Autofocus methods for single-beam and multi-beam systems

Abstract

With the first astigmatism applied, the electron beam is guided to the workpiece on the stage. The electron beam reflected from the workpiece is detected, and an image of the workpiece is generated with the first astigmatism applied. An autofocus model is applied to the image, and the autofocus model determines the focus offset relative to the image.

Description

Technical Field

[0001] This disclosure relates to an electron beam system.

Background Art

[0002] With the development of the semiconductor manufacturing industry, the requirements for yield management, particularly measurement and inspection systems, have been increasing significantly. Although the critical dimensions continue to shrink, the industry requires time reduction to achieve high-yield and high-value production. By minimizing the total time from detecting a yield problem to correcting it, the return on investment for semiconductor manufacturers is maximized.

[0003] Fabricating semiconductor devices such as logic and memory devices typically involves processing semiconductor wafers using a number of fabrication processes to form various features and multiple levels of the semiconductor device. For example, lithography is a semiconductor fabrication process that involves transferring a pattern from a reticle to a photoresist disposed on a semiconductor wafer. Further examples of semiconductor fabrication processes include, but are not limited to, chemical mechanical polishing (CMP), etching, deposition, and ion implantation. The placement of multiple semiconductor devices fabricated on a single semiconductor wafer can be separated into individual semiconductor devices.

[0004] An optical system can be used to inspect a workpiece. However, some of the dimensions of the features of the workpiece, or the size of the defects to be identified, may be below the optical resolution, and some of the defects may become undetectable by the optical system. To inspect the surface of the workpiece in this situation, charged particles such as electrons can be used. This can provide better resolution than an optical system.

[0005] In inspection and other imaging applications, such as those in the semiconductor industry, processing capacity is a critical consideration. Proper focusing of a focused electron beam onto a workpiece improves imaging and inspection quality. Some systems offer semi-automatic focusing, which includes user assistance with focusing. Some autofocus concepts have also been described in relation to fully automated focusing of the beam onto the workpiece. The time used to automatically focus the charged particle beam increases the total time for imaging areas on the workpiece, reducing processing capacity. This may be particularly relevant to imaging with multi-beam inspection tools where multiple beamlets are focused onto the workpiece.

[0006] The quality of electron beam images is degraded by factors such as astigmatism and defocusing of the electron beam system, leading to reduced resolution. Poor quality electron beam images result in decreased inspection sensitivity and performance of length measurements. In addition, such images exhibit pattern width variations, making it difficult to reliably detect image edges. Measurements of dimensions with such poor quality beams are often unsatisfactory.

[0007] Conventionally, the focus and astigmatism of electron beam optics may be adjusted by changing the control current of the objective lens while visually observing the electron beam image. The focus may be adjusted by varying the beam focusing height. The beam focusing height is changed by adjusting the current flowing through the objective lens.

[0008] Conventional image-based autofocusing methods for electron beam systems may involve stepping a focus knob (e.g., lens current or voltage) through a series of focus points encompassing both sides of the optimal focal plane. An image is captured at each step. A sharpness score is calculated from each step image. The dataset of sharpness scores versus focus knob values ​​is approximated by a curve, and the point from which the maximum score is derived is identified. The focus knob value corresponding to the best derived sharpness score is then assigned to the lens power supply as the new focal value.

[0009] This conventional method has processing limitations. Acquiring the desired focus value involves stepping the power supply and acquiring multiple images. This entire operation requires a time proportional to the number of steps in the autofocus sequence. Finely stepped autofocus has smaller step sizes and more steps, requiring longer operation times. In many automated, high-load industrial applications of electron microscopy, autofocus operations are performed periodically and frequently. The time required for autofocus operations directly impacts the system's processing capacity for these applications.

[0010] This conventional method also cannot be implemented in real-time continuous applications such as sweeping. In sweeping operations, the stage moves continuously while the electron beam apparatus continuously acquires images. Continuously tracking the image focus and correcting it in real time, as is done using a servo control loop, is impossible with conventional autofocusing methods due to the length of time required to complete the autofocusing process. [Prior art documents] [Patent Documents]

[0011] [Patent Document 1] U.S. Patent Application Publication No. 2019 / 0004298 [Patent Document 2] U.S. Patent Application Publication No. 2020 / 0365364 [Overview of the project] [Problems that the invention aims to solve]

[0012] Improved systems and techniques are needed. [Means for solving the problem]

[0013] An electron beam system is provided in a first embodiment. The electron beam system includes an electron beam source that generates at least one electron beam, a stage configured to hold a workpiece in the path of the electron beam, at least one lens located in the path of the electron beam and configured to control astigmatism of the electron beam, a detector configured to image the workpiece using the electron beam, and a processor that electronically communicates with the detector and the electron beam source. The processor is configured to generate an image of the workpiece with a first astigmatism applied, to apply an autofocus model to the image, and to determine a focus offset relative to the image using the autofocus model.

[0014] The amount and direction of the first astigmatism can be the same as those used during training of the autofocus model.

[0015] The processor may include at least one graphics processing unit (GPU).

[0016] The autofocus model can be trained using at least one model for a device on a workpiece.

[0017] In one example, only one of the images with the first astigmatism is used to determine the focus offset.

[0018] The electron beam source can generate three or more electron beams.

[0019] The processor can be further configured to send instructions to a lens for adjusting a focus offset. The lens can be an objective lens. The processor can also be further configured to send instructions for generating a second image of the workpiece after the lens has adjusted the focus offset.

[0020] A method is provided in a second embodiment. This method includes directing an electron beam onto a workpiece on a stage. A first spherical aberration is applied to the electron beam. A detector is used to detect the electron beam reflected from the workpiece. A processor is used to generate an image of the workpiece with the first spherical aberration applied. A autofocus model is applied to the image using the processor. A focus offset for the image is determined by the autofocus model using the processor.

[0021] This method can further include training the autofocus model by at least one model for a device on the workpiece.

[0022] The amount and direction of the first spherical aberration can be the same as those used during the training of the autofocus model.

[0023] In one example, only the image with the first spherical aberration is used to determine the focus offset.

[0024] This method can further include using a processor to send instructions for adjusting a focus offset in the electron beam. This method can also include generating a second image of the workpiece after the focus offset has been adjusted.

[0025] Directing can include three or more electron beams. The stage can move at a constant speed during the directing. The stage can also move before the generation.

[0026] A non-transitory computer-readable storage medium is provided in a third embodiment. The non-transitory computer-readable storage medium includes one or more programs for performing the following steps on one or more computing devices. An image of the workpiece is generated with a first aberration applied. An autofocus model is applied to the image. Using the autofocus model, a focus offset for the image is determined.

[0027] The step can include sending an instruction to adjust the focus offset.

[0028] The amount and direction of the first aberration can be the same as those used during the training of the autofocus model.

[0029] To more fully understand the nature and objects of the present disclosure, please refer to the following detailed description in conjunction with the accompanying drawings.

Brief Description of the Drawings

[0030] [Figure 1] It is a diagram of a focus series of a circular beam. [Figure 2] It is a diagram of a focus series of an aberration beam. [Figure 3] It is a flowchart of an embodiment of an autofocus method according to the present disclosure. [Figure 4] It is a diagram showing real-time autofocus during scanning for a multi-beam system. [Figure 5] It is a flowchart of an embodiment of an autofocus method for multi-beam scanning according to the present disclosure. [Figure 6] It is a block diagram of an electron beam system according to the present disclosure.

Modes for Carrying Out the Invention

[0031] While specific embodiments of the claimed subject matter are described herein, other embodiments, including those that do not provide all of the benefits and features described herein, are also within the scope of this disclosure. Various structural, logical, process steps, and electronic modifications can be made without departing from the scope of this disclosure. Therefore, the scope of this disclosure is defined solely by reference to the appended claims.

[0032] Embodiments disclosed herein provide image-based methods for increasing autofocus speed for single and multi-beam electron beam systems. A single-shot image-based autofocus method uses an astigmatic beam to generate an asymmetric focus series. A machine learning-based method can infer focus offsets based on out-of-focus astigmatic images. An auxiliary autofocus beam can be added to the primary inspection beam array to perform real-time autofocus during sweeping. The same or varying astigmatic ("stig") values ​​can be added to the autofocus beam to increase robustness to the focusing result.

[0033] Embodiments disclosed herein can speed up autofocus operation by reducing the number of imaging steps. A training step during a set time is used to train the autofocus model. In the training step, a series of focused images from various target locations on the workpiece are acquired. Training can be performed within a user-care area to capture characteristic pattern features that may be encountered later during operation, as relevant to the imaging area of ​​interest.

[0034] The focus knob is stepped to obtain the image's focal series. For example, the lens power value can be stepped. A known, controlled amount of astigmatism is added to the electron beam during the stepping of the focus knob. The focal series of a circular beam is unaffected by which side of the imaginary focal plane it lies on (Figure 1), while the focal series of an astigmatic beam (Figure 2) is asymmetrical depending on whether the plane is above or below the imaginary focal plane. Thus, the sign of the focal offset can be determined.

[0035] The image focus series can be specific to a particular workpiece, a particular device on a workpiece, or one or more devices within a particular care area. The image focus series can also be more common to the relevant type of semiconductor device (e.g., logic or memory devices). An autofocus model may include images with multiple different patterns, patterns specific to a particular manufacturer or manufacturing facility ("fab"), or patterns specific to the particular device being inspected.

[0036] The image focus series can include images of the actual device. The image focus series can also include composite images. If design pattern files are available from the semiconductor manufacturer, the model can be trained by simulating out-of-focus, astigmatic images without actually imaging the wafer.

[0037] The number of images in the image focus series can be varied depending on the wafer pattern. If the object has a very repeating pattern, such as a memory array, 1 to 3 locations may be required. If it contains more unique features, such as within a logical device, it may be necessary to obtain training data from 5 to 10 unique structures. A point spread function can be extracted from these images, and therefore the sharpness calculation does not depend on the pattern itself.

[0038] For example, if the user selects a care area, training images can be automatically captured. These images can be specific to the care area that will be later examined during the procedure. The captured images can become the focus series of images.

[0039] The astigmatic focus series of an image is assigned to a set of known focus knob values. Typically, the knob value controls a focusing lens, which can be a magnetic lens controlled by lens current or an electrostatic lens controlled by lens voltage. When the knob is adjusted, this means that the current or voltage of that lens is adjusted.

[0040] Next, image-knob value pairs are provided to the machine learning training model to associate individual images with knob values. Alternatively, instead of sending image-knob value pairs to the training model, the point spreading function (PSF) can be extracted first from the focus series images. These PSF-knob value pairs can then be provided to the machine learning training model. The goal of the machine learning training model is to enable it to infer focus knob values ​​when a single out-of-focus astigmatic image is provided during runtime, also known as single-shot autofocus. After the model training is complete, the autofocus is ready to operate.

[0041] In one example, the knob value is stepped continuously. For instance, the objective lens voltage can be stepped.

[0042] Astigmatism can be controlled using voltage in electrostatic astigmatism correction devices, or using current in magnetic astigmatism correction devices. The astigmatism value for voltage can be between 1 and 10V. The astigmatism value for current can be between 0.1 and 10mA. Other values ​​are possible, and these are merely examples.

[0043] Figure 3 is a flowchart of Method 100, which can be used for step-and-scan autofocus. Method 100 uses a trained autofocus model. Some of the steps in Method 100 can be performed using a processor. During the step-and-scan automated imaging operation, the stage moves, autofocus is performed, and image acquisition is performed. This process can be repeated one or more times across the workpiece.

[0044] In 101, the electron beam is directed onto a workpiece such as a semiconductor wafer. The electron beam is subjected to a first astigmatism. In 102, the electron beam reflected from the surface of the workpiece is detected using a detector. Then, in 103, an image of the workpiece is generated. The image of the workpiece uses an electron beam with the first astigmatism. This image can be a single image of the workpiece used to determine the focus offset. The first astigmatism can help determine whether the image is overfocused or underfocused.

[0045] The first astigmatism may be large enough so that the focal series can show the difference between above and below the optimal focal plane. However, the first astigmatism cannot be so large that the focal series is completely distorted by the astigmatism. The astigmatism value can be selected to ensure that a reliable focal series is created to produce a consistent focal score or point spreading function. This can be done empirically or by image simulation.

[0046] When the electron-optical conditions (which determine the ray paths of electrons traveling through the microscope tube) are calibrated, a given astigmatism and point spreading function relative to the focal point can be simulated. If it is known that a certain amount of astigmatism can generate a good autofocus training series, the voltage required to produce such astigmatism in the electron-optical system can be determined.

[0047] In step 104, an autofocus model is applied to the image. This can use a trained autofocus model 108 that has been trained using an astigmatism focus series of the image. In one example, a deep learning neural network can be used with the autofocus model. The astigmatism focus series of the image can cover or include a model of a device on a workpiece imaged using method 100. The amount and direction of the first astigmatism are the same as those used during training the autofocus model.

[0048] As a result of the training step, the training output is an autofocus model that pairs out-of-focus astigmatic images with known focus offsets. In the inference step, the input to the model is an out-of-focus image with known astigmatic aberration and an unknown amount of defocus, and the output of the model is the focus offset.

[0049] This image can be sent to a high-speed machine learning processor (e.g., a GPU) with minimal latency. Such a low-latency potential data path can be a coupling between data acquisition devices directly connected to a GPU on the same Peripheral Interconnection (PCI) bus, without routing through the system's central processing unit (CPU) or system memory. This type of coupling can be called direct memory access (DMA). In one example, both a data acquisition field-programmable gate array (FPGA) and a data processing GPU are integrated on the same printed circuit board (PCB) by an optimized direct interconnect between them. The machine learning processor can be the processor used in other steps of Method 100, or it can be a separate processor.

[0050] Since the time from imaging to focusing results affects the processing load, shorter processing times are generally preferred. The time to results can be less than 1 / 10 of the image acquisition time, which is typically 100 microseconds.

[0051] In step 105, the focus offset is determined for the image. An autofocus model operating within the GPU processor can output the focus offset corresponding to the input image. This method can achieve the desired focused image, capturing only a single autofocus image. For robustness, the autofocus process can be repeated to improve accuracy (as shown in Figure 3). Repeating steps 104-106 provides iterative process control.

[0052] In the training model, the point spreading function extracted from each step image is paired with a known focal offset. In the interference step, the point spreading function is first extracted from the runtime astigmatism image and matched with the PSF of a series of training images by this model. Then, the training image with the PSF closest to the runtime image is selected as the image with the closest focal offset condition. The corresponding focal offset is output as a result.

[0053] Next, in step 106, the focus offset is applied to the electron beam system. This can be a signed Z focus offset. Commands can be sent to components within the electron beam system to adjust the focus offset. For example, the lens value can be updated. The focus offset can be applied to the same knob from which the training series was acquired.

[0054] "Signed Z" refers to a signed Z offset of ±. Z=0 means the lens is at optimal focus. +Z indicates being Zμm above optimal focus, and -Z indicates being Zμm below optimal focus. A lens sensitivity table may be available to convert Z in μm units to lens control current or voltage. By using Z in μm units as the focus offset, the actual lens current / voltage can be removed from the algorithm, and therefore the algorithm does not need to know which lens is used to achieve a particular focus offset distance Z, as long as such a focus offset distance Z is achieved.

[0055] In 107, image acquisition is performed using the autofocus result, without the first astigmatism. This image acquisition may include detecting the electron beam reflected from the surface of the workpiece and generating an image of the workpiece. Thus, a second image of the workpiece can be generated after the focus offset has been adjusted. Because the focus offset has been adjusted, the image acquisition in 107 can be considered in focus.

[0056] Method 100 can repeat steps 104-106, but it can also operate from a single image. Therefore, only images from step 103 may be used. Existing autofocus techniques involve 6-10 steps, but these can be replaced with a single step using the embodiments disclosed herein. Therefore, the embodiments disclosed herein can be 6-10 times faster than previous techniques.

[0057] Some embodiments may utilize three or more electron beams. The stage may move at a constant speed while the electron beams are guided onto the workpiece, or the stage may move between imaging steps of the workpiece (for example, the stage may move before generating an image).

[0058] In a multi-beam system, the imaging system has two or more beams or beamlets for simultaneous image acquisition. Embodiments of autofocus methods disclosed herein can be applied to step-and-scan or sweep operations. In step-and-scan operation, the stage moves and stops alternately. In sweep operation, the stage moves continuously at a constant speed while one or more beams perform a line scan over a region of interest. The simultaneous movement of the stage and the line scanning of the beams create a 2D raster scan image.

[0059] In a multi-beam system, in addition to the main beam array used for inspection imaging, several auxiliary beam arrays can be placed alongside the main array for autofocus purposes. As shown in Figure 4, two auxiliary beam arrays are positioned on either side of the main inspection beam array. These arrays are called autofocus beam arrays and are used for autofocus purposes. Astigmatism is always applied to the autofocus beam arrays while acquiring images.

[0060] When the beam array is sweeping from left to right (i.e., the beam array is stationary and the stage is sweeping from right to left), the right autofocus (AF) array is used. Images acquired by the beam in this autofocus array can be sent to an autofocus model trained during the set time specified herein, allowing the focus offset value to be updated in real time. Similarly, when the beam array is sweeping from right to left, the left autofocus array can be used.

[0061] Since the autofocus is positioned to the right of the inspection array, it first checks for focus changes in front of the inspection array, giving the inspection array time to adjust its focus based on the results discovered by the autofocus beam array. Therefore, the autofocus array can see the focus variations in advance.

[0062] The same astigmatism value can be applied to all beams in an autofocus array. Each beam can generate a focus score, and the average focus score from all autofocus beams can be used as the final score. The final score can be the focus offset applied to the beamlet used for imaging. The final score can be applied collectively to the focus of the entire beam array. This method increases the redundancy of the autofocus process when encountering blank or featureless regions.

[0063] In another example, a different astigmatism value can be applied to each beam in an autofocus array. Each astigmatism value corresponds to a unique training dataset (trained with this astigmatism value applied to the beam). This approach increases the redundancy of the machine learning model to prevent one model from overestimating. In this case as well, the average result is output as the final focus score.

[0064] Such an autofocus sequence for multi-beam sweeping is shown in Figure 5, where multiple autofocus images are constantly acquired and sent to the autofocus model, and equal or varying astigmatism is applied to the autofocus beams. Focus offsets from this model can be applied sequentially to the entire beam array (including the autofocus beam array and the inspection beam array). Steps 201-207 are similar to steps 101-107, except that multiple electron beams are used. The trained autofocus model 208 may be the same as the trained autofocus model 108. In step 201, the electron beams may have the same or different astigmatism values.

[0065] Figure 6 is a block diagram of one embodiment of the electron beam system 300. The electron beam system 300 includes a workpiece inspection system (including an electron column 301) configured to generate an image of a workpiece 304 such as a semiconductor wafer.

[0066] The workpiece inspection system includes an output acquisition subsystem that includes at least an energy source and a detector. The output acquisition subsystem may be an electron beam-based output acquisition subsystem. For example, in one embodiment, the energy induced to the workpiece 304 includes electrons, and the energy detected from the workpiece 304 also includes electrons. Thus, the energy source may be an electron beam source. In one such embodiment shown in Figure 6, the output acquisition subsystem includes an electron column 301, which is coupled to a computer subsystem 302. A stage 310 can hold the workpiece 304. The stage 310 can be translated.

[0067] As also shown in Figure 6, the electron column 301 includes an electron beam source 303 configured to generate electrons that are focused onto the workpiece 304 by one or more elements 305. The electron beam source 303 may include, for example, a cathode source or an emitter tip. The one or more elements 305 may include, for example, a gun lens, an anode, a beam aperture diaphragm, a gate valve, a beam current selective aperture, an objective lens, and a scanning subsystem, all of which may include any such suitable elements known in the art.

[0068] Electrons returned from the workpiece 304 (e.g., secondary electrons) can be focused to the detector 307 by one or more elements 306. One or more elements 306 may include, for example, a scanning subsystem, which may be the same scanning subsystem as that included in element 305.

[0069] The electron column 301 may also include any other suitable elements known in the art.

[0070] Figure 6 shows the electron column 301 configured such that electrons are guided to the workpiece 304 at an oblique angle of incidence and scattered from the workpiece 304 at another oblique angle; however, the electron beam may be guided to the workpiece 304 at any preferred angle and scattered from the workpiece 304. In addition, the electron beam-based output acquisition subsystem may be configured to generate an image of the workpiece 304 using multiple modes (e.g., different irradiation angles, focusing angles, etc.). The multiple modes of the electron beam-based output acquisition subsystem may differ in any of the image generation parameters of the output acquisition subsystem.

[0071] The computer subsystem 302 may be coupled to the detector 307 as described above. The detector 307 can detect electrons returned from the surface of the workpiece 304, thereby forming an electron beam image of the workpiece 304. The electron beam image may include any preferred electron beam image. The computer subsystem 302 may be configured to perform any of the functions described herein using the output of the detector 307 and / or the electron beam image. The computer subsystem 302 may be configured to perform any additional steps described herein. The system 300, including the output acquisition subsystem shown in Figure 6, may be further configured as described herein.

[0072] Note that Figure 6 is provided herein to show the overall configuration of an electron beam-based power acquisition subsystem that can be used in the embodiments described herein. The electron beam-based power acquisition subsystem configuration described herein may be modified to optimize the performance of the power acquisition subsystem, as is commonly done when designing commercially available power acquisition systems. In addition, the systems described herein may be implemented using existing systems (for example, by adding the functions described herein to an existing system). In some such systems, the methods described herein may be provided as optional functions of the system (for example, in addition to other functions of the system). Alternatively, the systems described herein may be designed as entirely new systems.

[0073] While the output acquisition subsystem was described above as an electron beam-based output acquisition subsystem, the output acquisition subsystem may also be an ion beam-based output acquisition subsystem. Such an output acquisition subsystem may be configured as shown in Figure 6, except that the electron beam source can be replaced with any suitable ion beam source known in the art. In addition, the output acquisition subsystem may also be any other suitable ion beam-based output acquisition subsystem, such as those included in commercially available focused ion beam (FIB) systems, helium ion microscope (HIM) systems, and secondary ion mass spectrometry (SIMS) systems.

[0074] The computer subsystem 302 includes a processor 308 and an electronic data storage unit 309. The processor 308 may include a microprocessor, a microcontroller, or other device. For example, the processor 308 may include a GPU, or may be a GPU.

[0075] The computer subsystem 302 may be coupled to components of system 300 in any preferred manner (for example, via one or more transmission media, which may include wired and / or wireless transmission media) so that the processor 308 can receive its output. The processor 308 may be configured to use this output to perform multiple functions. The workpiece inspection system may receive instructions or other information from the processor 308. The processor 308 and / or the electronic data storage unit 309 may optionally communicate electronically with another workpiece inspection system, workpiece measurement system, or workpiece reinspection system (not shown) to receive additional information or send instructions.

[0076] The processor 308 communicates electronically with the workpiece inspection system, such as the detector 307, or other components within it. The processor 308 may be configured to process images generated using measurements from the detector 307. For example, the processor may implement embodiments of Method 100 or Method 200.

[0077] The computer subsystem 302 described herein, other systems, or other subsystems may be part of a variety of systems, including personal computer systems, image computers, mainframe computer systems, workstations, network appliances, internet appliances, or other devices. A subsystem or system may also include any suitable processor known in the art, such as a parallel processor. In addition, a subsystem or system may include a platform with high-speed processing and software, either as a standalone or networked tool.

[0078] The processor 308 and the electronic data storage unit 309 may be located within the system 300 or another device, or they may be part of the system 300 or another device in other ways. For example, the processor 308 and the electronic data storage unit 309 may be part of a standalone control unit or they may be located within a centralized quality control unit. Multiple processors 308 or electronic data storage units 309 may be used.

[0079] The processor 308 may, in practice, be implemented by any combination of hardware, software, and firmware. Furthermore, its functions as described herein may be implemented by a single unit or divided among different components, each of which may be implemented by any combination of hardware, software, and firmware. Program code or instructions for the processor 308 to implement various methods and functions may be stored in a readable storage medium such as memory in the electronic data storage unit 309 or other memory.

[0080] If the system 300 includes two or more computer subsystems 302, different subsystems may be coupled to one another so that images, data, information, instructions, etc., can be transmitted between subsystems. For example, one subsystem may be coupled to an additional subsystem by any suitable transmission medium, such transmission medium may include any suitable wired and / or wireless transmission medium known in the art. Two or more such subsystems may be substantially coupled by a shared computer-readable storage medium (not shown).

[0081] The processor 308 may be configured to perform multiple functions using the outputs of the system 300 or other outputs. For example, the processor 308 may be configured to send outputs to the electronic data storage unit 309 or another storage medium. The processor 308 may be further configured as described herein.

[0082] The processor 308 or computer subsystem 302 may be part of a defect re-inspection system, inspection system, measurement system, or any other type of system. Accordingly, the embodiments disclosed herein describe several configurations that can be prepared in multiple forms for systems having different capabilities that are more or less suitable for different applications.

[0083] The processor 308 may be configured according to any of the embodiments described herein. The processor 308 may also be configured to perform other functions or additional steps using the output of the system 300 or using images or data from other sources.

[0084] The processor 308 may be communicatively coupled to any of the various components or subsystems of system 300 in any form known in the art. Furthermore, the processor 308 may be configured to receive and / or acquire data or information from other systems (e.g., test results from an inspection system such as a retesting tool, a remote database containing design data, etc.) via a transmission medium which may include wired and / or wireless portions. In this way, the transmission medium may act as a data link between the processor 308 and other subsystems of system 300 or systems outside of system 300.

[0085] Various steps, functions, and / or operations of the System 300 and methods disclosed herein are performed by one or more of the following: electronic circuits, logic gates, multiplexers, programmable logic devices, ASICs, analog or digital controls / switches, microcontrollers, or computing systems. Program instruction implementation methods, such as those described herein, may be transmitted over a carrier medium or stored on a carrier medium. The carrier medium may include storage media such as read-only memory, random-access memory, magnetic or optical disks, non-volatile memory, solid memory, and magnetic tape. The carrier medium may include transmission media such as wires, cables, or wireless transmission links. For example, various steps described throughout this disclosure may be performed by a single processor 308 (or computer subsystem 302), or alternatively, by multiple processors 308 (or multiple computer subsystems 302). Furthermore, different subsystems of System 300 may include one or more computing or logic systems. Therefore, the above description should be construed as illustrative and not as a limitation to this disclosure.

[0086] In one example, the processor 308 communicates with the electron beam system 300. The processor 308 is configured to generate an image of the workpiece 304 with a first astigmatism applied, apply an autofocus model to the image so that a Z focus offset is applied to the image, and determine the focus offset relative to the image. The processor 308 may be further configured to send a command for the lens in the electron beam system 300 to adjust the focus offset, and after the lens has adjusted the focus offset, to generate a second image of the workpiece 304.

[0087] Additional embodiments, as disclosed herein, relate to a non-temporary computer-readable medium for storing program instructions executable on a controller to implement a computer-implemented autofocus method. In particular, as shown in Figure 6, the electronic data storage unit 309 or other storage medium may include a non-temporary computer-readable medium containing program instructions executable on a processor 308. The computer implementation method may include any step of any of the methods described herein, including method 100 or method 200. The electronic data storage unit 309 may include a trained autofocus model.

[0088] While this disclosure describes one or more specific embodiments, it will be understood that other embodiments of this disclosure may be made without departing from the scope of this disclosure. Therefore, this disclosure is deemed to be limited only by the appended claims and their reasonable interpretation.

Claims

1. An electron beam source that generates at least one electron beam, A stage configured to hold a workpiece within the path of the electron beam, At least one lens located within the path of the electron beam and configured to control the astigmatism of the electron beam, A detector configured to image the workpiece using the electron beam, The system comprises a processor that communicates electronically with the detector and the electron beam source, and the processor An image of the workpiece is generated with the first astigmatism applied. Applying an autofocus model to the aforementioned image, The system is configured to determine the focus offset for the image using the aforementioned autofocus model. An electron beam system characterized by the following features.

2. An electron beam system according to claim 1, characterized in that the amount and direction of the first astigmatism are the same as those used during training of the autofocus model.

3. An electron beam system according to claim 1, wherein the processor includes at least one GPU.

4. An electron beam system according to claim 1, wherein the autofocus model is trained using at least one model for the workpiece device.

5. An electron beam system according to claim 1, characterized in that only one of the images having the first astigmatism is used to determine the focal offset.

6. An electron beam system according to claim 1, characterized in that the electron beam source generates three or more electron beams.

7. An electron beam system according to claim 1, wherein the processor is further configured to transmit commands to the lens for adjusting the focal offset, and the lens is an objective lens.

8. An electron beam system according to claim 7, wherein the processor is further configured to transmit a command for generating a second image of the workpiece after the lens has adjusted the focal offset.

9. The electron beam is guided onto the workpiece on the stage, and a first astigmatism is applied to the electron beam. Using a detector, detect the electron beam reflected from the workpiece, Using a processor, generate an image of the workpiece with the first astigmatism applied, Using the aforementioned processor, an autofocus model is applied to the image, Using the aforementioned processor, the focus offset for the image is determined by the autofocus model. A method characterized by including the following.

10. A method according to claim 9, further comprising training the autofocus model with at least one model for a device on the workpiece.

11. A method according to claim 10, characterized in that the amount and direction of the first astigmatism are the same as those used during the training of the autofocus model.

12. A method according to claim 9, characterized in that only the image having the first astigmatism is used to determine the focus offset.

13. A method according to claim 9, further comprising using the processor to transmit an instruction for adjusting the focus offset in the electron beam.

14. A method according to claim 13, further comprising generating a second image of the workpiece after the focus offset has been adjusted.

15. A method according to claim 9, characterized in that the induction includes three or more electron beams.

16. A method according to claim 15, characterized in that the stage moves at a constant speed during induction.

17. A method according to claim 15, characterized in that the stage moves before the generation.

18. On one or more computing devices, A step of generating an image of the workpiece with a first astigmatism applied, The steps include applying an autofocus model to the aforementioned image, A non-temporary computer-readable storage medium, comprising one or more programs for performing the step of determining a focus offset for an image using the autofocus model.

19. A non-temporary computer-readable storage medium according to claim 18, wherein the step includes transmitting an instruction to adjust the focus offset.

20. A non-temporary computer-readable storage medium according to claim 18, characterized in that the amount and direction of the first astigmatism are the same as those used during training of the autofocus model.

Images