Method and system for reconstructing an image from electron diffraction data

Abstract

The present specification provides a method for generating an image (500) reconstructed from electron diffraction data of a test specimen (100), the method comprising: receiving electron diffraction data as electron diffraction data of a test specimen (100), the electron diffraction data comprising a plurality of test event detection coefficients (210a, 210b, 210c); For each test event detection coefficient (210a, 210b, 210c), the method comprises the step of selecting a kernel image (e.g., 310b, e, h) — selected kernel image (e.g., 310b, e, h) — from a kernel image library (300) containing a plurality of kernel images (e.g., 310a to i), wherein each kernel image (e.g., 310a to i) of the kernel image library (300) is to be added to a reconstruction grid (400) of a reconstructed image (500) as a pre-calculated mosaic, and wherein a plurality of selected added mosaics can mosaic and render the reconstructed image (500); and the method comprises the step of generating a reconstructed image (500), wherein each selected kernel image (310b, 310e, 310h) is added to the reconstruction grid (400) to progressively mosaic and render the reconstructed image (500).

Description

Technology Field

[0001] The present invention relates to the field of image reconstruction from electron diffraction data. Background Technology

[0002] According to recent trends in the field of scanning transmission electron microscopy (STEM), powerful methods are being implemented to directly reconstruct the electrostatic potential of solid specimens. These include center of mass integration (iCoM) of far-field intensity and ptychography, which was first pioneered in the field of X-ray microscopy. Electron ptychography exists in various forms, including iterative reconstruction methods and analytical approaches. Analytical ptychography includes side-band integration (SBI)—also known as single side-band reconstruction (SSB)—and Wigner distribution deconvolution (WDD).

[0003] A primary concern regarding these potential recovery methods is their inherent dose efficiency. STEM experiments are difficult to image because of the high risk of beam damage. In this context, performing measurements by utilizing each detected electron most efficiently is crucial for expanding applications into new scientific fields. One such field is structural biology, where less sophisticated TEM approaches are currently commonly used, and the main experimental limitation is the high probability of beam damage. In this context, any breakthrough that increases the information recovered per input electron could lead to significant advancements.

[0004] Generally, both iCoM and peticography require the pre-recording of a set of diffraction patterns corresponding to the lateral scan raster of a converging electron probe. This experimental setup is commonly referred to as 4D-STEM because it provides resolution in both real and inverse space, meaning it has four recording dimensions. Even in 1-bit mode, the frame time of a Medipix3-type detector is at least 80 μs. Consequently, 4D-STEM is time-consuming and susceptible to specimen drift and beam damage. This is one of the reasons why it is important to introduce faster options for both data acquisition and processing for live imaging. A second challenge is the volume of data generated as part of the experiment, which can reach tens of GB in just a few seconds. Transmitting this data for further processing entails burdensome requirements in terms of data bandwidth and computational power, necessitating very expensive, specialized, and difficult-to-maintain hardware. The problem to be solved

[0005] The present invention aims to provide a new framework for performing analytical peticography that includes the iCoM method as well as SBI and WDD approaches, while minimizing electron dose, acquisition time, and the need for data storage. To this end, new advancements in event-driven electron detection are utilized. means of solving the problem

[0006] The present specification provides a method for generating an image (500) reconstructed from electron diffraction data of a test specimen (100), said method comprising:

[0007] - A step of receiving electron diffraction data including a plurality of test event detection coefficients (210a, 210b, 210c) as electron diffraction data of a test specimen (100);

[0008] - For each test event detection coefficient (210a, 210b, 210c), a step of selecting a kernel image (e.g., 310b, e, h) from a kernel image library (300) containing a plurality of kernel images (e.g., 310b, e, h), wherein the kernel image (e.g., 310b, e, h) is a mosaic that has been calculated and is to be added to a reconstruction grid (400) of a reconstructed image (500), and wherein a plurality of selected added mosaics can mosaic and render the reconstructed image (500); and

[0009] - A step of generating a reconstructed image (500), comprising adding each selected kernel image (310b, 310e, 310h) to a reconstructed grid (400) to progressively mosaic and render the reconstructed image (500);

[0010] According to a preferred embodiment, each kernel image (310a to i) of the kernel image library (300) is:

[0011] - It has a size corresponding to a part of the size of the reconstruction grid (400);

[0012] - Position of the mock coefficient grid vector on the mock coefficient grid of the mock coefficient detector ( It is simulated from a single simulated event detection coefficient having ).

[0013] According to a desirable mode:

[0014] - Electron diffraction data is based on multiple different scanning grid vector positions (100a, 100b, 100c, ) additionally includes, and each scanning grid vector position is the vector position of the scanning grid (120) where the test electronic probe ( It is a location that interacts with the test specimen (100) to generate an exit wave (120a) from the test specimen (100);

[0015] - Among the multiple test event detection coefficients (210a, 210b, 210c), each test event detection coefficient (210a, 210b, 210c) is a test coefficient grid vector position (on the coefficient grid (205) of the event-driven detector (200) Having );

[0016] - Test coefficient grid vector position( ) is the scanning grid vector position (100a, 100b, 100c) that generated the exit wave (120a) causing the corresponding test event detection coefficients (210a, 210b, 210c) in the electron diffraction data It is connected to );

[0017] - Kernel images (310b, 310e, 310h) are ( )go ( Selected when it matches );

[0018] - Adding each selected kernel image (310b, 310e, 310h) to the reconstruction grid (400) is to the scan grid vector position (120a) that generated the exit wave (120a) causing the selected test event detection coefficients (210a, 210b, 210c) where the kernel image (310b, 310e, 310h) is selected Reconstruction grid vector position centered on ) It is performed in the set of ).

[0019] According to a preferred embodiment, the step of adding each selected kernel image (310b, 310e, 310h) to the reconstruction grid (400) comprises each reconstruction grid vector position of the added kernel image (310b, 310e, 310h) ( Individual pixel intensities at the same reconstruction grid vector location ( It includes a step of summing the existing individual pixel intensity in ).

[0020] According to a preferred embodiment, the reconstructed grid vector position ( Adding each selected kernel image (310b, 310e, 310h) to the reconstruction grid (400) in ) is done in an adjacent, overlapping, or superimposing manner, and

[0021] - Adjacent addition includes placing the selected kernel images (310b, 310e, 310h) next to the existing kernel images on the reconstruction grid (400);

[0022] - Overlapping addition includes positioning one or more edges of the selected kernel images (310b, 310e, 310h) to overlap with parts of one or more existing kernel images on the reconstruction grid (400);

[0023] - Overlay placement involves placing selected kernel images (310b, 310e, 310h) exactly overlaid on one or more existing kernel images on a reconstruction grid (400).

[0024] The existing kernel image on the reconstruction grid (400) is a selected kernel image previously added to the reconstruction grid (400) during the progressive mosaicking and rendering process of the reconstructed image (500).

[0025] According to a preferred embodiment, after the selected kernel images (310b, 310e, 310h) are added to the reconstruction grid (400), the reconstruction grid is updated, and only a portion of the reconstruction grid (400) to which the kernel images (310b, 310e, 310h) have been added is updated.

[0026] According to a preferred embodiment, the size of each kernel image (310a to i) of the kernel image library (300) is 50% or less of the size of the reconstruction grid (400) based on area.

[0027] According to a preferred embodiment, the mock coefficient grid vector position (in the image library (300) Simulating each kernel image (310a to i) from a single simulation coefficient having ) includes analytical peticography.

[0028] According to a desirable mode,

[0029] - Analytical peticography is center of mass integral (iCoM), sideband integral (SBI), or Wigner distribution deconvolution (WDD);

[0030] - The simulations of kernel images (310a to i) represent partial measurements of an electrostatic potential (μ) map weighted by an interaction constant (σ), partial measurements of a charge density (ρ) map weighted by the same interaction constant (σ), or the real or imaginary part of a transfer function (T) map, where the transfer function is a complex exponential function of the electrostatic potential (μ) weighted by the interaction constant (σ).

[0031] According to a preferred embodiment, the generation of the reconstructed image (500) is performed in real time.

[0032] Additionally, a method for interactively generating an image (500) reconstructed from electron diffraction data of a test specimen (100) is provided, comprising the method described herein, wherein:

[0033] - The above method further includes the step of transmitting an instruction to an electron microscope (EM) to direct a test electron probe toward a scanning grid vector position (100a, 100b, 100c, rs), thereby causing the electron probe to interact with a test specimen (100) and generate an exit wave (120a);

[0034] - In the next step:

[0035] A step of receiving electron diffraction data of a test specimen (100) including a plurality of test event detection coefficients (210a, 210b, 210c) induced by a generated exit wave (120a),

[0036] The electron diffraction data of the above test specimen (100) is generated by an electron microscope (EM);

[0037] - In the next step:

[0038] A step of progressively mosaicking and rendering the reconstructed image (500) by adding each selected kernel image (310b, 310e, 310h) to the reconstruction grid (400),

[0039] Each selected kernel image (310b, 310e, 310h) is a reconstructed grid vector position centered on the scan grid vector position (rs) that generated the exit wave (120a) causing the selected test event detection coefficient (210a, 210b, 210c) of the corresponding kernel image (310b, 310e, 310h). It is added to the reconstruction grid (400) from the set of );

[0040] - The above method is:

[0041] The method includes the step of repeating the transmission, reception, selection, and addition steps, wherein in at least one iteration, different scan grid vector positions (100a, 100b, 100c, It further includes a step of progressively mosaicking and rendering the reconstructed image (500) using ).

[0042] Additionally, an (interactive) method for interactively generating an image (500) reconstructed from electron diffraction data of a test specimen (100) is provided, said method:

[0043] - Transmit instructions to the electron microscope (EM) to scan the test electron probe to grid vector positions (100a, 100b, 100c, A step of directing the electron probe toward the test specimen (100) so that it interacts with the test specimen (100) and generates an exit wave (120a);

[0044] - A step of receiving electron diffraction data of a test specimen (100) including a plurality of test event detection coefficients (210a, 210b, 210c) induced by a generated exit wave (120a);

[0045] - For each test event detection coefficient (210a, 210b, 210c), a step of selecting a kernel image (e.g., 310b, e, h) from a kernel image library (300) containing a plurality of kernel images (310a to i), wherein each kernel image (e.g., 310a to i) of the kernel image library (300) is a pre-calculated mosaic to be added to a reconstruction grid (400) of a reconstructed image (500), and the plurality of selected added mosaics can mosaic and render the reconstructed image (500);

[0046] - Scan grid vector position (120a) that generates an exit wave (120a) that causes selected test event detection coefficients (210a, 210b, 210c) of the corresponding kernel image (310b, 310e, 310h) Reconstruction grid vector position centered on ) Step of adding to the reconstruction grid (400) from the set of );

[0047] - A step of repeating the transmission, reception, selection, and addition steps, wherein in at least one iteration, different scan grid vector positions (100a, 100b, 100c, It includes the step of progressively mosaicking and rendering the reconstructed image (500) using ).

[0048] According to a more preferred embodiment, the (interactive) method comprises any embodiment of the additional claimed subject matter of a method for generating an image (500) reconstructed from electron diffraction data of a test specimen (100) described herein.

[0049] The method according to the present invention is a computer-implemented method.

[0050] Additionally, a computing device or system configured to perform the (computer-implemented) method described in this specification is provided.

[0051] Additionally, a computer program or computer program product is provided having instructions that, when executed by a computing device or system, cause said device or system to perform the (computer implementation) method described in this specification.

[0052] Additionally, a computer-readable medium is provided that stores a computer program having instructions that, when executed by a computing device or system, cause said device or system to perform the method (computer implementation) described in this specification. Brief explanation of the drawing

[0053] FIG. 1 is a flowchart illustrating a method and system described in the present specification. Specific details for implementing the invention

[0054] Before describing the systems and methods of the present invention, it should be understood that the present invention is not limited to the specific systems, methods, or combinations described, and that such systems, methods, and combinations may naturally vary. Furthermore, it should be understood that the terms used herein should not be interpreted restrictively, as the scope of the present invention is limited only by the appended claims.

[0055] In this specification, the singular forms "one (a)," "one (an)," and "the" include both singular and plural references unless the context clearly indicates otherwise.

[0056] As used herein, the terms “comprising,” “comprises,” and “comprised of” are synonymous with “including,” “includes,” or “containing,” and “contains,” and are inclusive or open so as not to exclude additional unlisted members, elements, or method steps. These terms will be understood to include the terms “consisting of,” “consists,” and “consists of.”

[0057] The description by endpoint of a numerical range includes all numbers and fractions within each range, and the described endpoint.

[0058] In this specification, the terms “about” or “approximately” used to refer to measurable values, such as parameters, quantities, and temporal durations, encompass variations of ±10% or less of the specified value, preferably ±5% or less, more preferably ±1% or less, and even more preferably ±0.1% or less, where suitable for carrying out the disclosed invention.

[0059] The terms “one or more” or “at least one” include references to any one of the above members, or any two or more of the above members, for example, ≥3, ≥4, ≥5, ≥6, or ≥7 of the above members, and up to all members.

[0060] All references cited in this specification are incorporated herein by reference in their entirety.

[0061] Unless otherwise defined, all terms used to disclose the present invention, including technical and scientific terms, have the meaning generally understood by a person skilled in the art to which the present invention pertains.

[0062] Throughout this specification, references to "one embodiment" or "one embodiment" mean that specific features, structures, or characteristics described in relation to such embodiment are included in at least one embodiment of the present invention.

[0063] In the description of the invention, reference is made to the attached drawings, which constitute part of this specification, in which only specific embodiments in which the invention may be practiced are illustrated as examples. Unless otherwise indicated, all drawings in this document are not scaled and have been selected for the purpose of illustrating various embodiments of the invention.

[0064] It should be understood that other embodiments may be used and structural or logical changes may be made without departing from the scope of the invention. Accordingly, the following detailed description should not be interpreted in a limiting sense, and the scope of the invention is defined by the appended claims.

[0065] Key Introductory Content

[0066] The present specification provides a method or system for generating a reconstructed image (500) from electron diffraction data of a test specimen (100). A flowchart illustrating the method is disclosed in FIG. 1. The method or system receives electron diffraction data of a test specimen (100) collected by an event-driven detector (200), progressively mosaics and renders the reconstructed image (500) on a reconstruction grid (400) using a plurality of pre-calculated kernel images (310b, e, h), each pre-calculated kernel image (310b, e, h) is selected from a kernel image library (300), and the selection is made of the counting grid vector position (210a, 210b, 210c) of a detected event on the counting grid (210) of the event-driven detector (200). It is based on ). Vector location of selected kernel image (310b, e, h) on the reconstruction grid (400) ( ) is a scanning grid vector position (where the test electron probe (105) interacts with the test specimen (100) on the scanning grid (120) It is determined from ).

[0067] Accordingly, according to one embodiment, a method for generating an image (500) reconstructed from electron diffraction data of a test specimen (100) is provided, said method:

[0068] - A step of receiving electron diffraction data of a test specimen (100) including a plurality of test event detection coefficients (210a, 210b, 210c);

[0069] - For each test event detection coefficient (210a, 210b, 210c), a step of selecting a kernel image (e.g., 310b, e, h) from a kernel image library (300) containing a plurality of kernel images (e.g., 310a to i),

[0070] - A kernel image (e.g., 310a to i) of a kernel image library (300) is a previously calculated mosaic to be added to a reconstruction grid (400) of a reconstructed image (500), and a plurality of selected added mosaics can mosaic and render the reconstructed image (500); and

[0071] - A step of generating a reconstructed image (500), comprising adding each selected kernel image (310b, 310e, 310h) to a reconstructed grid (400) to progressively mosaic and render the reconstructed image (500);

[0072] According to a desirable mode:

[0073] - Each kernel image (310a to i) of the kernel image library (300) has a size corresponding to a part of the size of the reconstruction grid (400); and the mock coefficient grid vector position on the mock coefficient grid of the mock coefficient detector ( It is simulated from a single simulated event detection coefficient having ).

[0074] According to a desirable mode:

[0075] - Electron diffraction data is based on multiple different scanning grid vector positions (100a, 100b, 100c, ) additionally includes, and each scanning grid vector position is the vector position of the scanning grid (120) where the test electronic probe ( It is a location that interacts with the test specimen (100) to generate an exit wave (120a) from the test specimen (100);

[0076] - Among the multiple test event detection coefficients (210a, 210b, 210c), each test event detection coefficient (210a, 210b, 210c) is a test coefficient grid vector position (on the coefficient grid (205) of the event-driven detector (200) Having );

[0077] - Test coefficient grid vector position( ) is the position of the corresponding test coefficient grid vector in the electron diffraction data ( Scanning grid vector positions (100a, 100b, 100c) that generated an exit wave (120a) causing test event detection coefficients (210a, 210b, 210c) at ) It is connected to );

[0078] - Kernel images (310b, 310e, 310h) are ( )go ( Selected when it matches );

[0079] - Adding each selected kernel image (310b, 310e, 310h) to the reconstruction grid (400) is to the scan grid vector position (120a) that generated the exit wave (120a) causing the selected test event detection coefficients (210a, 210b, 210c) where the kernel image (310b, 310e, 310h) is selected Reconstruction grid vector position centered on ) It is performed in the set of )).

[0080] According to a more detailed embodiment, a method for generating an image (500) reconstructed from electron diffraction data of a test specimen (100) is provided, said method:

[0081] - As a step of receiving electron diffraction data of a test specimen (100), the electron diffraction data is,

[0082] - Multiple different scan grid vector positions (100a, 100b, 100c, As such, each scanning grid vector position is the vector position of the test electronic probe on the scanning grid (120) A plurality of different scanning grid vector positions (100a, 100b, 100c), which are positions that interact with the test specimen (100) to generate an exit wave (120a) from the test specimen (100). ); and

[0083] - Each scan grid vector position ( With respect to ), at least one test event detection coefficient (210a, 210b, 210c) is induced by the exit wave (120a) interacting with the event-driven detector (200), wherein each test event detection coefficient (210a, 210b, 210c) is a test coefficient grid vector position (on the coefficient grid (205) of the event-driven detector (200) Step comprising at least one test event detection coefficient (210a, 210b, 210c) having );

[0084] - For each test event detection coefficient (210a, 210b, 210c), a step of selecting a kernel image (e.g., 310b, e, h) from the kernel image library (300), wherein:

[0085] - The kernel image library (300) includes a plurality of pre-calculated kernel images (310a to i), where:

[0086] Each of the above plurality of kernel images (310a to i) is:

[0087] - As a mosaic to be added to the reconstruction grid (400) of the reconstructed image (500), a plurality of selected added mosaics can mosaic and render the reconstructed image (500);

[0088] - It has a size corresponding to a part of the size of the reconstruction grid (400);

[0089] - Position of the mock coefficient grid vector on the mock coefficient grid of the mock coefficient detector ( Simulated from a single simulated event detection coefficient having );

[0090] - Kernel images (310b, 310e, 310h) are ( )go ( Step selected when matching ); and

[0091] - A step of generating a reconstructed image (500), wherein the scanning grid vector position (120a) that generates an exit wave (120a) causing test event detection coefficients (210a, 210b, 210c) Reconstruction grid vector position centered on ) The method includes the step of adding each selected kernel image (310b, 310e, 310h) from a set of ) to a reconstruction grid (400) to progressively mosaic and render the reconstructed image (500).

[0092] In FIG. 1, a portion of the reconstruction grid (400) is shown. This portion is represented by a dotted box on the reconstruction image (500).

[0093] In conventional typical methods or systems, a diffraction pattern is recorded using a 2D full-frame image sensor, and this sensor reads a full-frame 2D image of the diffraction pattern (2D-FF-DP, 2D full-frame diffraction pattern). Different scanning grid vector positions ( A reconstructed image can be determined using multiple 2D-FF-DPs acquired from the source. However, this determination is typically performed offline using multiple stored 2D-FF-DPs, and a high-performance processor is required to convert each 2D-FF-DP into a reconstructed image and combine them. Generally, processing time ranges from less than one hour to a full day. Due to the full-frame-based readout method, typical storage requirements are high (e.g., 2 to 32 GB).

[0094] By creating a kernel image library (300) composed of pre-calculated kernel images (310a to i), most of the necessary calculations are performed in advance, so a reconstructed image can be generated in real time during collection. Therefore, the computation time is significantly reduced, and a final reconstructed image can be obtained at the end of collection.

[0095] The advantage of this real-time reconstruction is that the test electron probe (105) can be redirected to an area of ​​the scanning grid (120) that generates low-quality data due to the specimen thickness, for example, thereby allowing the area to be locally improved. Thus, interactive responsiveness enables real-time adjustment. Such adjustment is not possible in an offline scenario.

[0096] Furthermore, the method and system of the present invention can stop collection when the reconstructed image reaches sufficient quality and reduce beam damage by exposing only local areas of the specimen requiring additional data to longer beam dwell times. In contrast, conventional technology may irradiate the specimen more than necessary across the entire area to avoid data shortage.

[0097] Since reconstruction is primarily performed through image / mosaic addition operations, simplified processors such as Field Programmable Gate Arrays (FPGAs) can be used. This reduces hardware costs and provides valuable live feedback to users through real-time reconstructed images due to the high processing speed of FPGAs.

[0098] Since the reconstruction grid (400) effectively functions as a storage space for each selected kernel image (e.g., 310b, e, h), there is no need to store any data related to the event-driven detector (200). In particular, there is no need to store records of the detected events (210a, 210b, 210c). Since the event-driven detector is used, the conventional full frame image is not generated, and data processing requirements are significantly reduced. Since the pre-calculated kernel image to be added to the reconstruction grid (400) is selected using each detected event (210a, 210b, 210c), there is no need to store intermediate reconstructed images, and the reconstruction grid (400) effectively functions as a storage. Therefore, data storage requirements are reduced or substantially eliminated, except for the reconstruction grid (400) and / or the reconstructed images.

[0099] According to one embodiment, a method for interactively generating an image (500) reconstructed from electron diffraction data of a test specimen (100) is provided, said method:

[0100] - A step of transmitting an instruction to an electron microscope (EM) to direct the test electron probe toward a scanning grid vector position (100a, 100b, 100c, rs), so that the electron probe interacts with the test specimen (100) and generates an exit wave (120a);

[0101] - A step of receiving electron diffraction data of a test specimen (100) including a plurality of test event detection coefficients (210a, 210b, 210c) induced by a generated exit wave (120a);

[0102] - For each test event detection coefficient (210a, 210b, 210c), a step of selecting a kernel image (e.g., 310b, e, h) from a kernel image library (300) containing a plurality of kernel images (e.g., 310a to i), wherein:

[0103] - A kernel image (e.g., 310b, e, h) is a pre-calculated mosaic to be added to the reconstruction grid (400) of the reconstructed image (500), and a plurality of selected added mosaics can mosaic and render the reconstructed image (500), step;

[0104] - Scan grid vector position (120a) that generates an exit wave (120a) that causes selected test event detection coefficients (210a, 210b, 210c) of the corresponding kernel image (310b, 310e, 310h) Reconstruction grid vector position centered on ) A step of adding each selected kernel image (310b, 310e, 310h) from the set of ) to the reconstruction grid (400);

[0105] - A step of repeating the transmission, reception, selection, and addition steps (multiple times), wherein different scan grid vector positions (100a, 100b, 100c, at least one repetition (preferably multiple repetitions) The method includes the step of using or transmitting ) to progressively mosaic and render the reconstructed image (500).

[0106] The comparison steps for the method for generating a reconstructed image (500) described in this specification may be applied to a method for interactively generating a reconstructed image (500) by modifying only the necessary parts (mutatis mutandis).

[0107] Optionally, the reconstruction grid (400) may be displayed on a display (e.g., a computer display, a smart device display). Optionally, the method may be used for different scan grid vector positions (100a, 100b, 100c, Input can be received from ). Display and input allow the user or machine to interactively monitor the progress of the reconstruction grid (400) generation.

[0108] If one or more locations of the reconstruction grid (400) show defects or require improvement, the test electronic probe (105) corresponds to the scan grid vector locations (100a, 100b, 100c, By moving to the location of the defective area or the location of the reconstruction grid (400) that requires improvement, additional kernel images or mosaics can be added. This reduces beam damage because only the area requiring additional data collection is scanned. In the prior art, when a defect appears in one area of ​​the reconstruction grid (400), the entire specimen area is beam-scanned to damage all areas equally and subsequently cause distortion in the collected results.

[0109] According to one embodiment, the location of the reconfiguration grid (400) that is defective or requires improvement is identified from the display, and the provided input is the test electronic probe (105) to scan grid vector location (100a, 100b, 100c) corresponding to the location. The location of the reconfiguration grid (400) that is defective or needs improvement can be identified automatically (e.g., using a machine learning protocol) or based on user input.

[0110] Electron diffraction data is acquired using an electron microscope (EM). An EM can be any device that emits an electron beam (test electron probe (105)) capable of generating electron diffraction data of a test specimen. Examples of EMs include scanning electron microscopes (SEM) or scanning transmission electron microscopes (STEM). Examples of EM providers include Zeiss, Tescan, JEOL, Hitachi, and ThermoFisher Scientific. Standard methods for collecting electron diffraction data of a specimen are known in the field from standard EM protocols and / or related textbooks.

[0111] The test electron probe (105) is preferably a convergent beam and generates a convergent beam electron diffraction (CBED) pattern on the counting grid (205) of an event-driven detector.

[0112] Generally, EM is:

[0113] - Electronic source;

[0114] - A focusing unit configured to form electrons into a test electron probe (105);

[0115] - A deflection unit configured to adjust the position of a test electronic probe (105) on a test specimen (100) or a scanning grid (120);

[0116] - One or more event-driven detectors (200) configured to detect electrons generated from the interaction of a test electron probe (105) with a test specimen (100) on a scanning grid (120) (one or more event-driven detectors (200) are placed after the test specimen to collect electron diffraction data);

[0117] - A controller configured to control the EM;

[0118] - A memory configured to store a computer-readable command that commands a controller to operate an EM according to the present invention;

[0119] - A test specimen stage for adjusting the position of the specimen or scanning grid (120) relative to the beam; and

[0120] - Includes a vacuum chamber, an attached vacuum pump, and a vacuum sensor.

[0121] The event-driven detector (200) detects an event that occurs as a test electron probe (105) interacts with a test specimen (100) on a scanning grid (120) and generates a test event detection coefficient (TEDC). The TEDC event is a scanning grid vector position ( ) arises from the interaction between the test electron probe (105) and the specimen (100). More specifically, TEDC occurs when exit electrons emitted after passing through the test specimen (100) are at the counting grid vector position ( It occurs by colliding with the counting grid (210) at ). TEDC is generated only when an event exists, in which case the counting grid vector position ( ) is assigned. TEDC does not have a gradual amplitude and can be represented by "1" or "Yes" or other binary indications. The event-driven detector (200) does not generate frame-based readout outputs, and only the area of ​​the counting grid (210) collided by the exit wave forms part of the output of the event-driven detector (200). Companies providing examples of event-driven detectors include Advacam, Amsterdam Scientific Instruments, and Quantum Detectors.

[0122] The test electron probe (105) is a convergent focused beam and generates a convergent beam electron diffraction (CBED) pattern on the counting grid (205) of the event-driven detector.

[0123] The test electronic probe moves in a scanning pattern on the scanning grid (120), and at each scanning grid vector position ( , 100a, 100b, 100c) are paused for a period of time, and during that period, one or more test event detection coefficients (TEDC) are detected by an event-driven detector (200).

[0124] Multiple different injection sites ( In acquiring electron diffraction data, the coil of the deflection unit controlling the lateral position of the test electron probe (105) is generally set to a scanning pattern. Types of scanning patterns include raster scanning, snake-pattern scanning, random position scanning, or user-controlled scanning. Raster scanning means that the test electron probe (105) is controlled to move across the specimen in an array of parallel sweeps. User-controlled scanning means that the user can freely move the scanning test electron probe (105) in any direction. Through user-controlled scanning, the user can "draw" in a sweep form on the reconstruction grid in real time, and for low-resolution areas, sweeps of the test electron probe (105) can be repeatedly applied under the user's control. The test electron probe (105) is preferably positioned at two spatially separated different scanning grid vector positions (100a, 100b, 100c, It is preferably blanked during repositioning between ).

[0125] The electron diffraction data includes a plurality of test event detection coefficients (TEDC), each TEDC being at least one coefficient grid vector position on the coefficient grid (210) of the event-driven detector (200). )(x, y coordinates) and, the scan grid vector location where the corresponding TEDC occurred ( It is expressed together with the indication of ). The TEDC does not have a gradual amplitude and can be represented by "1" or "Yes" or other binary indicators, or simply by the presence of at least one x, y coordinate within the electron diffraction data. The arrival time of the event on the event-driven detector (200) does not need to be recorded or present in the electron diffraction data. The TEDC is generated only when an event is present, in which case the counting grid vector position ( ) is granted.

[0126] Detection events are typically at the coefficient grid vector location ( ) is the activation of a single pixel on the counting grid (210). The detection event is the counting grid vector position ( It may be a cluster of adjacent (neighboring) pixels containing ).

[0127] Electron diffraction data is generally generated with a beam energy suitable for generating electron diffraction data. The beam energy may have an acceleration voltage of 1 to 300 keV, preferably 30 to 300 keV. Since the disclosure of the present invention allows for the collection of electron diffraction data using lower beam energies, the EM may provide an EM beam having an energy of 1 keV to 30 keV. A meaningful electron diffraction dataset is obtained within the specimen region. 2 50 to 10 per 6 A beam dose of receiving electrons can be obtained. This beam dose is generally achieved with a residence time of 1 μs to 1 ms per scanning point.

[0128] Scan grid vector position(100a, 100b, 100c, ) may differ from each other and may be visited only once by the test electronic probe (105). Scanning grid vector positions (100a, 100b, 100c, The points may differ from each other and may be visited at least twice by the test electronic probe (105). When visited multiple times with a low dose per point, beam damage is reduced compared to when the beam is irradiated only once per point and a higher dose is used.

[0129] Each selected kernel image (310b, 310e, 310h) is a reconstruction grid vector position ( It is added to the reconstruction grid (400) from the set of ). Reconstruction grid vector position ( The vector locations of the set of ) represent a continuous set of pixels (e.g., square boxes) having dimensions corresponding to the selected kernel image (310b, 310e, 310h). Added or adding means that the pixel intensity of the selected kernel image (310b, 310e, 310h) is combined (e.g., summed) with the existing pixel intensity at the corresponding location on the reconstruction grid (400).

[0130] The reconstructed image (500) is progressively mosaicked and rendered by adding each selected kernel image (310b, 310e, 310h). Mosaicing or mosaicking refers to the process of adding multiple selected kernel images (310b, 310e, 310h) (mosaics) at different, sometimes the same, reconstruction grid vector locations (r) over time, thereby progressively rendering the reconstructed image (500). The kernel image (or mosaic) has a size corresponding to a portion (a smaller proportion) of the size of the reconstruction grid (400). For example, the kernel image (or mosaic) may have an area that is 50% or less of the area of ​​the reconstruction grid (400). This area ratio depends on both the kernel image size and the number of visited scan points (e.g., the total size of the scan window).

[0131] The kernel image size is preferably user-defined and represents the area in the beam-irradiated specimen considered to participate in electron scattering by the test electron probe. Thus, this is a selected metric for the real-space extension of the electron probe. This can be expressed in nanometers or pixels, and the kernel image size can generally be 9×9 to 41×41 pixels.

[0132] Addition to the reconstruction grid (400) may be initial, meaning that the selected kernel image (310b, 310e, 310h) or mosaic occupies a position on the reconstruction grid (400) that has never been occupied by the existing kernel image and is adjacent to, overlaps with, or does not overlap with the existing kernel image.

[0133] The existing kernel image on the reconstruction grid (400) is a selected kernel image previously added to the reconstruction grid (400) during the progressive mosaicking and rendering process of the reconstructed image (500).

[0134] Additions to the reconstruction grid (400) may be adjacent additions, which means that selected kernel images (310b, 310e, 310h) or mosaics are placed next to existing kernel images on the reconstruction grid (400) (without overlap).

[0135] Addition to the reconstruction grid (400) may be additional or alternatively an overlapping addition, which means that a selected kernel image (310b, 310e, 310h) or mosaic is positioned such that one or more of its edges overlap with parts of one or more existing kernel images on the reconstruction grid (400).

[0136] Addition to the reconstruction grid (400) may be additional or alternatively superimposed, which means that a selected kernel image (310b, 310e, 310h) or mosaic is placed exactly on top of one or more existing kernel images on the reconstruction grid (400). For example, in FIG. 1, three kernel images (310b, 310e, 310h) are added to the reconstruction grid (400) by superimposing.

[0137] After a selected kernel image (310b, 310e, 310h) or mosaic is added to the reconstruction grid (400), the reconstruction grid is updated. Only a portion of the reconstruction grid (400) to which the kernel image (310b, 310e, 310h) or mosaic has been added may be updated. Multiple selected added mosaics can mosaic and render the reconstruction image (500).

[0138] Reconstruction grid vector location (400) where the selected kernel images (310b, 310e, 310h) are added to the reconstruction grid (400) ) is the scanning grid vector position ( It is determined from ). Generally, the reconstructed grid vector position ( ) is the scan grid vector position( It is determined from a proportional relationship with ), and optionally allows the reconstruction grid (400) to have a border (e.g., blank or black). The proportional relationship can be expressed according to Equation 1:

[0139]

[0140] Here ( ) is the reconstruction grid vector position to be determined, and

[0141] ( ) is the scan grid vector position,

[0142] Scale is a scaling factor constant (1 if no scaling), and

[0143] Offset x,y is an offset constant (0 if there is no border).

[0144] By adding and mosaicking pixels of multiple selected kernel images (310b, 310e, 310h), the signal-to-noise ratio of the reconstruction grid (400) and the reconstruction image (500) is improved.

[0145] The kernel image library (300) includes a plurality of pre-calculated kernel images (310a to i). Each of the plurality of kernel images (310a to i) is a mosaic to be added to the reconstruction grid (400) of the reconstructed image (500). Each kernel image (310a to i) is a mosaic to be added to the reconstruction grid (400) of the reconstructed image (500), and the mock coefficient grid vector position ( on the mock coefficient grid of the mock coefficient detector) It is simulated from a single simulated event detection coefficient having ).

[0146] Each kernel image (310a to i) of the kernel image library (300) is a mock coefficient grid vector location that generated the corresponding kernel image (310a to i) or mosaic ( It is connected to )(320a to 320i). Therefore, the test coefficient grid vector position ( By querying the kernel image library (300) with ), the equivalent mock coefficient grid vector location ( You can select kernel images (310a to i) connected to ). Therefore, kernel images (310b, 310e, 310h) are ( )go ( It is selected when it matches ).

[0147] The kernel images (310a to i) may represent partial measurements of an electrostatic potential (μ) map weighted by an interaction constant (σ), partial measurements of a charge density (ρ) map weighted by the same interaction constant (σ), or the real or imaginary part of a transfer function (T) map. The transfer function is a complex exponential function of the electrostatic potential (μ) weighted by the interaction constant (σ).

[0148] "Pre-calculated" means that the kernel image library (300) is created prior to the acquisition of electron diffraction data or prior to the data acquisition session. "Prior" may be, for example, within a few days (e.g., 1 to 3 days), hours, or minutes prior to the acquisition of electron diffraction data. Since the kernel image library (300) does not depend on the specimen, it can be retrieved from the database of the stored kernel image library (300). Such "prior" may be, for example, a few days (e.g., 1 to 3 days), hours, or a time longer than minutes prior to the acquisition of electron diffraction data.

[0149] Each kernel image (310a to i) of the kernel image library (300) is a mock coefficient grid vector location ( It is simulated from a single simulated coefficient having ). The simulation involves using analytical ptychography, the definition of which is extended herein to include the iCoM method. The simulation is performed using scan grid vector positions (100a, 100b, 100c, It does not require ).

[0150] The analytical peticography is preferably one of the center-of-mass integral (iCoM), sideband integral (SBI, also known as single-sideband (SSB) reconstruction), or Wigner distribution deconvolution (WDD). These methods are known in the art.

[0151] iCoM is “Phase contrast STEM for thin samples: Integrated differential phase contrast”, Laziζ et al It was first described in Ultramicroscopy, Volume 160, January 2016, pages 265-280 (https: / / doi.org / 10.1016 / j.ultramic.2015.10.011). It should be noted that this method is distinct from the integral derivative phase contrast (iDPC) first described in the same literature.

[0152] SBI “Experimental tests on double-resolution coherent imaging via STEM”, Rodenburg, et al It was first described in Ultramicroscopy, Volume 48, Issue 3, March 1993, Pages 304-314 (https: / / doi.org / 10.1016 / 0304-3991(93)90105-7).

[0153] WDD is "The theory of super-resolution electron microscopy via Wigner-distribution deconvolution", Rodenburg and Bates, Philosophical Transactions of the Royal Society of London. It was first described in Series A: Physical and Engineering Sciences, 15 June 1992, Volume 339, Issue 1655 (https: / / doi.org / 10.1098 / rsta.1992.0050).

[0154] Each kernel image (310a to i) may represent a partial measurement of an electrostatic potential (μ) map weighted by an interaction constant (σ), a partial measurement of a charge density (ρ) map weighted by the same interaction constant (σ), or the real or imaginary part of a transfer function (T) map. Thus, the reconstructed image (500) may be an electrostatic potential (μ) map weighted by the interaction constant (σ) of the specimen, a charge density (ρ) map weighted by the same interaction constant (σ), or the real or imaginary part of a transfer function (T) map.

[0155] To generate kernel images (310a to i) from a kernel image library (300), the conventional formalisms for the iCoM, SBI, and WDD methods are extended by describing the simulated diffraction data as a sum of Dirac functions, with one Dirac function used for each simulated coefficient. The existence of these Dirac functions enables the mathematical simplification of the integration and Fourier transform that must be explicitly performed on a complete four-dimensional electron diffraction dataset. Thus, each analytical peticography process is reduced to a sum of kernel images, each kernel image is pre-calculated and reduced in size in real space by a given kernel image size, enabling faster processing and reduced memory bandwidth.

[0156] Simulation of each kernel image preferably uses one or more of the following:

[0157] - One or more recording condition parameters (acceleration voltage V (kV), semi-convergence angle α (mrad), scan step size (nm));

[0158] - In the counting grid (205) of the event-driven detector (200) ( ) Coordinate calibration;

[0159] - Desired kernel size (in nm);

[0160] - Wiener number ε (e.g., a small number used to avoid division by zero, typically 10 -4 to 10 -8 Selected from a range, it acts as a noise filtering parameter);

[0161] Other optional parameters for band-pass filtering, aberration correction, incoherence compensation, contrast transfer function (CTF) compensation, and modulation transfer function (MTF) compensation may be selected.

[0162] The image kernel library (300) does not depend on the specimen.

[0163] The kernel image library (300) is stored in memory accessible to a standard computer system configured to generate a reconstructed image (500). The standard computer system is configured to perform one or more methods described herein. The standard computer system may be, for example, an Intel Architecture IA-32-based computer system. The method(s) may be implemented as programming instructions of one or more software modules stored in a non-volatile (e.g., hard disk or solid-state drive) storage device associated with said computer system. However, as mentioned elsewhere in this specification, at least some steps, in particular the selection of the kernel image and / or the generation of the reconstructed image, may be implemented in part or in whole by one or more dedicated hardware components, such as gate configuration data for one or more field programmable gate arrays (FPGAs) or application-specific integrated circuits (ASICs).

[0164] After an event is detected by the event-driven detector (200), the test coefficient grid vector position of the event ( ) is queried against the kernel image library (300). Equivalent mock coefficient grid vector location ( A kernel image (310a to i) of a kernel image library (300) having ) is selected. The selected kernel image (310a to i) is added to a reconstruction grid (400).

[0165] The method or system described herein may be used to generate an image (500) reconstructed from electron diffraction data in real time. Real time means that a detected event (210a, 210b, 210c) on the counting grid (210) of the event-driven detector (200) triggers the selection of a kernel image (e.g., 310b, e, h) from the kernel image library (300). The selection of the kernel image (e.g., 310b, e, h) occurs before the test electron probe (105) has completely scanned the scanning grid (120). Preferably, the selection of the kernel image (e.g., 310b, e, h) occurs when the test electron probe (105) reaches the next vector position ( It occurs before moving to ). Real-time may mean that electron diffraction data generated during the data acquisition session is not stored in storage memory such as a hard drive. During real-time processing, it is expected that multiple detected events (210a, 210b, 210c) may be buffered, for example, in transitory storage of limited capacity. Real-time generation enables the identification of one or more locations in the reconstruction grid (400) that show defects or require improvement during the acquisition of electron diffraction data, thereby enabling the test electron probe (105) to move to the corresponding scan grid vector location (100a, 100b, 100c, rs) to add additional kernel images or mosaics to the locations that are defective or require improvement.

[0166] As mentioned elsewhere in this specification, each kernel image (310a to i) is a mock coefficient grid vector position on the mock coefficient grid of the mock coefficient detector ( It is simulated from a single simulated event detection coefficient having ). Below, a detailed description regarding the generation of pre-calculated kernel images (310a to i) for the kernel image library (300) is provided.

[0167] The simulated diffraction dataset is the spatial frequency on the simulated count detector Mock scan position by pixels It includes a set of simulated event detection coefficients detected in, for example, a set of indicators d for each indicator c.

[0168] represents an aperture inserted into the focal plane of the microscope. spatial frequency satisfying It is 1 for, and 0 otherwise. Is It is a spatial frequency limit given by, where α is the semi-convergence angle of the illumination and λ is the electromagnetic wavelength determined by the acceleration voltage U.

[0169] is a kernel limitation term that includes a Hann window. The limitation radius, or kernel radius, is selected by the user and expressed in nm units. This distance is directly converted into the number of pixels of the kernel image size.

[0170] is a band-pass filtering window. The frequency limit is selected by the user. This band-pass filtering window can alternatively be replaced by a frequency weighting function, for example, spatial frequency It can be replaced by assigning a value between 0 and 1 that directly depends on it.

[0171] ε is a Wiener parameter selected by the user.

[0172] p is a CTF reward parameter selected by the user, with a value between 0 and 1.

[0173] is the effective source size to be corrected (spatial incoherence correction), and is the focus dispersion (time incoherence correction) that is subject to correction. and It is provided by the user as a single value (expressed in nm units).

[0174] is the aberration function to be corrected, and is given as follows:

[0175] ,

[0176] Here, parameter C a,b (expressed in nm units) and β a,b (expressed in radians) is provided by the user for a specific pair of integers [a, b], and the corresponding integer pair [a, b] is also provided by the user. is spatial frequency It is the angle direction of, and is the modulus (absolute value).

[0177] represents the camera's modulation transfer function. This is provided by the user if correction is required, otherwise it is set to 1.

[0178] F and F -1 represent the Fourier transform and the inverse Fourier transform, respectively. An arbitrary function The Fourier transform of is given as follows.

[0179]

[0180] And, an arbitrary function The inverse Fourier transform of is given as follows.

[0181]

[0182] Center of Mass Integration (iCoM)

[0183] Electrostatic potential:

[0184] Charge density:

[0185] Here:

[0186]

[0187]

[0188] CTF Reward Items:

[0189] Sideband Integral (SBI)

[0190] Electrostatic potential:

[0191] Charge density:

[0192] Preferably:

[0193]

[0194]

[0195]

[0196] More preferably:

[0197]

[0198]

[0199] and is a double nested function and is given as follows.

[0200]

[0201]

[0202] and The illumination correction term can be given as follows.

[0203]

[0204]

[0205] More preferably, and The illumination correction term can be given as follows.

[0206]

[0207]

[0208] CTF Reward Items:

[0209] Wigner distribution deconvolution (WDD)

[0210] Imaginary part:

[0211] Real part:

[0212] Here:

[0213]

[0214]

[0215] Lighting Convolutional Term:

[0216] Lighting Correction Item: Preferably:

[0217] Based on the initial calculation using the formula presented above, the library , , , , and It can be used to generate images through the process described elsewhere in this document. Spatial frequency grid is the mock coefficient grid vector position( Corresponds to ). These match the spatial frequency coordinates of the event-driven detector coefficient grid corrected prior to calculation. Real-space grid is a reconstruction grid. Previously used notation Is, ( )and ( Scan the kernel image first selected by the correspondence of ). It indicates centering on.

[0218] Electron diffraction data is collected during the data acquisition session. During the data acquisition session, the specimen maintains the same position relative to the specimen holder.

[0219] The present specification provides an EM (and a computing device or system) configured to perform the method described herein. The EM may provide an EM beam having an energy of 1 keV to 300 keV, preferably 30 keV to 300 keV. Since the disclosure of the present invention can collect electron diffraction data using lower beam energies, the EM may provide an EM beam having an energy of 1 keV to 30 keV.

[0220] The method according to the present invention is a computer-implemented method. The method can be performed by a standard (general) computer system.

[0221] In addition, a computing device or system configured to perform the method or part thereof described in this specification is provided.

[0222] In addition, a computer program or computer program product is provided having instructions that cause said device or system to perform the method or part thereof described in this specification when executed by a computing device or system.

[0223] Additionally, a computer-readable medium is provided that stores a computer program (product) having instructions that, when executed by a computing device or system, cause said device or system to perform the method (each step of) or part thereof described in this specification.

[0224] Additionally, a data stream representing a computer program or computer program product having instructions that, when executed by a computing device or system, cause said device or system to perform the method (each step of) or part thereof described in this specification is provided.

[0225] The method described in this specification may optionally be performed by an electron microscope equipped with a standard computer system.

[0226] Additionally, an electron microscope configured to perform the method or part thereof described in this specification is provided.

[0227] In addition, a computer program or computer program product is provided having instructions that cause said electron microscope to perform the method or part thereof described in this specification when executed by an electron microscope.

[0228] Additionally, a computer-readable medium is provided that stores a computer program (product) having instructions that cause said electron microscope to perform each step or part thereof of the method described in this specification when executed by an electron microscope.

[0229] Additionally, a data stream is provided representing a computer program or computer program product having instructions that cause said electron microscope to perform each step or part thereof of the method described herein when executed by an electron microscope.

Claims

Claim 1 A method for generating a reconstructed image (500) from electron diffraction data of a test specimen (100), comprising: receiving electron diffraction data including a plurality of test event detection coefficients (210a, 210b, 210c) as electron diffraction data of a test specimen (100); and for each test event detection coefficient (210a, 210b, 210c), selecting a kernel image (e.g., 310b, e, h) — selected kernel image (310b, 310e, 310h) — from a kernel image library (300) including a plurality of kernel images (e.g., 310a to i), wherein each kernel image (e.g., 310a to i) of the kernel image library (300) is a previously calculated mosaic to be added to a reconstruction grid (400) of the reconstructed image (500), and the plurality of selected added mosaics are said to be the reconstructed image (500). A method comprising: a step capable of mosaicking and rendering; and a step of generating the reconstructed image (500), wherein each selected kernel image (310b, 310e, 310h) is added to the reconstructed grid (400) to progressively mosaick and render the reconstructed image (500). Claim 2 In claim 1, each kernel image (310a to i) of the kernel image library (300) has a size corresponding to a part of the size of the reconstruction grid (400); and a mock coefficient grid vector position on the mock coefficient grid of the mock coefficient detector A method simulated from a single simulated event detection coefficient having Claim 3 In claim 1 or 2, the electron diffraction data further includes a plurality of different scanning grid vector positions (100a, 100b, 100c), each scanning grid vector position being a vector position of the scanning grid (120) where the test electron probe A location that interacts with the test specimen (100) to generate an exit wave (120a) from the test specimen (100); and each of the plurality of test event detection coefficients (210a, 210b, 210c) is a test coefficient grid vector position on the coefficient grid (205) of the event-driven detector (200). having; the above test coefficient grid vector position is the test coefficient grid vector position in the above electron diffraction data. The selected kernel images (310b, 310e, 310h) are connected to the scan grid vector positions (100a, 100b, 100c) that generated the exit wave (120a) causing the test event detection coefficients (210a, 210b, 210c) in the above; go Selected when it matches; adding each selected kernel image (310b, 310e, 310h) to the reconstruction grid (400) is a reconstruction grid vector position centered on the scan grid vector position that generated the exit wave (120a) causing the selected test event detection coefficients (210a, 210b, 210c) of the selected kernel image (310b, 310e, 310h). A method performed in a set of. Claim 4 In paragraph 3, the step of adding each selected kernel image (310b, 310e, 310h) to the reconstruction grid (400) comprises the reconstruction grid vector positions of each added selected kernel image (310b, 310e, 310h). Individual pixel intensity at the same reconstructed grid vector position A method comprising the step of summing with the existing individual pixel intensity in. Claim 5 In paragraph 3 or 4, the reconstructed grid vector position Adding each selected kernel image (310b, 310e, 310h) to the reconstruction grid (400) is performed in an adjacent, overlapping, or superimposing manner: the adjacent addition includes placing the selected kernel image (310b, 310e, 310h) next to an existing kernel image on the reconstruction grid (400); the superimposing addition includes placing one or more edges of the selected kernel image (310b, 310e, 310h) so as to overlap with a portion of one or more existing kernel images on the reconstruction grid (400); the superimposing addition includes placing the selected kernel image (310b, 310e, 310h) exactly overlapping on one or more existing kernel images on the reconstruction grid (400); and the existing kernel image on the reconstruction grid (400) is previously added to the reconstruction grid (400) during the progressive mosaicking and rendering of the reconstructed image (500). Method, which is the selected kernel image. Claim 6 A method according to any one of claims 1 to 5, wherein after the selected kernel image (310b, 310e, 310h) is added to the reconstruction grid (400), the reconstruction grid is updated, and only a portion of the reconstruction grid (400) to which the selected kernel image (310b, 310e, 310h) is added is updated. Claim 7 A method according to any one of claims 1 to 6, wherein the size of each kernel image (310a to i) of the kernel image library (300) is 50% or less of the size of the reconstruction grid (400) based on area. Claim 8 In any one of paragraphs 2 to 7, the mock coefficient grid vector position in the image library (300) A method comprising simulating each kernel image (310a to i) from a single simulation coefficient having an analytical ptychography. Claim 9 In claim 8, the analytical peticography is a center of mass integral (iCoM), a sideband integral (SBI), or a Wigner distribution deconvolution (WDD); and the simulation of kernel images (310a to i) in the image library (300) represents a partial measurement of an electrostatic potential (μ) map weighted by an interaction constant (σ), a partial measurement of a charge density (ρ) map weighted by the same interaction constant (σ), or the real or imaginary part of a transfer function (T) map, wherein the transfer function is a complex exponential function of the electrostatic potential (μ) weighted by the interaction constant (σ). Claim 10 A method in which, in any one of claims 1 to 9, the generation of the reconstructed image (500) is performed in real time. Claim 11 A method for interactively generating an image (500) reconstructed from electron diffraction data of a test specimen (100), comprising the method of any one of claims 1 to 10, wherein the method further comprises the step of transmitting an instruction to an electron microscope (EM) to direct a test electron probe toward a scanning grid vector position (100a, 100b, 100c) so that the electron probe interacts with the test specimen (100) and generates an exit wave (120a); wherein, in the step of receiving electron diffraction data of the test specimen (100) including a plurality of test event detection coefficients (210a, 210b, 210c) induced by the generated exit wave (120a), the electron diffraction data of the test specimen (100) is generated by the electron microscope (EM); and wherein the reconstructed image (500) is generated by adding each selected kernel image (310b, 310e, 310h) to the reconstruction grid (400). In the step of progressively mosaicking and rendering, each selected kernel image (310b, 310e, 310h) is a reconstruction grid vector position centered on the scan grid vector position that generated the exit wave (120a) causing the selected test event detection coefficients (210a, 210b, 210c) of the kernel image (310b, 310e, 310h). A method for adding to the reconstruction grid (400) in a set of ); the method further comprises the step of repeating the transmission, reception, selection, and addition steps, wherein different scan grid vector positions (100a, 100b, 100c) are used in at least one iteration to progressively mosaic and render the reconstruction image (500). Claim 12 A method for interactively generating an image (500) reconstructed from electron diffraction data of a test specimen (100), comprising: transmitting an instruction to an electron microscope (EM) to direct a test electron probe toward a scanning grid vector position (100a, 100b, 100c), thereby causing the electron probe to interact with the test specimen (100) and generate an exit wave (120a); receiving electron diffraction data of the test specimen (100) including a plurality of test event detection coefficients (210a, 210b, 210c) induced by the generated exit wave (120a); and for each test event detection coefficient (210a, 210b, 210c), a kernel image (e.g., 310b, e, h) from a kernel image library (300) including a plurality of kernel images (e.g., 310a to i) — a selected kernel image (310b, 310e, 310h). — A step of selecting, wherein each kernel image (e.g., 310a to i) of the kernel image library (300) is to be added to the reconstruction grid (400) of the reconstructed image (500) as a pre-calculated mosaic, and a plurality of selected added mosaics can mosaic and render the reconstructed image (500); a reconstructed grid vector position centered on the scan grid vector position that generated the exit wave (120a) causing the selected test event detection coefficients (210a, 210b, 210c) A method comprising: a step of adding each selected kernel image (310b, 310e, 310h) to the reconstruction grid (400) in a set of; and a step of repeating the transmission, reception, selection, and addition steps, wherein different scan grid vector positions (100a, 100b, 100c) are used in at least one iteration to progressively mosaic and render the reconstruction image (500). Claim 13 A method comprising the subject of a claim in any one of paragraphs 2 through 10, in paragraph 12. Claim 14 A method that is a computer-implemented method in any one of paragraphs 1 through 13. Claim 15 A computing device or system configured to perform the method of paragraph 14. Claim 16 A computer program or computer program product having instructions that cause said computing device or system to perform the method of claim 14 when executed by said computing device or system. Claim 17 A computer-readable medium storing a computer program having instructions that, when executed by a computing device or system, cause said computing device or system to perform the method of claim 14.

Images