, the local operation may be performed at a selected location, at a predetermined lateral offset from the selected location, at a predetermined depth offset from the selected location, over a region that includes the selected location, or over a region that has a lateral or depth offset from the selected location. The lateral offset may be either a beam coordinate or a sample coordinate. The working depth set in block 660 may be dynamically adjusted as the local operation is performed in block 670, with or without additional sample imaging.

[0099] The image may be acquired by or from an imaging device, which may include an ion microscope, an electron microscope, an optical microscope, a secondary electron detector, a secondary ion detector, or a photodetector. The imaging device may illuminate the sample using one or more beams, such as an ion beam, a plasma focused ion beam, an electron beam, or a light beam. Such beams may be directed at the sample along a beam axis having an angle of incidence in any of the following ranges: 70°-85°, 60°-87°, 0°-15°, 15°-40°, or 40°-60°.

[0100] The sample of interest may include a cryogenic sample, a plunge frozen sample, a biological sample, a semiconductor sample, or a nanostructure.

[0101] Exemplary Focus Stacking FIG. 7 is a flowchart 700 of a focus stacking method that may be used with the disclosed technology. The method begins with a set of source images focused at different depths. For ease of explanation, the term "slice" may be used to refer to each imaged working depth, the corresponding source image, and associated processing. The set of source images is processed to obtain (i) a composite image in which features at the different depths are all in focus, and (ii) a depth map indicating the depth values ​​of the features in the composite image. Although FIG. 7 describes one exemplary method of focus stacking using a Laplacian pyramid technique, the use of this technique is not a requirement and other focus stacking methods may be used as part of the disclosed technology.

[0102] The described technique can be conceptually viewed as including input-side processing, intermediate processing, and output-side processing, all of which are represented in Figure 7. First, multiple source images 701 can be acquired at each working (focus) depth.

[0103] In general, the set of images 701 contains information in the form of features. The information may be distributed across three spatial dimensions (two in the plane of each image 701, the third being the depth across slices 770A-770N) and may have variable scale lengths that may not be known a priori. When images are analyzed at fine resolution, coarse grain features may appear as smoothly varying pixel intensities and coarse features may be missed. Conversely, when images are analyzed at coarse resolution, fine features may be lost due to smoothing. To overcome such problems, decomposition can separate information in the original images 701 according to length scales, and each length scale can be processed at an appropriate resolution. Intermediate results at various length scales can be combined to obtain one composite image and, optionally, a corresponding depth map.

[0104] A. Input side processing The input side processing performs the decomposition shown for the first slice in dashed outline 770A, with the other slices shown after slice 770A. That is, each slice 770A-770N has a respective source image 701 that can be decomposed into a set of images 711-716.

[0105] As illustrated, a low pass filter (LPF) 801 is applied to image 701 and the output, subtracted from image 701 by subtractor block 811, leaves 711 containing information in the highest spatial frequency band (called an "octave") of source image 701. Because low pass filtering removes high frequency components from source image 701, the output of LPF 801 can be decimated or downsampled by block 821 to generate a smaller image 702. Illustratively, a 1024x1024 source image 701 can be downsampled by 1 / 2 to obtain a 512x512 image 702.

[0106] This procedure can be applied iteratively to obtain successively lower frequency bands of information in high-pass images 711-715, and successively lower frequency versions of the source image 701 in downsampled images 702-706. Details of the repetition of several iterations are omitted, except to note that the LPF blocks 802-805, subtraction blocks 812-815, and downsampler blocks 822-825 operate similarly to the blocks 801, 811, 821 described above, but with smaller images. The final downsampled image 706 can be retained as image 716, which contains all the residual low frequency information. In the described embodiment, the decomposition can produce images of size 1024x1024 (711), 512x512, 256x256, 128x128, 64x64, and 32x32 (716).

[0107] By suitable choice of cutoff spatial frequency for LPF 801, any residual high-pass information at its output may be insignificant and the downsampling in block 821 may be substantially lossless, which means that the decomposition of source image 701 into image pair 711, 702 may also be lossless, as well as for subsequent stages of decomposition. That is, images 711-716 can be combined (with suitable upsampling) to reconstruct source image 701 with high accuracy.

[0108] This decomposition with five stages of 1 / 2 downsampling is merely illustrative. Fewer or more stages can be used. Downsampling by smaller or larger factors can be used. Different stages can use different downsampling factors.

[0109] The decomposition described may be replicated across all slices 770A-770N to complete input side processing.

[0110] B. Intermediate Processing Next, we describe intermediate processing, which performs slice-to-slice fusion. While the input side processing was organized by slice (depth), the intermediate processing can be organized by scale length or spatial frequency band. That is, all slice 711 images are processed together, and similarly for image 712, image 713, etc.

[0111] Dashed line 793 illustrates intermediate processing of image 713. Within each image 713, the variance can be calculated at each pixel to obtain the variance matrix 723. As illustrated, there is one variance matrix 723 for each slice 770A-770N. The variance can be calculated, for example, over a 5×5 pixel zone of image 713 centered on the instant pixel. To illustrate, the variance for a pixel at (row, column)=(10,20) can be calculated over a zone of pixels in rows 8-12 and columns 18-22. Near the edge of image 713, the variance can be calculated over pixel locations that lie within the image. To illustrate, the variance for pixel (2,1) can be calculated over a zone of pixels in rows 1-4 and columns 1-3.

[0112] Then, at each pixel location, the variances across the slices can be compared and weights assigned accordingly. The slices 723 with the highest variance may have the most information and may be given the highest weight, while slices 723 with progressively lower variances may be assigned progressively lower weights. In some embodiments, the weights may be equal to the variance, while in other embodiments, other monotonic functions of the variance may be used, such as power laws, exponential functions, piecewise linear functions, or other functions. The weights may be stored in a weight matrix 733 having the same organization as the image 713 and the variance matrix 723. As will be described, there is one weight matrix 733 for each slice 770A-770N. Optionally, the weights may be normalized such that for any given pixel, the sum of the weights across all slices 733 is 1. Given the decomposed image 713 and weight matrix 733 as B and W, respectively, the weight W for an N×M pixel zone around pixel (r,c) may be calculated using the following formula, where M / 2 and N / 2 are the integer parts, respectively:

[0113]

number

[0114]

number

[0115] The covariance matrix 723 and weight matrix 733 can then be combined across slices. The subimage 743 can be calculated as a weighted sum across the decomposed image slices 713. Let E be the subimage 743, s, r, and c be the slice, pixel row, and pixel column, respectively, and assuming normalized weights,

[0116]

number

[0117] The partial depth map F763 is the depth Z of each slice. s can be calculated using a formula such as:

[0118]

number

[0119] In a variant, the index s is Z s can be used instead of , with a linear transformation from slice index to physical depth performed afterwards. Sub-images E (e.g. 741, 746) and sub-depth maps F (e.g. 761, 766) can be similarly computed for other scales from the corresponding decomposed images (e.g. 711, 716), as illustrated by processing modules 791, 796 shown in dashed outline.

[0120] To combine several sub-images 741-746 into a single composite image, an information matrix 753 can be derived and used to assign weights to pixels of sub-image 743. In some embodiments, an element of information matrix 753 may be the largest among the corresponding elements of covariance matrix 743 across all slices. Let V be the covariance matrix 743 and G be the information matrix 753, then an equation such as G(r,c)=max s (V s (r,c)

[0121] In other embodiments, other formulas can be used. As an illustration, the variance value V s (r,c) is Z s can be fitted to a Gaussian function (or another bell-shaped curve) and the peak of the fitted curve can be used as the information value G(r,c).

[0122] The information matrices for the other scales (eg, 751, 756) can be calculated similarly.

[0123] An intermediate processing stage is completed with the partial images E 741-746, the partial depth maps F 761-766, and weights for combining them in the form of partial information matrices G 751-756.

[0124] C. Output side processing The discussion continues with output-side processing, where the results of the intermediate processing can be fused across length scales to obtain one composite image and, optionally, its corresponding depth map.

[0125] The intermediate results are partially combined at each scale, as indicated by multiplier blocks 841-846, and then summed with suitable upsampling in block 848 to obtain the composite image 730. Let the subscript L denote some length scales of the intermediate processing modules 791-796 (e.g., the length scales of the decomposed images 711-716). Let (r',c') denote the upsampled pixel coordinates matching the source image 701, U denote the upsampling function from (r,c) to (r',c'), and H denote the composite image 730, and the overall calculation can be expressed by an equation such as:

[0126]

number

[0127] In this equation, the inner multiplication can be performed by multipliers 841 to 846 , and the upsampling and summation can be performed by block 848 .

[0128] A similar procedure can be used to generate the depth map 750. Denoting the depth map 750 as J, the overall calculation can be expressed by the following equation:

[0129]

number

[0130] All or part of the technique of Figure 7 may be used to implement focus synthesis in various methods and applications described herein, however, the disclosed techniques are not so limited and variations or extensions of Figure 7 may also be employed.

[0131] D. Explanation of focus stacking results 8 is a diagram illustrating the generation of a focused image and corresponding depth map, in which a set of three images are combined to obtain a composite image having a focused representation of the object from each of the three images and a depth map showing the depth of each of the objects. The annotations are provided for clarity of explanation and are not part of the image shown.

[0132] On the left, image set 800 includes images 881-883 acquired at working depths Z1, Z2, and Z3, respectively, as shown. The imaged scene includes three geometric shapes: a cross, a square, and a rectangle. For clarity, each shape is annotated with an "F" to indicate in focus, or an "X" to indicate out of focus. Thus, the rectangle, square, and cross are in focus at working depths Z1, Z2, and Z3, respectively. Focus stacking can be applied to image set 800 as described herein to obtain a composite image 830 and depth map 850. In composite image 830, all three shapes appear to be in focus, as indicated by the "F" annotation. Depth map 850 stores a depth Z1 corresponding to image 881, where the rectangle shape is in focus, as indicated by the annotation "Z1". Similarly, the stored depth values ​​"Z2", "Z3" for the square and cross indicate the depths at which these shapes are in focus in image set 800.

[0133] Second Exemplary Application 9 is a set of images 920, 930, 940 illustrating a second exemplary application of focus stacking. In this example, focus stacking is applied to a set of source images to generate a composite image that is used to place a pattern for milling a lamella.

[0134] Initially, image 920 is one of a set of source images as acquired by an imaging device at a particular working depth. As shown, the foreground is in focus while the background is out of focus. Rectangle 921 represents the other source images of the set, each focused to its respective working depth.

[0135] Through the application of focus stacking as described herein, a composite image 930 can be obtained in which all features, from foreground to background, appear to be in focus.

[0136] Image 940 is a copy of the composite image 930 marked with four FIB milling patterns 941-944. The execution of these milling patterns can form lamellae between the patterns 942, 944, as explained in the context of image 505 of FIG. 5 and FIG.

[0137] Third Exemplary Application 10A-10B together provide a set of images 1011-1017, 1030, and 1050 illustrating a third exemplary application of focus stacking. Images 1011-1017 are a set of FIB images for each working depth, with each zone being in focus. As a visual aid, exemplary zones 1021-1027 are marked in images 1011-1017, respectively, to show features that are in focus in some images. It can be seen that the in-focus zones 1021-1027 progressively recede from the foreground to the background as the working depth increases from images 1011 to 1017. Image 1011 has the lowest value of working depth among images 1011-1017, and in-focus zone 1021 is the most foreground of zones 1021-1027. It can further be seen that zone 1021 degrades in focus as one progresses from image 1012 to image 1017. Image 1017 has the greatest working depth and background zone 1027 is in focus. Zone 1027 is out of focus in image 1011.

[0138] Application of focus stacking to images 1011-1017 can produce a composite image 1030 and a depth map 1050. Composite image 1030 (included in both Figures 10A and 10B) shows all zones that are in focus, from foreground to background. Depth map 1050 shows the topographical relief of features visible in composite image 1030 relative to a reference plane. In general, cells protruding above the perforated film visible in image 1030 also have lower depth values ​​than those zones would have if the cells were not present (e.g., if only the perforated film were imaged). This depth reduction appears as lighter colored zones in depth map 1050. For visual effect, depth map 1050 uses a reference plane that is slightly tilted relative to the major surface of the bare support film, which accounts for the systematic shading variation from the darker foreground to the lighter background.

[0139] Fourth Exemplary Application 11A-11C together provide a set of images 1101-1120 and 1130 illustrating focus stacking applied to fluorescence imaging, which may be done in the context of beam 150 of FIG. 1 or in configurations as discussed elsewhere herein. Images 1101-1120 are a set of fluorescence images for each working depth, with each zone in focus. As a visual aid, exemplary zones 1142, 1145, 1149, 1151, and 1153 are shown in focus in each image in the set of images 1101-1120. These features can be seen to be out of focus in other images a few slices away from the marked image. To illustrate, zone 1142 is out of focus in image 1105, and zone 1153 is out of focus in image 1115.

[0140] Application of focus stacking to images 1101-1120 can produce a composite image 1130, where all zones are seen to be in focus.

[0141] Additional Examples The following numbered paragraphs describe additional embodiments of the disclosed technology.

[0142] Example 1 is an apparatus including a tool configured to perform a local action on a sample using a first beam, an imaging device configured to acquire a plurality of images of the sample using a second beam, and a controller configured to focus-combine the plurality of images to obtain a composite image, receive an indication of a selected location in the composite image, and cause the tool to perform a local action based on the selected location.

[0143] Example 2 includes the subject matter of example 1 and further specifies that the controller is further configured to generate a depth map from the plurality of images, use the depth map to determine a working depth from the selected location, and configure the tool according to the working depth for localized operation.

[0144] Example 3 includes the subject matter of example 1 or 2, wherein the controller is further configured to form a lamella incorporating the selected location based at least in part on the local motion.

[0145] Example 4 includes the subject matter of any of examples 1-3, wherein the tool includes one or more of a mill, an etcher, an electron microscope, an electron spectrometer, an ion microscope, an optical microscope, or an optical spectrometer.

[0146] Example 5 includes the subject matter of any of examples 1-4, wherein the localized operation includes one or more of milling, etching, microscopy, or spectroscopy.

[0147] Example 6 includes the subject matter of any of examples 1-5, wherein the first beam includes one or more of an ion beam, a plasma focused ion beam, an electron beam, or a light beam.

[0148] Example 7 includes the subject matter of any of Examples 1-6, wherein the first beam is directed at the sample along a beam axis oriented between 60° and 87° from a surface normal of the sample.

[0149] Example 8 includes the subject matter of any of examples 1-7, wherein the imaging device includes one or more of an ion microscope, an electron microscope, an optical microscope, a secondary electron detector, a secondary ion detector, or a photodetector.

[0150] Example 9 includes the subject matter of any of Examples 1-8, wherein the second beam is directed at the sample along a viewing axis oriented between 60° and 87° from a surface normal of the sample.

[0151] Example 10 includes the subject matter of any of examples 1-9, wherein the second beam includes one or more of an ion beam, a plasma focused ion beam, an electron beam, or a light beam.

[0152] Example 11 includes the subject matter of any of examples 1-10, wherein the first beam and the second beam are in a common beam column; or It is generated in a separate beam column.

[0153] Example 12 includes the subject matter of any of examples 1-11, wherein focus stacking is performed using one or more of a multi-scale transform, an image pyramid, a Gaussian image pyramid, a Laplacian image pyramid, or a wavelet transform.

[0154] Example 13 includes the subject matter of any of examples 1-12, wherein the location is selected by a user or by automated software.

[0155] Example 14 includes the subject matter of any of Examples 1-13, wherein the location corresponds to a feature within the sample.

[0156] Example 15 includes the subject matter of any of examples 1-14, wherein the localized operation is performed at the selected location, at a predetermined lateral offset from the selected location, at a predetermined depth offset from the selected location, over a region that includes the selected location, or over a region that has a lateral or depth offset from the selected location.

[0157] Example 16 includes the subject matter of any of Examples 2-15, and further specifies that the working depth is equal to a first value stored in the depth map for the selected location, a first average of values ​​stored in the depth map in a first zone that includes the selected location, a second value stored in the depth map at a predetermined lateral offset from the selected location, or a second average of values ​​stored in the depth map in a second zone having a lateral offset from the selected location, or a predetermined linear transformation thereof.

[0158] Example 17 is a computer-implemented method that includes acquiring a plurality of images of a sample at respective focal depths, focus-compositing the plurality of images to obtain a composite image, receiving an indication of a selected location in the composite image, and performing a local action on the sample based on the selected location.

[0159] Example 18 includes the subject matter of example 17, and further specifies that the focus synthesis further generates a depth map of the sample, and the method further includes using the depth map to determine a working depth from the selected location, setting a tool according to the working depth, and causing the tool to perform a localized operation.

[0160] Example 19 includes the subject matter of any of examples 17-18, further including forming a lamella incorporating the selected location, where forming includes performing a localized action.

[0161] Example 20 includes the subject matter of any of Examples 17-19, further specifying that the image is acquired using one or more of an ion microscope, an electron microscope, an optical microscope, a secondary electron detector, a secondary ion detector, or a photodetector.

[0162] Example 21 includes the subject matter of any of Examples 17-20, further specifying that the imaging is performed by guiding a first beam toward the sample, and the first beam includes an ion beam, a plasma focused ion beam, an electron beam, or a light beam.

[0163] Example 22 includes any of the subject matter of example 21, further specifying that the first beam is directed toward the sample along a beam axis oriented between 60° and 87° from a surface normal of the sample.

[0164] Example 23 includes the subject matter of any of Examples 17-22, further specifying that the sample comprises a cryogenic sample, a plunge frozen sample, a biological sample, a semiconductor sample, or a nanostructure.

[0165] Example 24 includes the subject matter of any of Examples 17-23, further specifying that the samples have a surface height variation in the range of 1 to 2, 2 to 5, 5 to 10, or 10 to 20 μm over a surface lateral extent of 2, 5, 10, 20, 50, or 100 μm.

[0166] Example 25 includes the subject matter of any of Examples 17-24, and further specifies that the focus synthesis is performed using one or more of a multi-scale transform, an image pyramid, a Gaussian image pyramid, a Laplacian image pyramid, or a wavelet transform.

[0167] Example 26 includes the subject matter of any of Examples 17-25, further specifying that the location is selected by a user or by an automated program.

[0168] Example 27 includes the subject matter of any of Examples 17-26, further specifying that the location corresponds to a feature within the sample.

[0169] Example 28 includes the subject matter of any of examples 17-27, further specifying that the localized action is performed by one or more of milling, etching, microscopy, or spectroscopy.

[0170] Example 29 includes the subject matter of any of examples 17-28, further specifying that the localized action is performed by directing a second beam at the sample, the second beam comprising an ion beam, a plasma focused ion beam, an electron beam, or a light beam.

[0171] Example 30 includes any of the subject matter of example 29, further specifying that the second beam is directed to the sample along a second axis oriented between 60° and 87° from a surface normal of the sample.

[0172] Example 31 includes the subject matter of any of Examples 17-30, and further specifies that the local operation occurs at the selected location, at a predetermined lateral offset from the selected location, at a predetermined depth offset from the selected location, over a region that includes the selected location, or over a region that has a lateral or depth offset from the selected location.

[0173] Example 32 includes the subject matter of any of Examples 18-31, and further specifies that the working depth is equal to a first value stored in the depth map for the selected location, a first average of values ​​stored in the depth map in a first zone that includes the selected location, a second value stored in the depth map at a predetermined lateral offset from the selected location, or a second average of values ​​stored in the depth map in a second zone having a lateral offset from the selected location, or a predetermined linear transformation thereof.

[0174] Example 33 includes the subject matter of any of examples 18-32, further specifying that the tool comprises one or more of a mill, an etcher, an electron microscope, an electron spectrometer, an ion microscope, an optical microscope, or an optical spectrometer.

[0175] Example 34 is one or more computer-readable media that store instructions, the instructions being non-transitory computer-readable media that, when executed by one or more hardware processors, cause a device to perform any one of the methods of Examples 17 to 33.

[0176] Generalized Computer Environment 12 illustrates a generalized example of a suitable computing system 1200 that can implement the described examples, techniques, and technologies for applying focus stacking to a sample preparation or analysis procedure. The computing system 1200 is not intended to suggest any limitations on the scope of use or functionality of the present disclosure, as the innovations can be implemented in a variety of general-purpose or special-purpose computing systems. The computing system 1200 can control a FIB imaging tool, a FIB milling tool, a SEM, a stage, an analytical instrument, or other similar equipment; can perform focus stacking, metrology, or other analysis on images or other acquired data representative of a sample; can control a stage, an ion beam column, or an electron beam column; or can acquire, process, output, or store measurement data.

[0177] Referring to Figure 12, a computing environment 1210 includes one or more processing units 1222 and memory 1224. In Figure 12, this basic configuration 1220 is included within a dashed line. The processing unit 1222 may execute computer-executable instructions, such as for control, measurement, or other functions described herein. The processing unit 1222 may be a general-purpose central processing unit (CPU), an application-specific integrated circuit (IC), or a combination of both. The tangible memory 1224 may be a processor in a general-purpose integrated circuit (ASIC), or any other type of processor. In a multiprocessing system, multiple processing units execute computer-executable instructions to increase processing power. The computing environment 1210 may also include a graphics processing unit or co-processing unit 1230. The tangible memory 1224 may be volatile memory (e.g., registers, cache, or RAM), non-volatile memory (e.g., ROM, EEPROM, or flash memory), or some combination thereof, accessible by the processing unit 1222, the processing unit 1230. The memory 1224 stores software 1280, in the form of computer-executable instructions suitable for execution by the processing unit 1222, the processing unit 1230, that implements one or more of the innovations described herein. For example, software 1280 may include software 1281 for controlling a FIB, SEM or other imaging tool, software 1282 for controlling a FIB or other milling tool, software 1283 for controlling a stage on which a sample is supported, software 1284 for focus stacking, software 1286 for performing metrology or other analysis of sample data, or other software 1285 (including user interfaces, host interfaces, or fault detection). The inset shown for software 1280 in storage 1240 is equally applicable to software 1280 in other locations in FIG. 12. Memory 1224 may also store control parameters, calibration data, measurement data, other database data, configuration data, or operational data.

[0178] Computing system 1210 may have additional functionality, such as one or more of storage devices 1240, input devices 1250, output devices 1260, or communication ports 1270. An interconnection mechanism (not shown), such as a bus, controller, or network, interconnects the components of computing environment 1210. Typically, operating system software (not shown) provides an operating environment for other software 1280 executing in computing environment 1210 and coordinates the operation of the components of computing environment 1210.

[0179] The tangible storage 1240 may be removable or non-removable and includes magnetic 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 1210. The storage 1240 stores instructions (including instructions and / or data) for the software 1280 that implements one or more of the innovations described herein. The storage 1240 may also store image data, measurement data, workflow programs, reference data, calibration data, configuration data, sample data, or other databases or other data structures described herein.

[0180] The input devices 1250 may be mechanical, touch-sensitive, or proximity-sensitive input devices, such as a keyboard, mouse, pen, touch screen, or trackball, voice input devices, scanning devices, or another device that provides input to the computing environment 1210. The output devices 1260 may be a display, printer, speakers, optical disk burner, or another device that provides output from the computing environment 1210. The input or output may also be communicated to or from a remote device (as described in the context of FIG. 1) over a network connection via communications port 1270.

[0181] The communications port 1270 enables communication over a communications medium to another computing entity. The communications medium conveys 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 characteristics set or changed in such a manner as to encode information in the signal. By way of example, and not limitation, the communications medium may use an electrical, optical, RF, acoustic, or other carrier wave.

[0182] A data acquisition system can be integrated into the computing environment 1210 either as an input device 1250 or coupled to a communication port 1270 and can include analog-to-digital converters or connections to an instrumentation bus. An instrumentation control system can be integrated into the computing environment 1210 either as an output device 1260 or coupled to a communication port 1270 and can include digital-to-analog converters, switches, or connections to an instrumentation bus.

[0183] In some embodiments, computer system 1200 may also include a computing cloud 1290 in which instructions for performing all or a portion of the disclosed techniques are implemented. Any combination of memory 1224, storage 1240, and computing cloud 1290 may be used to store software instructions and data for the disclosed techniques.

[0184] The innovation may be described in the general context of computer-executable instructions, such as instructions contained in a program module, executed in a computing system on a target real or virtual processor. Generally, program modules or components include routines, programs, libraries, objects, classes, components, data structures, etc. that perform particular tasks or implement particular data types. The functionality of the program modules may be combined or split between program modules as desired in various embodiments. The computer-executable instructions of the program modules may be executed in a local computing system or in a distributed computing system.

[0185] The terms "computing system," "computing environment," and "computing device" are used interchangeably herein. Unless the context clearly dictates otherwise, none of the terms limit 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 dedicated hardware and / or general-purpose hardware and / or virtualized hardware, along with 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 thus include both the software associated with the virtualization and the underlying hardware.

[0186] General Considerations As used in this application and the claims, the singular forms "a," "an," and "the" include the plural forms unless the content clearly dictates otherwise. Furthermore, the term "includes" means "comprises." Furthermore, the term "coupled" does not exclude the presence of intermediate elements between the coupled items. Furthermore, as used herein, the terms "or" and "and / or" refer to any one item or combination of items in a phrase.

[0187] The systems, methods, and devices described herein should not be construed as limiting in any manner. Instead, the present disclosure is directed to all novel and non-obvious features and aspects of the various disclosed embodiments, alone and in various combinations and subcombinations with each other. The disclosed systems, methods, and devices are not limited to any particular aspect or feature or combination thereof, and the disclosed systems, methods, and devices do not require that any one or more particular advantages exist or problems be solved. The techniques of any example can be combined with the techniques described in any one or more of the other examples.

[0188] Although some operations of the disclosed methods are described in a particular order for convenience of presentation, it should be understood that this method of description encompasses reordering, unless a particular order is required by specific language described herein. For example, operations described in sequence may be reordered or performed simultaneously in some cases. Furthermore, for the sake of simplicity, the accompanying drawings may not show various ways in which the disclosed systems, methods, and apparatuses may be used with other systems, methods, and apparatuses. Furthermore, the description may use terms such as "obtain," "apply," "average," "configure," "control," "correlate," "decimate," "decompose," "determine," "direct," "downsampling," "filtering," "fuse," "generate," "image," "generate," "search," "scaling," "select," "set," "store," "convert," or "upsampling" to describe the disclosed methods. These terms are high-level abstractions of the actual operations that are performed. The actual operations that correspond to these terms vary depending on the particular implementation and are readily discernible by those skilled in the art.

[0189] In some examples, values, procedures, or devices are described as "minimum," "best," "maximum," "optimum," "extreme," etc. It will be understood that such descriptions are intended to indicate that a selection from few or many alternatives is possible, and that such a selection is not necessarily lower, better, lesser, or otherwise preferred over other options.

[0190] Any theories of operation, scientific principles, or other theoretical explanations presented herein with reference to the disclosed devices or methods are provided for ease of explanation to aid in understanding and are not intended to be limiting in scope, i.e., the disclosed systems, methods, and devices are not limited to such theories of operation, and the appended claims are not limited to embodiments that function in a manner described by such theories of operation.

[0191] Any of the disclosed methods may be controlled by or implemented as computer-executable instructions or computer program products stored in one or more computer-readable storage media, such as tangible non-transitory computer-readable storage media, and executed on a computing device (e.g., any available computing device, including a tablet, smartphone, or other mobile device that includes computing hardware). A tangible computer-readable storage medium is an available tangible medium that can be accessed within a computing environment (e.g., one or more optical media disks, such as DVDs or CDs, a volatile memory component (such as DRAM or SRAM), or a non-volatile memory component (such as flash memory or hard drive)). As an example, referring to FIG. 12, a computer-readable storage medium includes memory 1224 and storage device 1240. The term computer-readable medium or computer-readable storage medium does not include signals and carrier waves. Furthermore, the term computer-readable medium or computer-readable storage medium does not include a communication port (e.g., 1270).

[0192] Any of the computer-executable instructions for carrying out the disclosed techniques, as well as any data created and used during the practice of the disclosed embodiments, can be stored on one or more computer-readable storage media. The computer-executable instructions can be, for example, a dedicated software application, or part of a software application accessed or downloaded via a web browser or other software application (such as a remote computing application). Such software can be executed, for example, on a single local computer (e.g., any suitable commercially available computer) or in a networked environment using one or more networked computers (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).

[0193] For clarity, only certain 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 a particular computer language or program. For example, the disclosed technology can be implemented by software written in Adobe Flash, C, C++, C#, Curl, Dart, Fortran, Java, JavaScript, Julia, Lisp, Matlab, Octave, Perl, Python, Qt, R, Ruby, SAS, SPSS, SQL, WebAssembly, any derivatives thereof, or any other suitable programming language, or in some embodiments, a markup language such as HTML or XML, or any combination of suitable languages, libraries, and packages. Similarly, the disclosed technology is not limited to a particular computer or type of hardware. Specific details of suitable computers and hardware are well known and need not be described in detail in this disclosure.

[0194] Additionally, any of the software-based embodiments (e.g., including computer-executable instructions for causing a computer to perform any of the disclosed methods) may be uploaded, downloaded, side-loaded, or remotely accessed through suitable communication means, including, for example, the Internet, the World Wide Web, an intranet, a software application, cable (including fiber optic cable), magnetic communication, electromagnetic communication (including RF communication, microwave communication, infrared communication, and optical communication), electronic communication, or other such communication means.

[0195] In view of the numerous possible embodiments to which the principles of the presently disclosed subject matter may be applied, it should be recognized that the described embodiments are merely preferred examples and should not be construed as limiting the scope of the claims. Rather, the scope of the claimed subject matter is defined by the following claims. We therefore claim all that comes within the scope and spirit of those claims.

Claims

1. An apparatus comprising: a tool configured to perform a localized action on the specimen using the first beam; an imaging device configured to acquire a plurality of images of the sample using a second beam; and A controller, focus-compounding the plurality of images to obtain a composite image; receiving an indication of a selected location in the composite image; a controller configured to cause the tool to perform the local action based on the selected location.

2. The controller: generating a depth map from the plurality of images; Using the depth map to determine an operating depth from the selected location; The apparatus of claim 1 , further configured to set the tool according to the working depth for the localized operation.

3. The apparatus of claim 1 , wherein the first beam comprises a plasma focused ion beam.

4. 10. The apparatus of claim 1, wherein the first beam is directed at the sample along a beam axis oriented between 60° and 87° from a surface normal of the sample.

5. The apparatus of claim 1 , wherein the imaging device comprises an ion microscope.

6. The apparatus of claim 1 , wherein the second beam is directed at the sample along a viewing axis oriented between 60° and 87° from a surface normal of the sample.

7. The apparatus of claim 1 , wherein the second beam comprises a plasma focused ion beam.

8. The apparatus of claim 1 , wherein the second beam comprises a light beam.

9. The apparatus of claim 1 , wherein the first beam and the second beam are generated in a common beam column.

10. The apparatus of claim 1 , wherein the controller is configured to perform the focus stacking using a multi-scale transform.

11. 1. A computer-implemented method comprising: acquiring a plurality of images of the sample at each focal depth; focus-compositing the plurality of images to obtain a composite image; receiving an indication of a selected location in the composite image; and performing a local action on the sample based on the selected location.

12. The focus stacking further generates a depth map of the sample, the computer-implemented method comprising: determining an operating depth from the selected location using the depth map; and setting a tool according to said working depth; The computer-implemented method of claim 11 , further comprising: causing the tool to perform the local action.

13. The computer-implemented method of claim 12 , wherein the working depth is based on a value stored in the depth map at a predetermined lateral offset from the selected position.

14. The computer-implemented method of claim 11 , further comprising forming a lamella incorporating the selected location, said forming comprising causing the localized operation.

15. The computer-implemented method of claim 11 , wherein the sample comprises a cryogenic biological sample.

16. 12. The computer-implemented method of claim 11, wherein the sample has a surface height variation in the range of 2 to 10 μm over a lateral surface extent in the range of 5 to 20 μm.

17. The computer-implemented method of claim 11 , wherein the location is selected by an automated program.

18. The computer-implemented method of claim 11 , wherein the localized action is performed by milling.

19. The computer-implemented method of claim 11 , wherein the localized motion occurs at a predetermined lateral and / or depth offset from the selected location.

20. One or more computer-readable media storing instructions that, when executed by one or more hardware processors, cause the one or more hardware processors to operate in a manner that: acquiring a plurality of images of the sample at each focal depth; focus-compositing the plurality of images to obtain a composite image; receiving an indication of a selected location in the composite image; and performing a local action on the sample based on the selected location.

Images