Data processing device and method, charged particle assessment system and method

Abstract

A data processing device for detecting defects in sample image data generated by a charged particle assessment system, the device comprising: a first processing module configured to receive a sample image datastream from the charged particle assessment system, the sample image datastream comprising an ordered series of data points representing an image of the sample, and to apply a first defect detection test to select a subset of the sample image datastream as first selected data, wherein the first defect detection test is a localised test which is performed in parallel with receipt of the sample image datastream; and a second processing module configured to receive the first selected data and to apply a second defect detection test to select a subset of the first selected data as second selected data.

Description

[Technical Field]

[0001] The embodiments provided herein are generally related to data processing apparatus and methods, particularly data processing apparatus and methods used in or for use in charged particle evaluation systems, and methods for operating charged particle evaluation systems. [Previous Technology]

[0002] In the manufacture of semiconductor integrated circuit (IC) wafers, undesired pattern defects caused by factors such as optical effects and accidental particles inevitably occur on the substrate (i.e., wafer) or mask during the manufacturing process, thereby reducing yield. Therefore, monitoring the extent of undesired pattern defects is an important process in IC wafer manufacturing. More generally, the inspection and / or measurement of the surface of substrates or other objects / materials is an important process during and / or after their manufacture.

[0003] Pattern detection equipment with charged particle beams has been used to inspect objects, referred to as samples, for example, to detect pattern defects. These devices typically employ electron microscopy techniques, such as scanning electron microscopy (SEM). In SEM, a primary electron beam, oriented at relatively high energy, is used in a final deceleration step to land on the sample with relatively low landing energy. The electron beam is focused as a probe spot on the sample. The interaction between the material structure at the probe spot and the landing electrons from the electron beam causes signal electrons, such as secondary electrons, backscattered electrons, or Auger electrons, to be emitted from the surface. Signal electrons can be emitted from the material structure of the sample. By scanning the sample surface with the primary electron beam as a probe spot, signal electrons can be emitted across the sample surface. By collecting these emitted signal electrons from the sample surface, the pattern detection equipment can obtain an image representing the characteristics of the material structure of the sample surface.

[0004] When pattern inspection equipment is used to detect defects on samples at high throughput, a very large amount of image data is generated, and this image data must be processed to detect defects. Specifically, it is necessary to reduce noise from the image data. US 8,712,184 B1 and U 9,436,985 B1 describe methods for reducing noise or improving the signal-to-noise ratio in images acquired by a scanning electron microscope. In some cases, the data generation rate may be too high for real-time processing without an excessively high processing power, and prior art methods are not easily optimized for high-speed processing of such large amounts of data, such as real-time processing. Therefore, data processing can be an undesirable limitation on the throughput of pattern inspection equipment. [Summary of the Invention]

[0005] One objective of the present invention is to provide an embodiment for reducing the computational cost of processing images generated by charged particle evaluation equipment to detect defects.

[0006] According to a first state, a data processing apparatus is provided for detecting defects in sample image data generated by a charged particle evaluation system. The apparatus includes: a first processing module configured to receive a sample image data stream from the charged particle evaluation system, the sample image data stream including a series of ordered data points representing an image of the sample, and the first processing module applying a first defect detection test to select a subset of the sample image data stream as first selected data, wherein the first defect detection test is a localization test performed in parallel with receiving the sample image data stream; and a second processing module configured to receive the first selected data and applying a second defect detection test to select a subset of the first selected data as second selected data.

[0007] According to a second aspect of the present invention, a charged particle evaluation system is provided, which includes a charged particle beam system and a data processing device as described above.

[0008] According to a third aspect of the present invention, a method is provided for detecting defects in a sample having a plurality of pattern repeating regions using a charged particle beam system. The method comprises, in sequence: a first scan of a first portion of a first region of the pattern repeating regions to generate first scan image data; a second scan of a first portion of a second region of the pattern repeating regions to generate second scan image data, the second region being spaced apart from the first region; a third scan of a first portion of a third region of the pattern repeating regions to generate third scan image data, the third region being spaced apart from the first region and the second region; and a subsequent scan of a second portion of the first region to generate fourth scan image data.

[0009] According to a fourth aspect of the present invention, a method is provided for detecting defects in a sample using a charged particle beam system with one charged particle multi-beam comprising a plurality of sub-beams, the sample having a plurality of pattern repeating regions nominally being identical regions. The method comprises, in sequence: using the charged particle multi-beam to perform a first scan of one of the first pattern repeating regions of the sample to generate a first scan image data; using the charged particle multi-beam to perform a second scan of one of the second pattern repeating regions of the sample to generate a second scan image data, the second pattern repeating region being spaced apart from the first pattern repeating region; and using the charged particle multi-beam to perform a third scan of one of the third pattern repeating regions of the sample to generate a second scan image data. A third scan image data is generated, the third pattern repeating region being spaced apart from the first pattern repeating region and the second pattern repeating region; and the first scan image data, the second scan image data, and the third scan image data are processed by comparing the first scan image data, the second scan image data, and the third scan image data to identify a defect; wherein each sub-beam is assigned a portion of the first pattern repeating region, the second pattern repeating region, and the third pattern repeating region, such that each scan includes scanning corresponding portions of the respective pattern repeating regions using the same sub-beam; wherein the processing includes comparing scan image data obtained from corresponding portions of the respective pattern repeating regions using the same sub-beam.

[0010] According to a fifth aspect of the present invention, a method is provided for detecting defects in a sample using a charged particle beam system with one charged particle multi-beam comprising a plurality of sub-beams, the sample having a plurality of nominally identical pattern repeating regions, the method comprising: continuously scanning at least three different pattern repeating regions of the sample using the charged particle multi-beam to generate different first sample image data sets; processing the different first sample image data sets by comparing the image data of each region to identify a defect; wherein each sub-beam is assigned a corresponding portion of each of the pattern repeating regions such that each scan comprises scanning the corresponding portion of the respective pattern repeating regions using the same sub-beam.

[0011] According to a sixth aspect of the present invention, a data processing method for detecting defects in sample image data generated by a charged particle evaluation system is provided. The method includes: receiving a sample image data stream from the charged particle evaluation system, the sample image data stream including a series of ordered data points representing an image of the sample; applying it as a first defect detection test as part of a localization test, the application of the first detection test including selecting a subset of the sample image data stream as first selected data, wherein the application of the first defect detection test is performed in parallel with receiving the sample image data stream; receiving the first selected data; and applying a second defect detection test, the application of the second detection test including selecting a subset of the first selected data as second selected data.

Implementation Method

[0034] Detailed reference will now be made to exemplary embodiments, examples of which are illustrated in the accompanying drawings. The following description refers to the accompanying drawings, wherein, unless otherwise indicated, the same reference numerals in different figures denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments of the invention. In fact, they are merely examples of apparatuses and methods conforming to the nature of the invention as enumerated in the appended claims.

[0035] Enhanced computing power in electronic devices can be achieved by significantly increasing the packing density of circuit components (such as transistors, capacitors, diodes, etc.) on IC chips, thereby reducing the physical size of the device. This has been achieved through increased resolution, enabling the fabrication of smaller structures. For example, a smartphone IC chip (about the size of a thumbnail and available in 2019 or earlier) could include more than 2 billion transistors, each smaller than 1 / 1000th the size of a human hair. Therefore, it is not surprising that semiconductor IC manufacturing involves a complex and time-consuming process with hundreds of individual steps. Even an error in one step can significantly affect the functionality of the final product. A single "fatal defect" can cause device failure. The goal of the manufacturing process is to improve the overall yield of the process. For example, to achieve a 75% yield in a 50-step process (where steps can indicate the number of layers formed on the wafer), the yield of each individual step must be higher than 99.4%. If each individual step has a yield of 95%, the overall process yield will be as low as 7%.

[0036] While high process yields are desirable in IC chip manufacturing facilities, maintaining high substrate (i.e., wafer) output (defined as the number of substrates processed per hour) is also essential. High process yields and high substrate outputs can be affected by the presence of defects. This is especially true if operator intervention is required to inspect for defects. Therefore, high-volume detection and identification of micron and nanometer-level defects by inspection equipment (such as scanning electron microscopes ("SEM")) is crucial for maintaining high yields and low costs.

[0037] A SEM includes a scanning device and a detector device. The scanning device includes: an illumination device containing an electron source for generating primary electrons; and a projection device for scanning a sample, such as a substrate, using one or more focused primary electron beams. At least the illumination device or system and the projection device or system can be collectively referred to as an electron optical system or device. Primary electrons interact with the sample and generate secondary electrons. The detector device captures the secondary electrons from the sample while scanning it, allowing the SEM to produce an image of the scanned area of ​​the sample. For high-volume inspection, some inspection devices use multiple focused beams of primary electrons, i.e., multi-beam inspection. The constituent beams of a multi-beam inspection can be called sub-beams or fine beams. Multi-beam inspection can scan different parts of the sample simultaneously. Multi-beam inspection devices can therefore inspect samples at much higher speeds than single-beam inspection devices.

[0038] The following describes the implementation of a known multi-beam detection device.

[0039] Although the description and figures are directed to an electron optical system, it should be understood that the embodiments are not intended to limit the invention to specific charged particles. Therefore, more generally, reference to electrons throughout the present invention can be considered as reference to charged particles, wherein charged particles are not necessarily electrons.

[0040] Referring now to FIG1, which is a schematic diagram illustrating an exemplary charged particle beam detection system 100, which may also be referred to as a charged particle beam evaluation system or simply an evaluation system. The charged particle beam detection system 100 of FIG1 includes a main chamber 10, a loading and locking chamber 20, an electron beam system 40, an equipment front-end module (EFEM) 30, and a controller 50. The electron beam system 40 is located within the main chamber 10.

[0041] EFEM 30 includes a first loading port 30a and a second loading port 30b. EFEM 30 may include additional loading ports. The first loading port 30a and the second loading port 30b may, for example, receive a front-opening unit cassette (FOUP) containing a substrate to be tested (e.g., a semiconductor substrate or a substrate made of other materials) or a sample (the substrate, wafer, and sample are collectively referred to below as "sample"). One or more robotic arms (not shown) in EFEM 30 transport the sample to the loading locking chamber 20.

[0042] Loading-locking chamber 20 is used to remove gas surrounding the sample. This creates a vacuum, i.e., a local gas pressure lower than the pressure in the surrounding environment. Loading-locking chamber 20 can be connected to a loading-locking vacuum pump system (not shown), which removes gas particles from loading-locking chamber 20. Operation of the loading-locking vacuum pump system enables the loading-locking chamber to reach a first pressure below atmospheric pressure. After reaching the first pressure, one or more robotic arms (not shown) can transport the sample from loading-locking chamber 20 to main chamber 10. Main chamber 10 is connected to a main chamber vacuum pump system (not shown). The main chamber vacuum pump system removes gas particles from main chamber 10, causing the pressure around the sample to reach a second pressure below the first pressure. After reaching the second pressure, the sample is transported to an electron beam system capable of detecting the sample. Electron beam system 40 may include multi-beam electron optics.

[0043] The controller 50 is electronically connected to the electron beam system 40. The controller 50 may be a processor (such as a computer) configured to control the charged particle beam detection device 100. The controller 50 may also include processing circuitry configured to perform various signal and image processing functions. Although the controller 50 is shown in FIG. 1 as being external to a structure including the main chamber 10, the loading locking chamber 20, and the EFEM 30, it should be understood that the controller 50 may be part of that structure. The controller 50 may be located in one of the components of the charged particle beam detection device or may be distributed above at least two of the components. While the present invention provides an example of housing the main chamber 10 of the electron beam system, it should be noted that the nature of the invention is not limited in its broadest sense to housing the chamber of the electron beam system. In fact, it should be understood that the foregoing principles may also be applied to other configurations of other devices and equipment operating under a second pressure.

[0044] Referring now to FIG2, it is a schematic diagram illustrating an exemplary electron beam system 40 including a multi-beam electron optical system 41 as part of the exemplary charged particle beam detection system 100 of FIG1. ​​The electron beam system 40 includes an electron source 201 and a projection device 230. The electron beam system 40 further includes a motorized stage 209 and a sample holder 207. The electron source 201 and the projection device 230 may be collectively referred to as the electron optical system 41 or as an electron optical column. The sample holder 207 is supported by the motorized stage 209 to hold a sample 208 (e.g., a substrate or mask) for detection. The multi-beam electron optical system 41 further includes a detector 240 (e.g., an electronic detection device).

[0045] The electron source 201 may include a cathode (not shown) and an extractor or anode (not shown). During operation, the electron source 201 is configured to emit electrons from the cathode as primary electrons. The primary electrons are extracted or accelerated by the extractor and / or anode to form a primary electron beam 202.

[0046] The projection device 230 is configured to convert the primary electron beam 202 into a plurality of sub-beams 211, 212, 213 and to direct each sub-beam onto the sample 208. Although three sub-beams are described for simplicity, there may be tens, hundreds, thousands, tens of thousands, or even thousands of sub-beams. Such sub-beams may be referred to as fine beams.

[0047] The controller 50 can be connected to various parts of the charged particle beam detection device 100 of FIG. 1, such as the electronic source 201, the detector 240, the projection device 230, and the motorized stage 209. The controller 50 can perform various image and signal processing functions. The controller 50 can also generate various control signals to control the operation of the charged particle beam detection device (including charged particle multi-beam device).

[0048] The projection device 230 can be configured to focus the primary sub-beams 211, 212, and 213 onto the sample 208 for detection and can form three detection spots 221, 222, and 223 on the surface of the sample 208. The projection device 230 can be configured to deflect the primary sub-beams 211, 212, and 213 across individual scanning areas in a segment of the surface of the sample 208 to scan the detection spots 221, 222, and 223. In response to the primary sub-beams 211, 212, and 213 incident on the detection spots 221, 222, and 223 on the sample 208, electrons are generated from the sample 208. These electrons include secondary electrons, which can be referred to as signal particles, and backscattered electrons. Secondary electrons typically have an energy of less than or equal to 50 eV. Actual secondary electrons may have an energy of less than 5 eV, but anything below 50 eV is considered a secondary electron. Backscattered electrons typically have electron energies between 0 eV and the landing energies of primary sub-beams 211, 212, and 213. Since electrons with detected energies less than 50 eV are generally considered secondary electrons, a portion of actual backscattered electrons will be considered secondary electrons.

[0049] The detector 240 is configured to detect signal particles such as secondary electrons and / or backscattered electrons and generate corresponding signals that are sent to the signal processing system 280, for example, to construct an image of the corresponding scanned area of ​​the sample 208. The detector 240 may be incorporated into the projection device 230.

[0050] Signal processing system 280 may include circuitry (not shown) configured to process signals from detector 240 to form an image. Signal processing system 280 may also be referred to as an image processing system. The signal processing system may be incorporated into components of electron beam system 40, such as detector 240 (as shown in FIG. 2). However, signal processing system 280 may be incorporated into any number of components of detection device 100 or electron beam system 40, such as as part of projection device 230 or controller 50. Signal processing system 280 may include an image acquisition unit (not shown) and a storage device (not shown). For example, signal processing system may include a processor, computer, server, mainframe computer, terminal, personal computer, any kind of mobile computing device and the like, or combinations thereof. Image acquisition unit may include at least a portion of the processing functionality of controller. Therefore, image acquisition unit may include at least one or more processors. The image acquisition device can be communicatively coupled to a detector 240 that allows signal communication, such as an electrical conductor, fiber optic cable, portable storage medium, IR, Bluetooth, Internet, wireless network, radio, or a combination thereof. The image acquisition device can receive signals from the detector 240, process the data contained in the signals, and construct an image based on that data. The image acquisition device can thus acquire an image of sample 208. The image acquisition device can also perform various post-processing functions, such as generating outlines, overlaying indicators onto the acquired image, and similar functions. The image acquisition device can be configured to perform adjustments to the brightness and contrast of the acquired image. The storage medium can be a storage medium such as a hard disk, flash drive, cloud storage, random access memory (RAM), other types of computer-readable memory, and similar media. The storage medium can be coupled to the image acquisition device and can be used to store scanned raw image data as the original image and post-processed images.

[0051] The signal processing system 280 may include a measurement circuitry (e.g., an analog-to-digital converter) to obtain the distribution of detected secondary electrons. Electron distribution data collected during the detection time window can be combined with corresponding scan path data from each of the primary sub-beams 211, 212, and 213 incident on the sample surface to reconstruct an image of the structure of the sample under test. The reconstructed image can be used to reveal various features of the internal or external structure of the sample 208. The reconstructed image can thereby be used to reveal any defects that may exist in the sample. For convenience, the above-described functions of the signal processing system 280 can be performed in the controller 50 or shared between the signal processing system 280 and the controller 50.

[0052] Controller 50 can control the motorized stage 209 to move the sample 208 during the detection of the sample 208. Controller 50 enables the motorized stage 209 to move the sample 208 more continuously (e.g., at a constant speed) in one direction, at least during the sample detection, which can be referred to as a type of scan. Controller 50 can control the movement of the motorized stage 209 such that the controller varies the speed of movement of the sample 208 depending on various parameters. For example, controller 50 can control the stage speed (including its direction) depending on the detection steps of the scanning procedure and / or the characteristics of the scanning procedure, as disclosed in EPA 21171877.0 filed May 3, 2021, which is hereby incorporated with respect to at least a combination of stepping and scanning strategies of the stage.

[0053] Known multi-beam systems (such as the electron beam tool 40 and charged particle beam detection device 100 described above) are disclosed in US2020118784, US20200203116, US2019 / 0259570 and US2019 / 0259564, which are hereby incorporated by reference.

[0054] The electron beam system 40 may include a projection assembly to adjust the accumulated charge on the sample by illuminating the sample 208.

[0055] Figure 3 is a schematic diagram of an exemplary electron optical column 41 used for evaluating the system. For ease of illustration, the lens array is schematically depicted herein by an array of elliptical shapes. Each elliptical shape represents one of the lenses in the lens array. By convention, elliptical shapes are used to represent lenses, similar to the biconvex form often used in optical lenses. However, in the context of charged particle configurations such as those discussed herein, it should be understood that the lens array will generally operate electrostatically and therefore may not require any physical element with a biconvex shape. Instead, as described below, the lens array may comprise multiple plates having apertures. Each plate having an aperture may be referred to as an electrode. Electrodes may be provided in series along the sub-beam paths of the sub-beams of the multi-beam array.

[0056] An electron source 201 guides electrons toward an array of condenser lenses 231 (also referred to as a condenser lens array). The electron source 201 is ideally a high-brightness thermal field emitter with a good trade-off between brightness and total emission current. There may be tens, hundreds, or thousands of condenser lenses 231. The condenser lenses 231 may comprise multi-electrode lenses and have a construction based on EP1602121A1, the document of which is hereby incorporated by reference, in particular, the disclosure of a lens array for splitting an electron beam into a plurality of sub-beams, wherein the array provides a lens for each sub-beam. The array of condenser lenses 231 may be in the form of at least two plates (preferably three plates) acting as electrodes, wherein the apertures in each plate are aligned with each other and correspond to the positions of the sub-beams. During operation, at least two of these plates are maintained at different potentials to achieve the desired lensing effect. Between the plates of the condenser lens array are electrically insulating plates made of insulating materials such as ceramic or glass, having one or more apertures for the sub-beams. An alternative configuration of one or more of the plates may be characterized by apertures, each aperture having its own electrode, each aperture having an electrode array around its periphery, or a group of apertures having a common electrode.

[0057] In one configuration, the array of condenser lenses 231 is formed by three plate arrays in which charged particles have the same energy when entering and leaving each lens; this configuration may be referred to as an ensemble lens. Therefore, dispersion occurs only within the ensemble lens itself (between the entry and exit electrodes of the lens), thereby limiting off-axis chromatic aberration. When the thickness of the condenser lens is low, for example, a few millimeters, such aberrations have a small or negligible effect.

[0058] Each focusing lens in the array guides electrons into individual sub-beams 211, 212, 213, which are focused at individual intermediate focal points 233. Collimators or collimator arrays can be positioned to operate on the individual intermediate focal points 233. The collimator can take the form of a deflector 235 disposed at the intermediate focal point 233. The deflector 235 is configured to bend the individual sub-beams 211, 212, 213 by a certain amount to effectively ensure that the main ray (which may also be referred to as the beam axis) is substantially normal to the sample 208 (i.e., substantially 90° to the normal surface of the sample).

[0059] Below the deflector 235 (i.e., in the downstream direction or further away from the source 201), there is a control lens array 250, which includes control lenses 251 for each sub-beam 211, 212, 213. The control lens array 250 may include two or more (preferably at least three) plate electrode arrays connected to the respective potential sources, preferably wherein an insulating plate contacts the electrodes, for example, between the electrodes. Each of the plate electrode arrays may be referred to as a control electrode. The function of the control lens array 250 is to optimize the beam angle relative to the beam reduction rate and / or control the beam energy delivered to the objective lens 234, each of which guides the respective sub-beams 211, 212, 213 onto the sample 208.

[0060] Where appropriate, a scanning deflector array 260 is provided between the array of control lens array 250 and objective lens 234 (objective lens array). The scanning deflector array 260 includes a scanning deflector 261 for each sub-beam 211, 212, 213. Each scanning deflector is configured to deflect the respective sub-beam 211, 212, 213 in one or both directions to scan the sub-beam throughout the sample 208 in one or both directions.

[0061] The detector module 240 of the detector is disposed within or between the objective lens 234 and the sample 208 to detect signal electrons / particles emitted from the sample 208. An exemplary configuration of this detector module 240 is described below. It should be noted that the detector may additionally or alternatively have a detector element in the countercurrent direction along the primary beam path of the objective lens array or even the control lens array.

[0062] Figure 4 is a schematic diagram of an exemplary electron beam system having an alternative electron optical column 41'. The electron optical column 41' includes an objective lens array 241. The objective lens array 241 includes a plurality of objective lenses. The objective lens array 241 may be an interchangeable module. For the sake of simplicity, the features of the electron beam system described above are not repeated here.

[0063] As shown in Figure 4, the electron optical column 41' includes a source 201. Source 201 provides a beam of charged particles (e.g., electrons). A multi-beam source focused onto sample 208 is provided by the beam from source 201. Sub-beams can be derived from the beam, for example, using a beam limiter defining a beam-limiting aperture array. The beam can be split into sub-beams upon encountering control lens array 250. The sub-beams are substantially parallel as they enter control lens array 250. The sub-beams of the multi-beam configuration can be configured in a pattern that can be referred to as a multi-beam configuration. The pattern can form a grid. The grid can be hexagonal, linear, rhomboid, or square. In the illustrated example, a collimator is provided in the counter-current direction of the objective lens array assembly.

[0064] The collimator may include a macrocollimator 270. The macrocollimator 270 acts on the beam before it has split into multiple beams from the source 201. The macrocollimator 270 bends individual portions of the beam by a certain amount to effectively ensure that the beam axis of each of the sub-beams derived from the beam is substantially perpendicular to the sample 208 (i.e., substantially 90° to the nominal surface of the sample 208). The macrocollimator 270 includes a magnetic lens and / or an electrostatic lens. In another configuration (not shown), the macrocollimator may be partially or entirely replaced by an array of collimator elements disposed in the downstream direction of the upper beam limiter.

[0065] In the electron optics column 41' of FIG4, a giant scanning deflector 265 is provided to scan the sub-beams over the sample 208. The giant scanning deflector 265 deflects individual portions of the beam to scan the sub-beams over the sample 208. In an embodiment, the giant scanning deflector 265 comprises a macroscopic multipole deflector, for example, having eight or more poles. The deflection is to cause the sub-beams originating from the beam to scan across the sample 208 in one direction (e.g., parallel to a single axis, such as the X-axis) or in two directions (e.g., relative to two non-parallel axes, such as the X-axis and Y-axis). The giant scanning deflector 265 acts macroscopically on all beams, rather than comprising an array of deflector elements each configured to act on different individual portions of the beam. In the illustrated embodiment, the giant scanning deflector 265 is disposed between the giant collimator 270 and the control lens array 250. In another configuration (not shown), the giant scan deflector 265 may be partially or completely replaced by a scan deflector array, for example, as a scan deflector for each sub-beam. In other embodiments, both the giant scan deflector 265 and the scan deflector array are provided, and they can operate synchronously.

[0066] In some embodiments, the electro-optical system 41 further includes an upper beam limiter 252. The upper beam limiter 252 defines a beam limiting aperture array. The upper beam limiter 252 may be referred to as an upper beam limiting aperture array or a counter-current beam limiting aperture array. The upper beam limiter 252 may include a plate (which may be plate-shaped) having a plurality of apertures. The upper beam limiter 252 forms sub-beams from the charged particle beam emitted from the free source 201. The upper beam limiter 252 may block (e.g., absorb) portions of the beam other than those contributing to the formation of sub-beams to avoid interfering with the sub-beams in the downstream direction. The upper beam limiter 252 may be referred to as a sub-beam defining aperture array.

[0067] In some embodiments, as illustrated in FIG4, the objective array assembly (which is a unit including objective array 241) further includes a beam shaping limiter 262. The beam shaping limiter 262 defines a beam limiting aperture array. The beam shaping limiter 262 may be referred to as a lower beam limiter, a lower beam limiting aperture array, or a final beam limiting aperture array. The beam shaping limiter 262 may include a plate (which may be plate-shaped) having a plurality of apertures. The beam shaping limiter 262 can control the flow direction of at least one electrode (in case all electrodes) of the lens array 250. In some embodiments, the beam shaping limiter 262 is in the flow direction of at least one electrode (in case all electrodes) of the objective array 241. In one configuration, the beam shaping limiter 262 is structurally integrated with the electrodes of the objective array 241. Desiredly, the beam shaping limiter 262 is positioned in an area with a low electrostatic field strength. The alignment of the beam limiting aperture with the objective lens array allows a portion of the sub-beams from the corresponding objective lens to pass through the beam limiting aperture and irradiate the sample 208, such that only a selected portion of the sub-beams incident on the beam shaping limiter 262 passes through the beam limiting aperture.

[0068] Any of the objective array assemblies described herein may further include a detector 240. The detector detects electrons emitted from the sample 208. The detected electrons may include any of the electrons detected by the SEM, including signal particles such as secondary and / or backscattered electrons emitted from the sample 208. An exemplary configuration of the detector 240 is described in more detail below with reference to Figures 6 and 7.

[0069] FIG5 schematically depicts an electron beam system 40 including an electron optical column 41'' according to one embodiment. Features identical to those described above are given the same reference numerals. For simplicity, such features are not described in detail with reference to FIG5. For example, the source 201, condenser lens 231, giant collimator 270, objective lens array 241, and sample 208 may be as described above.

[0070] As described above, in one embodiment, the detector 240 is positioned between the objective lens array 241 and the sample 208. The detector 240 may face the sample 208. Alternatively, as shown in FIG5, in one embodiment, the objective lens array 241 comprising a plurality of objectives is positioned between the detector 240 and the sample 208.

[0071] In one embodiment, a deflector array 95 is positioned between the detector 240 and the objective lens array 241. In one embodiment, the deflector array 95 includes a Wien filter array, such that the deflector array can be referred to as a beam splitter. The deflector array 95 is configured to provide a magnetic field to separate charged particles projected onto the sample 208 from secondary electrons from the sample 208 toward the detector 240.

[0072] In one embodiment, detector 240 is configured to detect signal particles with reference to the energy of charged particles (i.e., depending on the bandgap). This detector 240 may be referred to as an indirect current detector. Secondary electrons emitted from sample 208 gain energy from the field between electrodes. The secondary electrodes have sufficient energy once they reach detector 240. In different configurations, detector 240 may be, for example, an array of scintillators in fluorescent strips between beams, positioned relative to the Wayne filter along the countercurrent direction of the primary beam path. The primary beam passing through the Wayne filter array (orthogonal to the magnetic and electrostatic strips of the primary beam path) has a path that is substantially parallel to both the countercurrent and forward current directions of the Wayne filter array; and the signal electrons from the sample are directed toward the scintillator array to the Wayne filter array. The generated photons are directed to a remote optical detector (e.g., an optical fiber array) via a photon delivery unit, which generates a detection signal when detecting the photons.

[0073] The objective array 241 of any embodiment may include at least two electrodes in which an aperture array is defined. In other words, the objective array includes at least two electrodes having a plurality of holes or apertures. FIG6 shows electrodes 242, 243 as portions of an exemplary objective array 241 having respective aperture arrays 245, 246. The position of each aperture in the electrodes corresponds to the position of a corresponding aperture in the other electrode. The corresponding aperture operates in use on the same beam, sub-beam, or beam group in multiple beams. In other words, the corresponding aperture in at least two electrodes is aligned with and arranged along a sub-beam path (i.e., one of the sub-beam paths 220). Thus, each electrode has an aperture through which the respective sub-beams 211, 212, 213 propagate.

[0074] The objective lens array 241 may contain two electrodes (as shown in Figure 6) or three electrodes, or may have more electrodes (not shown). An objective lens array 241 with only two electrodes may have lower aberrations than an objective lens array 241 with more electrodes. A three-electrode objective lens may have a larger potential difference between the electrodes and thus achieve a stronger lens. Additional electrodes (i.e., more than two electrodes) provide additional degrees of freedom for controlling the electron trajectory, such as focusing secondary electrons and the incident beam. The advantage of a two-electrode lens over a single lens is that the energy of the incident beam may not be the same as the output beam. Advantageously, the potential difference on this two-electrode lens array allows it to act as an accelerating or decelerating lens array.

[0075] The adjacent electrodes of the objective array 241 are spaced apart from each other along the sub-beam path. The distance between adjacent electrodes (where the insulating structure may be positioned as described below) is greater than that between the objectives of the objective array.

[0076] Preferably, each of the electrodes provided in the objective array 241 is a plate. The electrode may also be described as a flat sheet. Preferably, each of the electrodes is planar. In other words, each of the electrodes will preferably be provided as a thin plate in planar form. Of course, the electrode does not need to be flat. For example, the electrode may be bent due to forces induced by a high electrostatic field. Planar electrodes are preferred because this makes it easier to manufacture the electrode using known manufacturing methods. Planar electrodes are also preferred because they allow for more accurate alignment of the apertures between different electrodes.

[0077] The objective array 241 can be configured to reduce the charged particle beam by a factor greater than 10, and as needed in the range of 50 to 100 or greater.

[0078] A detector 240 is provided to detect signal particles emitted from the sample 208, i.e., secondary and / or backscattered charged particles. The detector 240 is positioned between the objective lens 234 and the sample 208. The detector generates a detection signal in the direction of the signal particles. The detector 240 may also be referred to as a detector array or a sensor array, and the terms "detector" and "sensor" are used interchangeably throughout this application.

[0079] An electro-optical device may be provided for the electro-optical system 41. The electro-optical device is configured to project an electron beam toward the sample 208. The electro-optical device may include an objective lens array 241. The electro-optical device may include a detector 240. The objective lens array (i.e., objective lens array 241) may correspond to either the detector array (i.e., detector 240) and / or the beam (i.e., sub-beam).

[0080] An exemplary detector 240 is described below. However, any reference to detector 240 may be a single detector (i.e., at least one detector) or multiple detectors as needed. Detector 240 may include detector elements 405 (e.g., sensing elements such as capture electrodes). Detector 240 may include any suitable type of detector. For example, capture electrodes may be used (e.g.) to directly detect electronic charges, scintillators, or PIN elements. Detector 240 may be a direct current detector or an indirect current detector. Detector 240 may be the detector described below with respect to FIG7.

[0081] The detector 240 may be positioned between the objective lens array 241 and the sample 208. The detector 240 is configured to be close to the sample 208. The detector 240 may be very close to the sample 208. Alternatively, there may be a large gap between the detector 240 and the sample 208. The detector 240 may be positioned in the apparatus so as to face the sample 208. Alternatively, the detector 240 may be positioned elsewhere in the electro-optical system 41 such that a portion of the electro-optical apparatus that is not a detector faces the sample 208.

[0082] Figure 7 is a bottom view of a detector 240, which includes a substrate 404 on which a plurality of detector elements 405, each surrounding a beam aperture 406, are provided. The beam aperture 406 can be formed by etching through the substrate 404. In the configuration shown in Figure 7, the beam aperture 406 is in the form of a hexagonal close-packed array. The beam aperture 406 can also be configured in different ways, such as a rectangular or rhomboid array. The hexagonal beam configuration in Figure 7 can be more densely packed than a square beam configuration. The detector elements 405 can be configured in a rectangular array or a hexagonal array. The beam aperture can correspond to a multi-beam configuration of sub-beams facing the substrate 404.

[0083] The capture electrode 405 forms the bottommost (i.e., closest to the sample) surface of the detector module 240. A logic layer is disposed between the capture electrode 405 and the body of the silicon substrate 404. The logic layer may include amplifiers (e.g., impedance amplifiers), analog-to-digital converters, and readout logic. In one embodiment, each capture electrode 405 has an amplifier and an analog-to-digital converter. Circuitry characterized by these elements may be contained in unit regions called aperture-associated cells. The detector model 240 may have several aperture-associated cells. A wiring layer is provided within or on the substrate to connect to the logic layer and externally to each cell, for example via power, control, and data lines. The integrated detector module 240 described above is particularly advantageous when used with systems having tunable landing energy, as secondary electron capture can be optimized for the landing energy range. Detector modules in array form can also be integrated into other electrode arrays, not just the lowest electrode array. This detector module may be characterized by detectors, which are, for example, scintillators or semiconductor detectors, such as PIN detectors, located above the surface of the objective lens in the most downstream direction. Such detector modules may also be characterized by a circuit architecture similar to that of detector modules containing current detectors. Further details and alternative configurations of detector modules integrated into the objective lens can be found in EP applications 20184160.8 and 20217152.6, which are incorporated herein by reference, at least for details concerning the detector modules.

[0084] The detector may have multiple parts, and more specifically, multiple detection parts. A detector comprising multiple parts may be associated with one of the sub-beams 211, 212, and 213. Thus, multiple parts of a detector 240 may be configured to detect signal particles emitted from the sample 208 with respect to one of the primary beams (which may also be referred to as sub-beams 211, 212, and 213). In other words, a detector comprising multiple parts may be associated with one of the apertures of at least one of the electrodes of the objective lens assembly. More specifically, a detector 405 comprising multiple parts may be configured around a single aperture 406, providing an example of such a detector. As mentioned, detection signals from the detector module are used to generate an image. With multiple detection parts, the detection signals include components from different detection signals, which can be used as a dataset or processed in the detection image.

[0085] In one embodiment, the objective array 241 is an interchangeable module, which may be used alone or in combination with other elements such as a control lens array and / or a detector array. This interchangeable module may be field-replaceable, meaning it can be replaced by a new module by a field engineer. In one embodiment, multiple interchangeable modules are included within the system and can be switched between operable and inoperable positions without activating the electron beam system.

[0086] In some embodiments, one or more aberration correctors are provided to reduce one or more aberrations in the sub-beams. An aberration corrector located in or directly adjacent to the intermediate focal point (or intermediate image plane) may include a deflector to correct for sources 201 appearing at different locations for different beams. The corrector can be used to correct macroscopic aberrations caused by the source that prevent proper alignment between each sub-beam and its corresponding objective. The aberration corrector can correct aberrations that prevent proper column alignment. The aberration corrector may be a CMOS-based individually programmable deflector disclosed in EP2702595A1 or a multi-pole deflector array disclosed in EP2715768A2, the descriptions of fine beam manipulators in these two documents are hereby incorporated by reference. The aberration corrector can reduce one or more of the following: field curvature; focus error; and astigmatism.

[0087] This invention can be applied to various system architectures. For example, the electron beam system can be a single-beam system, or it can include a plurality of single-beam pillars or a plurality of multi-beam pillars. The pillars can be included in the electron optical system 41 described in any of the above embodiments or examples. As a plurality of pillars (or a multi-pillar system), the device can be configured as an array, the number of which can be two to one hundred pillars or more. The electron beam system can take the form of the embodiments described and depicted with respect to FIG. 3 or the embodiments described and depicted with respect to FIG. 4, but preferably has an electrostatic scanning deflector array and an electrostatic collimator array.

[0088] FIG8 is a schematic diagram of an exemplary single-beam electron beam system 41''' according to one embodiment. As shown in FIG8, in one embodiment, the electron beam system includes a sample holder 207 supported by a motorized stage 209 to hold a sample 208 to be tested. The electron beam system includes an electron source 201. The electron beam system further includes a gun aperture 122, a beam limiting aperture 125, a condenser lens 126, a pillar aperture 135, an objective lens assembly 132, and an electron detector 144. In one embodiment, the objective lens assembly 132 may be a modified oscillating objective delayed immersion lens (SORIL) that includes a pole piece 132a, a control electrode 132b, a deflector 132c, and an excitation coil 132d. The control electrode 132b has an aperture formed therein for the passage of the electron beam. The control electrode 132b forms a facing surface 72, which is described in more detail below.

[0089] In the imaging process, the electron beam emitted from source 201 can pass through gun aperture 122, beam limiting aperture 125, and condenser lens 126, and is focused into a probe spot by a modified SORIL lens and then illuminates the surface of sample 208. The probe spot can be scanned across the surface of sample 208 by deflector 132c or other deflectors in the SORIL lens. Secondary electrons emitted from the sample surface can be collected by electron detector 144 to form an image of the region of interest on sample 208.

[0090] The condenser and illumination optics of the electro-optical system 41 may include or be supplemented by an electromagnetic quadrupole lens. For example, as shown in FIG8, the electro-optical system 41 may include a first quadrupole lens 148 and a second quadrupole lens 158. In one embodiment, the quadrupole lens is used to control the electron beam. For example, the first quadrupole lens 148 may be controlled to adjust the beam current and the second quadrupole lens 158 may be controlled to adjust the beam spot size and beam shape.

[0091] Images output from a charged particle evaluation device (e.g., electron beam system 40) need to be automatically processed to detect defects in the evaluated sample. Figure 9 depicts a data processing device 500 for detecting defects in images generated by the charged particle evaluation device. The data processing device 500 may be part of a controller 50, part of another computer in a wafer fab, or integrated elsewhere within the charged particle evaluation device. It should be noted that the configuration of the components of the data processing device 500 shown and described with reference to Figure 9 is illustrative and provided to aid in explaining the functionality of a data processor operating on images generated by the charged particle evaluation device. Any feasible configuration that can be conceived by a skilled data processor technician capable of achieving the functionality of the data processing device 500 as described herein may be used.

[0092] Charged particle evaluation devices can have high throughput and high resolution, meaning that data can be output at high rates. For example, a charged particle evaluation device can have thousands, or even tens of thousands, or more beams, each of which has one or more detector sections that output data points at a rate of kilohertz or megahertz. Data needs to be processed at a rate equal to or at least similar to the output rate of the self-contained charged particle evaluation system 40. When applied to multi-beam or multi-column charged particle evaluation devices, known image processing methods used for defect detection require excessively high processing power to keep up with the rate of image generation.

[0093] As depicted in Figure 10, the present invention addresses the problem of high-speed data processing by providing a two-step procedure or at least one of at least two steps. (Although characterized by more than two steps, the procedure may be referred to as a "two-step" procedure). A charged particle evaluation device generates a sample image data stream DS1 of data points S1, each data point representing the characteristics of a position on the sample when it influences the generation of signal particles (i.e., backscattering and / or secondary electrons). A first defect detection test S2 is applied to the sample image data stream DS1. The first detection test S2 is applied to the sample image to select data points and / or groups of data points representing sample areas that may contain defects. The selected data points are output as a reduced data stream DS2. The reduced data stream DS2 has a lower data rate than the sample data stream DS1 due to the selectivity of the first defect detection test. A second defect detection test S3 is applied to the selected data in the reduced data stream DS2. That is, the second detection test S3 determines with higher accuracy whether the selected defect is actually a defect. That is, a second defect detection test is used to determine whether any of the areas identified as potentially defective by the first defect detection test actually has a defect. The data points determined to be defective by the second defect detection test are output in the output data stream DS3.

[0094] Where necessary, the two-step procedure is performed simultaneously or in parallel with the generation of the sample image data stream, so that the data of interest is initially identified, i.e., identified as potentially containing one or more defects. Therefore, data not of interest does not require further storage or processing. However, in other cases, it may be necessary to perform the two-step procedure on the stored data. For example, the two-step procedure may be performed on data from at least two sources, one of which is the stored data. The other data source used for this two-step procedure may be a data stream.

[0095] Figure 11 depicts a hardware configuration according to one embodiment for implementing this method. A data processing device 540 for detecting defects in sample image data is connected to a charged particle evaluation system 40. The data processing device 540 includes a first processing module 550 and a second processing module 560. The first processing module 550 is configured to receive a sample image data stream DS1 from the charged particle evaluation system 40, the sample image data stream containing a series of ordered data points representing an image of the sample. The first processing module 550 applies a first defect detection test to select a subset of the sample image data stream as first selected data, for example, in the form of a reduced data stream DS2. The first defect detection test is a localized test performed in parallel with the received sample image data stream. In some embodiments, the received sample image data stream is compared with buffered data obtained from other parts of the sample. This buffered data can be considered as stored data transmitted via streaming for processing in the first defect detection test. "Localization" means performing the test on the target pixel and a small number of neighboring pixels (e.g., less than 50), rather than considering the entire image. Hardware configurations apply data processing methods for detecting defects in data, such as image data. That is, the data represents a sample image, such as a portion of the sample. The data is generated by an evaluation system for evaluating the sample. Therefore, this method is a data processing method for detecting defects in sample image data generated by the evaluation system. In this method: Data generated by the evaluation system is received. The data is received in a data stream. This can be a sample image data stream received by the evaluation system. The sample image data stream can contain a series of ordered data points representing the image of the sample. A first test is applied to select a subset of the data. The first test is applied to the data in the data stream. The first test includes a simple algorithm. In the first test, the simple algorithm is applied to the data. Such a simple algorithm is a localization test. The first test (or a first defect detection test, such as a localization test) can be performed in parallel with the received data (e.g., as a sample image data stream). The application of the first detection test may include selecting a subset of the sample image data stream as the first selected data.

[0096] The second processing module 560 is configured to receive the first selected data. The second processing module 560 applies a second defect detection test to select a subset of the first selected data as the second selected data. The second processing module 560 outputs the second selected data, for example, as an output data stream DS3, to the output buffer 570. The second processing module 560 can apply any suitable defect detection test to detect defects. Suitable defect detection tests may involve computationally intensive tests such as high-resolution image comparison, pattern recognition, or machine learning techniques. The second processing module can also perform other image processing steps, such as filtering or noise reduction. Therefore, the second test can be applied by means of hardware configuration application. The second test can be applied to a subset to verify whether at least a portion of the subset of data represents at least one defect. The second test is a complex or expensive algorithm used to apply to the data. The algorithm may cause relatively computationally intensive operations when applied to the data (as needed, a subset of the data). For the second application, the first selected data can be received. The second defect detection test can be applied to the first selected data. The application of the second defect detection test may include selecting a subset of the first selected data as the second selected data. The second selected data may represent the defect.

[0097] The two-step process can handle high data rates without excessive hardware requirements because the first processing module actually selects a portion of the sample image data for further analysis by the second processing module (or a later processing module). Therefore, a small proportion of the sample image data is streamed and subjected to further analysis by the second processing module. This proportion can be as low as less than 10%, less than 5% if necessary, and less than 2% if even more necessary.

[0098] Where necessary, the first processing module 550 uses a first number of operations per pixel to apply a first defect detection test. The second processing module uses a second number of operations per pixel to apply a second defect detection test. The first number of operations is less than the second number of operations. If the first defect detection test is simpler than the second defect detection test, the processing requirements are reduced.

[0099] Where appropriate, the first processing module 550 uses fewer than 200 operations per pixel, and more appropriately fewer than 100 operations, to apply the first defect detection test. When the first defect detection test is performed in a small number of operations, it is easy to execute at high speed, for example, using dedicated hardware. Specifically, the first processing module needs to use only simple operations, such as: AND, OR, NOT, NAND, XOR, addition, subtraction, and bit shifting.

[0100] The first defect detection test may have lower selectivity than the second defect detection test. That is, the first detection test produces a higher false positive rate (areas marked as potentially defective but actually not defective). If the first defect detection test rejects a high proportion of true negatives (defect-free areas), the overall data processing requirements can be reduced. This reduction can be achieved even if the first defect detection test produces a considerable proportion of false positives.

[0101] Where appropriate, the first processing module 540 includes a filter module 551 that performs convolution of data points of the sample image data stream with a kernel of a predetermined size. The application of this filter can be implemented with a few operations and is effectively used to reduce noise in the image data.

[0102] Where appropriate, the first processing module includes a comparator 552 that compares data points of the sample image data stream with first reference image data 553. The first reference image data can be obtained in various ways as described below. The comparison of data points in the sample image data stream with corresponding pixels in the reference image can be performed in a few operations.

[0103] Where appropriate, the first processing module includes an input buffer 554 configured to buffer and temporarily store sample image data streams. The input buffer 554 may have a capacity smaller than that required to store the entire pattern repeating region of the sample. The pattern repeating region of the sample is a region containing a pattern that repeats elsewhere on the sample. The pattern repeating regions are nominally identical. In many cases, the pattern repeating region will correspond to a die or target portion of the sample (a semiconductor substrate having means formed thereon), but in some cases the pattern repeating region may be smaller than the die or may correspond to a plurality of combined dies. The input buffer 554 is depicted as being located after the filter module 551, but may be before it or in parallel with it.

[0104] Where appropriate, the first processing module 550 includes a reference buffer 555 configured to buffer and temporarily store the first reference image data 553. The reference buffer may have a capacity smaller than the amount of first reference image data required to store the entire pattern repeating area of ​​the sample. The amount of buffer storage required can be minimized by scanning and processing a smaller area of ​​the sample at a time.

[0105] Where appropriate, the first reference image data has a lower resolution than the sample image. Preferably, the second processing module compares the first selected data with the second reference image data. Preferably, the first reference image data is a lower resolution version of the second reference image data. Using a lower resolution reference image reduces the amount of data that needs to be supplied to the first processing module for comparison.

[0106] Where appropriate, the first processing module includes a field-programmable gate array or application-specific integrated circuit. Dedicated hardware (such as a simple processor, like an FPGA or ASIC) can perform direct computation efficiently at high speed. The size of the simple processor can be small with low power consumption, making it easier to access the data source location in the charged particle evaluation device.

[0107] Where necessary, a first defect detection test generates a first defect score. The first defect score indicates the probability that a pixel in the sample image represents a defect. Pixels with defect scores indicating that they have a significant probability of being defects are called defect candidates. The first processing module includes an output buffer configured to accumulate data of image areas containing candidate defects. The accumulated data of areas containing candidate defects may be first selected data. The accumulated data of areas containing candidate defects may have the highest value of the first defect score.

[0108] Where appropriate, the charged particle beam system and data processing device are integrated into the charged particle evaluation system. Where appropriate, the charged particle beam system is located in a vacuum chamber. The first processing module is at least partially located in the vacuum chamber. If not entirely outside the vacuum chamber, the second processing module is at least partially located outside the vacuum chamber. A portion of the first processing module may be located outside the vacuum chamber or a portion of the second processing module may be located inside the vacuum chamber; the first and second processing modules may be partially inside and outside the vacuum chamber, and therefore may be distributed processing modules.

[0109] Desiredly, the charged particle beam system is a multi-column beam system. Multi-column systems generate data at high rates. Therefore, multi-column systems greatly benefit from the present invention. Desiredly, there are a plurality of first processing modules, each of which is associated with one of the columns of the multi-column beam system. Compared to the first processing modules, fewer second processing modules may exist. The overall requirements of the processing hardware can be reduced by having multiple first processing modules feed selected data into a single second processing module. This is achievable because the amount of data to be processed by the second processing module is substantially less than the amount of data processed by the first processing module.

[0110] Various methods can be used to detect defects in images generated by the charged particle evaluation device. A particularly suitable test for a first defect detection test performed by the first processing module is to compare an image of a portion of a sample (referred to herein as a sample image) with a reference image. In practice, data points representing a data stream of the sample image are compared with data points of a reference image retrieved from memory or delivered in a parallel data stream. For simplicity, this procedure may be referred to below as comparing the image with data points as pixels. Any pixel that is different from the corresponding pixel of the reference image may be considered a defect. This pixel that is different from the reference image, along with its neighboring pixels that are also different from the reference image, is considered a single defect. (In one embodiment, this may mean that a pixel that is different from the reference image and has neighboring pixels that are the same as (i.e., not different from) the reference image may be considered not to represent a defect.) The reference image can be obtained in various ways, as discussed below.

[0111] The false positive rate can be controlled, that is, marking a sample as defective when no significant defect actually exists. The false positive rate can be controlled by setting a threshold value for determining the presence of a defect before marking it. The threshold value can be the difference between a pixel and a reference image, such as the difference between the equivalent pixels of the reference image. The false positive rate is further controlled by applying noise reduction to either or both of the reference image and the sample image. However, noise reduction increases the amount of processing required to detect defects.

[0112] An efficient and effective method for reducing noise in sample images is to apply a simple filter by convolution (e.g., a uniform filter (convolution with a uniform kernel)). To reduce noise in a reference image, multiple source images can be averaged. In some cases, such as when the reference image is obtained by simulation using self-designed data (often in GDSII format), noise reduction of the reference image can be omitted.

[0113] The efficiency and effectiveness of noise reduction in sample images can be optimized by appropriately selecting the size of the (uniform) filter. The optimal filter size may depend on factors such as the resolution of the sample image and the size of the features on the sample being inspected. The size of the kernel used to implement the uniform filter can be equal to a non-integer number of pixels. A kernel width in the range of 1.1 to 5 pixels (and, where necessary, 1.4 to 3.8 pixels) is suitable for a variety of applications. The form of the uniform kernel is further discussed below. Using a uniform filter to reduce noise facilitates its implementation on dedicated hardware such as FPGAs or ASICs, thereby achieving efficient and fast processing.

[0114] The averaging of the source image to obtain the reference image may vary depending on the nature of the source image. When the source image is exported from a past scan library, a large number of images (e.g., more than 20, more than 30, or about 35) can be averaged to obtain the reference image, because the averaging can be performed offline. The source image can be aligned before averaging.

[0115] Alternatively, the sample image can be compared with a reference image derived from a "live" source image obtained from different portions of the same sample. In this case, fewer (e.g., two) source images can be averaged to obtain the reference image. Two source images can be obtained from corresponding regions of different grains of the sample. Alternatively, if the pattern to be detected has repeating elements, source images can be obtained from the same grain. In some cases, the source image can be a shifted portion of the sample image. When comparing the sample image with a reference image derived from a live source image, the roles of the different images can be rotated. For example, if three images A, B, and C are output by a charged particle evaluation device: A and B can be averaged to provide a reference image for comparison with C; A and C can be averaged to provide a reference image for comparison with B; and B and C can be averaged to provide a reference image for comparison with A.

[0116] Another possibility is that the reference image is obtained from one or more "known good" patterns. Another possibility is that the reference image is obtained by simulation, for example, based on pattern data (e.g., in GDSII format). At least by means of the first processing module (e.g., in the first step of a two-step procedure), this reference image may be one or more "known good" patterns or a simplification of the simulation.

[0117] The result of comparing the sample image and the reference image can be a simple binary value representing the difference or correspondence (i.e., match) between the sample and the reference image. More preferably, the comparison result is a difference representing the magnitude of the difference between the sample image and the reference image. More preferably, the comparison result is the difference for each pixel (or each group of neighboring pixels, which may be called a "pixel region"), so that the location of defects in the source image can be determined more accurately.

[0118] To determine whether the difference between pixels or pixel regions between a source image and a reference image indicates a defect in the pattern being inspected, a threshold value can be applied to the difference corresponding to the pixel or pixel region. The threshold value can be fixed in advance, for example, for a specific charged particle beam system or for a specific pattern to be inspected. The threshold value can be a user-defined parameter or dependent on other conditions, such as depending on the application or being updated periodically based on evaluation. The threshold value can be dynamically determined, updated during processing, or both. Alternatively, a predetermined number of locations with the highest difference values ​​can be selected as candidate defects for further inspection. Neighboring pixels with differences higher than the threshold value can be considered as single defects or candidate defects. All pixels of a single defect can be assigned the same difference value. Such neighboring pixels and all pixels of a single defect can be referred to as a pixel region.

[0119] An efficient approach to identifying a predetermined number of locations with the highest difference is to process pixels sequentially and write pixel information and differences to a buffer. Pixel information may include a pixel data area surrounding the pixel or group of pixels identified as a potential defect. This pixel data area may be referred to as a clip. If the buffer is full and a newly processed pixel has a higher difference than the pixel in the buffer with the minimum difference, the pixel information associated with the pixel with the minimum difference is rewritten. In one possible implementation, a threshold value for selecting pixels is set at a predetermined level until the buffer is full. When the buffer is full, the threshold value is updated to the minimum difference of the pixels stored in the buffer, and is updated whenever a pixel in the buffer is rewritten. In this way, only one comparison needs to be performed. Alternatively, the threshold value may be kept constant and initially the selected pixels may be tested individually to see if they have a difference higher than that of the pixels in the buffer. This method may select a predetermined proportion of pixels and a predetermined amount of data. Alternatively, a threshold value may be set to select pixels that exceed the threshold value, such that the selected pixels being tested have a value corresponding to or exceeding the set threshold value. In this method, the proportion of pixels is not predetermined; therefore, the amount of data selected is unpredictable. The proportion of selected pixels may depend on the set of data being processed. Because the number of selected pixels in such methods is much lower than the total number of pixels, further processing of the selected pixels can be performed asynchronously with the initial processing (e.g., by a different processor) without reducing output.

[0120] Figure 9 depicts a processing module 500 that can be used as the first processing module 550. The processing module 500 includes: a filter module 501 that receives a sample image from the charged particle evaluation system 40 and filters the sample image; a reference image generator 503 that generates a reference image based on the source image; a comparator 502 that compares the filtered sample image with the reference image; and an output module 504 that processes and outputs the comparison result.

[0121] The filter module 501 applies a filter (e.g., a uniform filter) of a predetermined size to the sample image. Applying the uniform filter involves convolving the sample image with a uniform kernel. The size of the uniform kernel is determined by the user, for example, based on factors such as the size of features on the sample, the size of the defect to be detected, the resolution of the charged particle evaluation device, the amount of noise in the image, and the desired trade-off between sensitivity and selectivity for detecting a given sample. The size of the uniform kernel does not necessarily have to be an integer number of pixels. For example, for pixel sizes in the range of 5 nm to 14 nm and defects of approximately 20 nm, a uniform kernel with a width in the range of 1.1 to 5 pixels (ideally in the range of 1.4 to 3.8 pixels) is advantageous by providing high selectivity and high sensitivity.

[0122] A square uniform kernel 505 with a non-integer size (width) is depicted in Figure 12. This uniform kernel includes a central region 505a with n×n values ​​(all values ​​are 1), and a peripheral region 505b consisting of a top column, a bottom column, a left row, and a right row. Except for the corner values ​​of f2, all values ​​of the peripheral region are f, where f < 1. The effective size of the uniform kernel is equal to n + 2f pixels. Depending on the situation, the uniform kernel may be normalized (i.e., all values ​​are divided by a constant so that the sum of all values ​​is 1). Alternatively or additionally, the filtered sample image may be normalized or scaled.

[0123] In some cases, such as the uniform kernel described above, a two-dimensional kernel can be decomposed in orthogonal directions into two sequentially applied one-dimensional convolutions. This can be advantageous because the number of operations used to perform an n×n two-dimensional convolution scales proportionally to the square of n, while the number of operations used to perform two n one-dimensional convolutions scales linearly with n.

[0124] The kernel need not be square and can be, for example, rectangular or any other convenient shape. The filtering function implemented by the kernel need not have the same shape and size as the kernel; kernels larger than the filtering function will include zero values. The filter can be symmetrical if desired, but this is not mandatory. Simulations performed by the inventors show that implementing a uniform kernel provides good results, but allows for some deviation from mathematically uniform filters. For example, a corner filter can have a value f, thus slightly overweighting other pixels but not significantly. Non-uniform filters, such as Gaussian filters, can be conveniently implemented by convolving with a suitable kernel.

[0125] In particular, when configured to apply a uniform filter of a predetermined size, the filter module 501 can be conveniently implemented using dedicated hardware such as an FPGA or ASIC. Such dedicated hardware can be more efficient and economical than programmed general-purpose computing devices such as standard or common types of CPU architectures. The processor has less power than a CPU, but can have an architecture suitable for processing software used to process detection signal data (i.e., images), and therefore can process images in the same or less time than a CPU. Although it has less processing power than most CPUs at the same time, this detection processing architecture can process data just as fast because of the more efficient data architecture of the dedicated processing architecture.

[0126] As an example, Figure 13 is an image or even a portion of an image of a sample generated by the charged particle evaluation device, and therefore is an edit. It will be seen that the tested sample has a characteristic repeating pattern, in which the size of the unit cell is indicated by the dimensions ShiftX and ShiftY.

[0127] The reference image generator 503 can operate in one or more modes, each mode representing a different way of generating the reference image.

[0128] In library mode, the reference image generator 503 averages a large number of source images obtained from previous scans of patterns that are nominally identical to the currently evaluated pattern. These images may have been generated earlier in the same batch of samples or from samples in previous batches. Library images can be exported from test samples or production samples. Before averaging, the images are aligned with each other as needed. Averaging the source images to generate the reference images has a noise reduction effect. Averaging the source images in this way also averages out any defects that may be visible in the source images.

[0129] When the detected pattern is a repeating pattern, as shown, for example, in Figure 13, it is possible to generate a reference image by averaging multiple shifted versions of the source image. Each version of the source image is shifted by an integer multiple of ShiftX and / or ShiftY. If any one or two dimensions of the unit cell are not equal to an integer number of pixels, the shift amount can be rounded to the nearest pixel, or fractional pixel shifting can be achieved by interpolation (e.g., linear or cubic interpolation, such as bicubic interpolation) or any other known interpolation technique. Another possibility is to shift the pitch of the repeating pattern by a multiple such that the multiple is an integer number of pixels. In practice, multiple instances of unit cells are extracted from the source image and averaged. This method can be referred to as an instance of an array pattern, and more specifically, provides a reference image for an array pattern.

[0130] In inter-grain mode, three pillars of the multi-pillar charged particle evaluation device are used to generate sample images and two reference images. An image aligner is provided to align the images before supplying them to the reference image generator 503 and filter module 501 as needed. This configuration is particularly efficient when the spacing between the pillars is equal to the grain size of the sample being tested, because the pillars will then automatically scan the corresponding pattern features simultaneously. In cases where there is a difference between the pillar spacing and the grain size, a buffer can be used to correct the timing of the images input to the data processing device.

[0131] In an alternative variant of the inter-grain mode, the same beam (for the same column in a multi-column system) is used to generate a sample image and two reference images. This has the advantages of eliminating the need for inter-column and inter-beam corrections and simplifying data routing. For example, this calibration is the relative position of the beam to the ideal beam position used for the other beam or even for each beam. Using the same beam avoids calibrations between beams used for different scans being compared, such as offset assessments. Using the same beam means that although the beam may be offset relative to the ideal beam position, there is no positional offset that needs to be corrected when comparing data acquired from the self-scan. However, this may increase the need for data buffering before starting to compare data from different grains, for example, in terms of frequency (if not in terms of the amount of buffered data).

[0132] An alternative version of the array pattern of a single column 507, for example, a single-column system, is used to provide a sample image, which is compared with a reference image derived from two shifted versions of itself as the source image. A buffer can be used to provide the shifted image.

[0133] It should be noted that a uniform filter may also be applied to the source image and / or reference image, especially when the reference image is derived from a small number of source images obtained simultaneously with the source image.

[0134] Referring again to Figure 9, comparator 502 can be any logic circuit capable of comparing two values, such as an XOR gate or a subtractor. Comparator 502 is also suitable for implementation using dedicated hardware such as an FPGA or ASIC. Such dedicated hardware can be more efficient and economical than a programmable general-purpose computing device such as a CPU. Where necessary, comparator 502 is implemented on the same dedicated hardware as filter module 501.

[0135] In some cases, the reference image generator 503 may also be implemented with dedicated hardware, especially where the reference image generator operates only in a mode that generates reference images from a small number (e.g., two) of source images. In such cases, the reference image generator needs to be implemented with dedicated hardware similar to that of the comparator and / or filter module. The mathematical operations of pixel averaging of the source images and comparison with the pixels of the sample images can be combined into a single logic circuit where appropriate.

[0136] Output module 504 receives the result output by comparator 502 and prepares it for output to a user or other wafer fab system. The output can take any of several different forms. In the simplest option, the output may simply be an indication of whether the sample has or does not have a defect. However, since almost all samples will have at least one potential defect, more detailed information is required. Therefore, the output may include, for example, defect location mapping, difference images, and / or information about the severity of a possible defect, represented by the magnitude of the difference between the sample image and the reference image. Output module 504 may also filter potential defects, for example, by outputting only the defect locations where the magnitude of the difference between the sample image and the reference is greater than a threshold or where the pixel density of the difference is higher than a threshold. Another possibility is to output only a predetermined number of the most severe defect sites indicated by the magnitude of the difference. This can be achieved by storing the defect sites in buffer 510, and when the buffer is full, if a higher magnitude defect is detected, the lowest magnitude defect is overwritten.

[0137] Any suitable format for outputting defect information can be used, such as a list or image. If needed, output module 504 can output a clip, i.e., an image of the area of ​​the sample where the potential defect has been detected. This allows for further examination of the potential defect to determine whether it is real and serious enough to affect the operation of the device formed on or present in the sample. The remaining portions of the source image (i.e., those not saved as clips) can be discarded to save on data storage and transmission requirements.

[0138] In the charged particle detection system, each or every electron optical system 41 is located within the main (vacuum) chamber 10 and can be any of the electron optical systems 41, 41', 41'', or 41''' described above. Therefore, a device is necessary to transmit data from the electron optical system to the outside of the vacuum chamber. An optical transceiver located near the detector module 240 of the electron optical system 41 can be used to transmit the raw data generated by the electron optical system out of the vacuum chamber. The optical transceiver is configured to convert the electrical signal output by the detector module 240 into an optical signal for transmission along an optical fiber. The optical fiber may be capable of transmitting multiple channels simultaneously (e.g., using different wavelengths). Therefore, the detection signal from each individual electrode of the detector module is converted into an appropriate number of data streams. Multiple optical fibers can be used, as single-channel or multi-channel optical fibers. Each or every optical fiber passes through the wall of the main chamber 10 (which is in a vacuum during use) via a vacuum feedthrough. Suitable vacuum feedthroughs are described in US 2018 / 0182514 A1, which is incorporated herein by reference at least in its case concerning feedthrough devices. An optical fiber is connected to a first processing unit 550, which can therefore be located outside the vacuum for easy access and to avoid the need to increase the size of the vacuum chamber to accommodate the data processing unit.

[0139] An alternative or additional configuration places the first processing module close to the detector module. This is particularly convenient when the first processing module compromises a simplified dedicated processor such as an FPGA and at least provides signal connection to an optical transceiver and fiber optic cable. The optical transceiver and fiber optic cable transmit the first selected data to a second processing module located outside the main chamber. This configuration reduces the number of fibers that must be transmitted through the walls of the vacuum chamber because the data rate of the first selected data is much lower than the output of the detector module.

[0140] In both single-column and multi-column systems, multiple optical transceivers and multiple optical fibers can be used per column if convenient.

[0141] In a defect detection method that compares a sample image with a reference image derived from other areas of the same sample, it has been previously proposed to scan the entire pattern repeating area (possibly excluding non-critical areas) to derive the reference image. Therefore, the reference image is derived before scanning the sample area to obtain the sample image. This sample image is compared with the previously derived reference image. The inventors have determined that this is not the most efficient method. The inventors alternatively propose that scanning be performed only on a portion of each of a plurality (e.g., two) pattern repeating areas that serve as the reference area, and the corresponding portion of the pattern repeating area to be detected. The pattern repeating area to be detected may be referred to as the sample area. The sample area and the reference area are scanned before data processing to begin defect detection, and the other portions of the reference area and the sample area are subsequently scanned.

[0142] Therefore, in one embodiment, a method uses a charged particle beam system to detect defects in a sample having a plurality of pattern repeating regions, the method comprising, in sequence, a first scan, a second scan, a third scan, and a subsequent scan. In the first scan, a first portion of a first region in the pattern repeating regions is scanned to generate first scan image data. In the second scan, a first portion of a second region in the pattern repeating regions is scanned to generate second scan image data. The second region is spaced apart from the first region. In the third scan, a first portion of a third region in the pattern repeating regions is scanned to generate third scan image data. The third region is spaced apart from both the first and second regions. The subsequent scan scans a second portion of the first region to generate fourth scan image data.

[0143] The first, second, and third scan image data can be used to detect defects using various different procedures. For example, two of the scan image data can be used as source images to derive a reference image, and the third scan image data can be compared with that reference image. Alternatively, a majority vote can be applied among the three scan image data. Detection methods based on machine learning techniques (e.g., trained on labeled defective and non-defective images) are also suitable.

[0144] From the above configuration, a number of advantages can be accumulated. For example, by selecting the size of each scanned region, the amount of buffer memory required can be determined; thus, the trade-off between equipment cost and output can be achieved.

[0145] As needed, the sample is mechanically scanned relative to the electro-optical system in at least a single direction during the first, second, third, and fourth scans. In this case, it is not necessary to stitch multiple scans together to produce an image of the sample, thereby avoiding the need for overscanning in the stitching process and reducing the computational workload. A small amount of overscanning is required to ensure that the images of different scan areas can be aligned with each other before comparison. This amount of overscanning depends on the positioning accuracy and the time between scans (e.g., considering the thermal expansion of the sample), but is likely to be less than the time required for stitching. The detailed implementation of overscanning is described below.

[0146] In the case of a charged particle beam system with multiple beams, a first scan, a second scan, and a third scan need to be performed so that the same beam is used to scan the corresponding portions of each pattern repeating region. In this way, differences between beams do not contribute to errors in image comparison. Furthermore, matching the beam or column spacing with the pitch of the pattern repeating region (e.g., grain) of the sample is unnecessary. Although this configuration allows for real-time comparison of data in multiple data streams, this comparison will require calibration between data streams to compensate for the use of different beams for each data stream.

[0147] The above method is particularly applicable when used in conjunction with a two-step processing method that includes a first subroutine and a second subroutine. The first subroutine for applying the first defect detection test selects a subset of one of the first scan image data, the second scan image data, and the third scan image data as the first selected data. The second subroutine for applying the second defect detection test selects a subset of the first selected data as the second selected data.

[0148] The processing method for these two steps can be described as described above with reference to Figures 9 to 13.

[0149] Where appropriate, when the two-step procedure is used to detect defects (i.e., the two-step procedure is a two-step defect detection test), the time spent by the first subroutine is less than or equal to the time interval between the start of the first scan and the start of the fourth scan.

[0150] The first scan image data needs to be buffered and temporarily stored, at least until the generation of the second scan image data begins. Processing of the first and second scan image data can begin immediately or even when the second scan image data is being generated. For example, filtering or denoising procedures can be performed. This can reduce net processing time. However, this processing is incomplete because a third set of image data is required for complete processing.

[0151] The first and second scan image data also need to be buffered and temporarily stored, at least until the generation of the third scan image data begins. Processing of all three scan image data sets can begin, for example, when the third image data scan is generated from the data stream of the third image data scan. This can reduce processing time.

[0152] It is also necessary to buffer and temporarily store the first scan image data, the second scan image data and the third scan image data.

[0153] A more detailed example of this partial scanning method is described below with reference to Figures 14A, 14B, 14C, 15 and 16.

[0154] Figures 14A, 14B, and 14C depict an example relationship between the geometry of the multi-beam configuration (as a grid of sub-beam fields of view 720) and the geometry of the sub-beam processing region 740 of a multi-beam electro-optics system. The relationship between the geometry of the multi-beam configuration and the geometry of the sub-beam processing region 740 enables continuous coverage of samples. The symmetry of the sub-beam grid need not be the same as the symmetry of the sub-beam processing region 740. In the example shown, sub-beams are provided on a hexagonal grid (i.e., a grid with hexagonal symmetry), and each sub-beam has a hexagonal field of view 720. The sub-beam processing regions 740 (each of which is scanned by a single sub-beam) are rectangular. Having rectangular sub-beam processing regions is desirable because the geometry used for scanning can be simpler than the geometry that the hexagonal grid of the sub-beam processing region might require. Continuous coverage of the sample surface parallel to the first direction (horizontally in the depicted orientation) is achieved by configuring the movement distance of the sample 208 during scanning to be equal to the pitch w of the sub-beam grid in the first direction (i.e., the width of the tessellated hexagon 720). The length of the elongated region and / or the length of the sub-beam processing area associated with each sub-beam can therefore be equal to the pitch w. Continuous coverage of the sample surface parallel to the second direction (vertically in the depicted orientation) is achieved by configuring the cumulative distance of step Nx between scans in the sub-beam processing area (e.g., width w). The dimension h of the sub-beam processing area 740 in the second direction is set to be equal to the pitch of the sub-beam grid in the second direction. The field of view 720 of the sub-beams of the multi-beam electro-optics system can be combined to form a multi-beam field of view 750, which is also hexagonal in shape. To scan a continuous region of the sample, the sample is stepped in a giant step 801, which is at an angle to the first direction, for example, in the range of 40 to 70 degrees. Scans at different steps can be at least connected, thus together covering a continuous region of the sample. (More generally, the sub-beam field of view 720 can have a similar size and shape embedded in a grid. The multi-beam field of view 750 can have a shape such that it is embedded above the surface of the sample with a similar field of view, thereby substantially covering the surface; that is, portions of the sample continuous in the multi-beam field of view 750 can be embedded together for a substantially continuous surface.)

[0155] In a specific instance where a sub-beam is provided using a hexagonal array, the pitch of the hexagonal array can range from 50 micrometers to 100 micrometers. The width of the rectangular sub-beam processing area 740 will be a multiple of the pitch of the hexagonal array. In an instance where the field of view of the objective lens in the objective array is approximately 1 micrometer, 50 to 100 mechanical scans of the elongated region 724 (or strip) are required to cover one sub-beam processing area. It should be noted that in different configurations, the sub-beam can be provided with an array of grids of different shapes, such as parallelograms, rhombuses, rectangles, or squares. For each shape of the beam configuration, the sub-beam processing area 740 can be rectangular.

[0156] To perform scanning, a processing area 740 is assigned to each beam of the multi-beam configuration. Depending on the electro-optical design, different processing strategies may be used by the assigned sub-beams to process their processing areas. In one embodiment, the sub-beams are mechanically scanned by a lateral stage, such that the sub-beams and the sample are scanned laterally across the processing area 740. For example, the sub-beams are electrostatically scanned orthogonal to or at least at an angle relative to the direction of lateral stage movement. Such electrostatic scanning can determine the width of elongated regions, for example, from 0.5 micrometers to 5 micrometers. The stage steps between the elongated regions of the processing area. All sub-beams are scanned on the sample surface of each assigned processing area. Thus, the sample surface of the grid assigned to the sub-beams is processed. The stage and the multi-beam array move relative to each other, for example, the stage is stepped, such that unprocessed areas of the sample surface are assigned to the grid of the sub-beams for processing, i.e., scanning. In another configuration, scanning is mechanical, for example, by stage movement. In an alternative configuration, scanning is electrical, for example, using a scanning deflector that is orthogonal to, or at least angular to, the direction of the elongated region, in order to define the breadth of the elongated region. In one embodiment, different combinations of mechanical and electrical scanning actuators are used (note that the scanning actuator may be a stage movement or an electrostatic scanning deflector).

[0157] Different types of overscanning, which can be used in the configurations shown in Figures 14A to 14C, are depicted in Figures 17 to 19. Generally, the term overscanning refers to scanning an area larger than the nominal area to be processed. Overscanning can be performed for various reasons: for example, to allow data from adjacent scans to be stitched together to form an image of a larger area; or to consider alignment errors (the location of the area of ​​interest is not precisely known); or for a combination of reasons. When using overscanning to avoid the stage alignment step in an image comparison method, the area scanned on the sample is larger than the image area being compared. In other words, image comparison is performed on a subset of scanned image data representing the overlapping area of ​​the scans. The size of the additional scanned area can be relatively small compared to the area of ​​the image being compared.

[0158] Figure 17 depicts a form of overscanning that can be termed strip overscanning. In strip overscanning, an additional boundary 741 is scanned by increasing the width of the strip 724 (or elongated region) in a direction perpendicular to the scanning direction on the sample surface. This can be achieved via electro-optical scanning of the sub-beam, where appropriate (e.g., using an electromagnetic scanning deflector and / or electronic scanning). The width of the scan can also be increased as needed by adjusting the mechanical scanning of the sample. Strip overscanning can address: beam or sample drift during scanning (e.g., drift in the relative position of the beam and the wafer); beam or sample drift between scans in inter-die mode; and / or stage movement matched to die spacing. As an example, the amount of strip overscanning can be up to about 10% of the strip width.

[0159] Figure 18 depicts one form of overscan, which may be referred to as pitch region overscan (or overscan for processing region 742). In pitch region overscan, an additional boundary 743 is scanned around the sub-beam processing region 742. This can be achieved by increasing the length of the scan strips 724 and / or increasing the width of one or more or all of the strips 724. Increasing the width of the strips 724 of the processing region can be achieved by applying strip overscan. Pitch region overscan may include strip overscan. Alternatively or additionally, pitch overscan may be achieved by one or more additional strips 724. If additional strips are used to achieve pitch region overscan, the additional strips should be included, for example, when deriving data for comparison by scanning the same beam for the pitch region (e.g., similar pitch regions in each of different grains). Pitch region overscan can resolve beam or sample drift and / or defects in the sub-beam grid during scanning of the entire sub-beam processing region. As an example, the amount of strip overscan can be as high as about 2% of the larger dimension of the sub-beam processing area.

[0160] This considers the comparison of images obtained from the same beam. Using the same beam may affect overscanning. By using the same beam to produce the images being compared, several errors are included in "pitch area overscanning". These errors can be defects in the beam grid and drift of the beam or wafer down to the processing area on the time scale. If different beams are used to produce the three images being compared, the defects will be characterized by "strip overscanning". This is advantageous because "pitch overscanning" can be much larger than "strip overscanning". Defects can be contained in a larger area, thus allowing for the generation of more defects, for example, that have a low impact on the comparison of image data.

[0161] Figure 19 depicts a form of overscanning that can be termed multi-beam field-of-view overscanning. In multi-beam field-of-view overscanning, the field of view is used selectively, rather than increasing the field of view of the multi-beam column (which would require the addition of additional beams). The width w or Wfov and height h or Hfov of the multi-beam field of view are maintained; the relative position of the multi-beam field of view 750 can be shifted to achieve overscanning. Multi-beam field-of-view overscanning can address: complete coverage of the selected area for detection, such as scanning a rectangular area on the sample using a hexagonal field of view (i.e., a shape different from the selected area for detection), inaccuracies in sample alignment, and / or malfunctioning beams. Therefore, adjacent scans of multiple multi-beam fields can be at least connected, such that adjacent scans scan a continuous area of ​​the sample surface. When the selected area for detection is completely covered, there is overlap between adjacent scans of the multi-beam field of view 750. The overlap between different adjacent scans of the multi-beam field of view is similar.

[0162] To implement the partial area scanning method, the scanning sequence is modified as depicted in FIG15. The first portion sr1-1 of the first pattern repeating area pr1, referred to as the scanning area, is scanned using a sub-beam of the electro-optical system. The sample 208 is displaced (stepped) under the electro-optical system such that the sub-beam is positioned at the starting point of the second scanning area sr2-1 of the second pattern repeating area pr2, which is scanned next. The sample 208 steps to the starting point of the third scanning area sr3-1 of the scanned third pattern repeating area pr3. All of the first to third scanning areas sr1-1, sr2-1, and sr3-1 are related to the corresponding portions of the pattern repeating. The data stream generated from the scanning of the first to third scanning areas sr1-1, sr2-1, and sr3-1 is processed as described above to detect defects. This procedure is repeated for the fourth to sixth scanning areas sr1-2, sr2-2, and sr3-2. It should be understood that the pattern regions pr1, pr2, and pr3 do not need to be positioned in a straight line as shown in Figure 15, but can be located anywhere on the sample and can be adjacent or spaced apart. The scanning areas sr1-n, sr2-n, and sr3-n of the three pattern repeating regions pr1, pr2, and pr3 can be scanned consecutively until selected portions (which may be all or part) of each pattern repeating region pr1, pr2, and pr3 are scanned. The scanning areas do not need to be in the same direction. In this method, the length of each scan may differ from the length discussed above with reference to Figures 14A, 14B, and 14C, and is more determined by the size (field pitch) of the pattern repeating region than by the characteristics of the electron-optical system 40. Therefore, selected portions of the pattern repeating region can be scanned as needed using at least a number of adjacent scans of the sub-beam grid. To ensure that no portion of the sample surface is missed, adjacent scans of the multi-beam configuration can overlap, preferably to a substantially minimal amount, to reduce the impact on yield. Figure 20 illustrates the steps and scanning sequence of an upper partial region scanning method using a hexagonal multi-beam field of view, wherein the sample has a linear grid of pattern repeating regions 760. Step 802 is performed between the first portions of the three regions, and step 803 is a return step to start a new portion of the first region.

[0163] Some charged particle evaluation systems are known to require alignment after each stage movement to address alignment drift between the stage (and therefore the sample held thereon) and the paths of multiple beams. In other systems, the drift may be sufficiently limited to require, but still exist, alignment. Therefore, precise stage movement when stepping between different sample portions for scanning can make it difficult to ensure that each beam scans the corresponding portion of each pattern repeating region (e.g., a grain). As mentioned above, it is preferable to use the same beam to scan the corresponding portion of each pattern repeating region because it avoids the need to calibrate the displacement of each beam path from the ideal beam path.

[0164] It is proposed to use overscanning (specifically, strip overscanning as described above) to avoid or reduce time-consuming alignment steps. Strip overscanning can be performed to increase the length and / or width of the scanned strip. The amount of overscanning required depends on the amount of drift experienced between scans. In cases where overscanning is used to avoid alignment steps in charged particle evaluation tools, additional image alignment steps may be required, such as during processing for analysis of scanned image data.

[0165] For defect detection purposes, a set of scan areas is scanned in relation to corresponding portions of a plurality of (e.g., three) different pattern repeating areas. Multiple sets of scan areas are scanned to cover more pattern repeating areas. Within each set, it is not necessary to scan in the same order. Nor is it necessary to scan the entire pattern repeating area. Scanning and defect detection may be limited to the portion of the pattern repeating area where defects are most likely to occur. Locations where defects may be present can be predicted from simulation, as disclosed in US 2019 / 0006147 A1, which is incorporated herein by reference to its disclosure of predicted defect hotspot locations.

[0166] It is also unnecessary to scan one scan area from each of the pattern repeating areas pr1 to pr3, and then immediately scan the other scan areas from the same set of pattern repeating areas. A different set of pattern repeating areas can be scanned first. However, to reduce data buffering, it is necessary to scan all scan areas to be used in a given defect detection test consecutively.

[0167] The sequence of scanning and data processing actions is depicted in Figure 16, which is a swimlane diagram illustrating the actions performed by the electron beam system 40, input buffer 554, comparator 552, second processing module 560, and output buffer 570. As the scan areas sr1-1, sr2-1, and sr3-1 are scanned sequentially, information accumulates in the input buffer 554. Depending on the nature of the first defect detection test, processing can begin once data related to the second defect in the scan area is available. For example, if two scan images are to be averaged to form a reference for comparison with a third scan image, or once data from the third scan area is available.

[0168] When a defect is detected by the first processing module, the data is sent to the second processing module 560. The second processing module 560 can then begin applying a second defect detection test. Information indicating the defect detected in the second defect detection test is sent to the output buffer 570. Where necessary, the processing of the first set of scan areas by the first processing module does not take more time than scanning those areas. The time spent stepping the sample to the start of scanning the next set of scan areas ensures that data processing does not limit output. (As discussed above, it should be noted that the next set of scan areas does not need to be related to the same pattern repetition area as the first set of scan areas.)

[0169] References to the upper and lower parts, upward and downward, above and below, etc., should be understood as directions parallel to the (usually but not always vertical) countercurrent and downstream directions of the electron beam or multiple beams irradiating the sample 208. Therefore, references to the countercurrent and downstream directions are intended to refer to directions independent of any current gravitational field relative to the beam path.

[0170] The embodiments described herein may take the form of a series of aperture arrays or electro-optical elements arranged in an array along a beam or multi-beam path. Such electro-optical elements may be electrostatic. In one embodiment, for example, all electro-optical elements in a sub-beam path prior to a sample, from the beam-limiting aperture array to the final electro-optical element, may be electrostatic and / or may be in the form of an aperture array or a plate array. In some configurations, one or more of the electro-optical elements are fabricated as microelectromechanical systems (MEMS) (i.e., using MEMS fabrication techniques). Electro-optical elements may have both magnetic and electrostatic elements. For example, a compound array lens may be characterized by a giant magnetic lens covering a multi-beam path, having upper and lower electrodes arranged within the magnetic lens and along the multi-beam path. An aperture array for the beam path of the multi-beam may be present in these electrodes. Electrodes may be present above, below, or between the electrodes to control and optimize the electromagnetic field of the compound lens array.

[0171] The evaluation tool or evaluation system according to the present invention may include equipment for performing qualitative evaluation of samples (e.g., pass / fail), equipment for performing quantitative measurement of samples (e.g., size of features), or equipment for generating an image of a sample. Examples of evaluation tools or systems are inspection tools (e.g., for identifying defects), inspection tools (e.g., for classifying defects), and metrology tools, or any combination of tools capable of performing evaluation functionality associated with inspection tools, inspection tools, or metrology tools (e.g., metrology inspection tools).

[0172] Reference to a system of components or components or elements for controlling the manipulation of a charged particle beam in a certain manner includes: configuring a controller or control system or control unit to control the components to manipulate the charged particle beam in the manner described, and using other controllers or devices (e.g., voltage supply) as appropriate to control the components thereby manipulating the charged particle beam in this manner. For example, a voltage supply may be electrically connected to one or more components to apply a potential to such components under the control of a controller or control system or control unit, such as to the electrodes of the control lens array 250 and objective lens array 241. Actuable components such as a stage may be controllable to actuate other components such as the beam path and thus move relative to other components such as the beam path using one or more controllers, control systems or control units for controlling the actuation of the component.

[0173] The functionality provided by the controller, control system, or control unit can be implemented by a computer. Any suitable combination of components can be used to provide the required functionality, including, for example, a CPU, RAM, SSD, motherboard, network connection, firmware, software, and / or other components known in the art that allow the execution of the required computational operations. The required computational operations can be defined by one or more computer programs. One or more computer programs can be provided on media, and in some cases, non-transitory media, storing the formation of computer-readable instructions. When the computer-readable instructions are read by the computer, the computer executes the required method steps. The computer can consist of a self-contained unit or a distributed computing system having a plurality of different computers interconnected via a network.

[0174] The terms "sub-beam" and "splitter" are used interchangeably herein and are both understood to encompass any radiation beam derived from the parent radiation beam by dividing or splitting the parent radiation beam. The term "manipulator" is used to encompass any element, such as a lens or deflector, that affects the path of a sub-beam or split beam. Reference to an element aligned along a beam path or sub-beam path should be understood to mean that the individual element is positioned along the beam path or sub-beam path. Reference to an optics device should be understood to mean an electro-optics device.

[0175] The method of the present invention can be executed by a computer system comprising one or more computers. The computer used to implement the present invention may include one or more processors, including a general-purpose CPU, a graphics processing unit (GPU), a field-programmable gate array (FPGA), an application-specific integrated circuit (ASIC), or other dedicated processors. As discussed above, in some cases, a particular type of processor may offer advantages in terms of reduced cost and / or increased processing speed, and the method of the present invention is applicable to the use of a particular processor type. Certain steps of the method of the present invention involve parallel computing suitable for implementation on processors capable of performing parallel computing (e.g., GPUs).

[0176] As used herein, the term "image" is intended to refer to any array of values, where each value is a sample of a location and the arrangement of values ​​in the array corresponds to the spatial arrangement of the sampled locations. An image may comprise a single layer or multiple layers. In the case of a multi-layer image, it may also refer to each layer of a channel representing a different sample of a location. The term "pixel" is intended to refer to a single value in the array, or, in the case of a multi-layer image, a group of values ​​corresponding to a single location.

[0177] The computer used to implement the present invention may be physical or virtual. The computer used to implement the present invention may be a server, client, or workstation. Multiple computers used to implement the present invention may be distributed and interconnected via a local area network (LAN) or a wide area network (WAN). The results of the method of the present invention may be displayed to a user or stored in any suitable storage medium. The present invention may be embodied in a non-transitory computer-readable storage medium storing instructions to carry out the method of the present invention. The present invention may be embodied in a computer system comprising one or more processors and memory or storage for storing instructions to carry out the method of the present invention.

[0178] The present invention is described in the following numbered clauses.

[0179] Clause 1: A data processing apparatus for detecting defects in sample image data generated by a charged particle evaluation system, the apparatus comprising: a first processing module configured to receive a sample image data stream from the charged particle evaluation system, the sample image data stream comprising a series of ordered data points representing an image of the sample, and the first processing module applying a first defect detection test to select a subset of the sample image data stream as first selected data, wherein the first defect detection test is a localized test performed in parallel with receiving the sample image data stream; and a second processing module configured to receive the first selected data and applying a second defect detection test to select a subset of the first selected data as second selected data.

[0180] Item 2: The apparatus of Item 1, wherein the first processing module uses a first number of operations per pixel to apply the first defect detection test, and the second processing module uses a second number of operations per pixel to apply the second defect detection test, wherein the first number of operations is less than the second number of operations.

[0181] Item 3: The apparatus as described in Item 1 or 2, wherein the first processing module uses less than 200 operations per pixel, and less than 100 operations as needed, to apply the first defect detection test.

[0182] Item 4: The device as described in Items 1, 2 or 3, wherein the operation performed by the first processing module is selected from the group consisting of: AND, OR, NOT, NAND, XOR, addition, subtraction, bit shift.

[0183] Item 5: The apparatus of item 1, 2, 3 or 4, wherein the first defect detection test has a selectivity lower than that of the second defect detection test.

[0184] Clause 6: The apparatus of any of the preceding clauses, wherein the first processing module performs a convolution of the data points of the sample image data stream with a kernel having a predetermined size.

[0185] Clause 7: The apparatus of any of the preceding clauses, wherein the first processing module includes an input buffer configured to buffer the sample image data stream, preferably wherein the input buffer has a capacity less than that required to store an image of an entire pattern repeating region of the sample.

[0186] Clause 8: The apparatus of any of the preceding clauses, wherein the first processing module compares the data points of the sample image data stream with the first reference image data.

[0187] Item 9: The apparatus of Item 8, wherein the first processing module includes a reference buffer configured to buffer and temporarily store the first reference image data, preferably wherein the reference buffer has a capacity less than that required to store the first reference image data of an entire pattern repeating region of the sample.

[0188] Item 10: The apparatus as described in Item 8 or 9, wherein the first reference image data has a lower resolution compared to the image of the sample, preferably, the second processing module compares the first selected data with the second reference image data, preferably, the first reference image data is a lower resolution version of the second reference image data.

[0189] Item 11: The apparatus as described in any of the preceding items, wherein the first processing module comprises a field-programmable gate array or an application-specific integrated circuit.

[0190] Clause 12: The apparatus as described in any of the preceding clauses, wherein the first selected data comprises a pixel region surrounding the pixels of the image data that satisfy the first defect detection test.

[0191] Clause 13: The apparatus of any of the preceding clauses, wherein the first defect detection test generates a first defect score indicating the probability that a pixel of the sample image represents a defect, and the first processing module further includes an output buffer of the first selected data configured to accumulate data of the region having the highest value of the first defect score.

[0192] Clause 14: A charged particle assessment system comprising a charged particle beam system and a data processing device as described in any of the preceding clauses.

[0193] Item 15: The charged particle evaluation system of Item 14, wherein the charged particle beam system and the first processing module are located in a vacuum chamber, and the second processing module is located outside the vacuum chamber.

[0194] Item 16: The charged particle evaluation system as described in Item 14 or 15, wherein the charged particle beam system is a multi-beam beam system.

[0195] Clause 17: The charged particle evaluation system of Clause 16, wherein there are a plurality of first processing modules, each first processing module being associated with one of the columns of the multi-column beam system, and there are fewer second processing modules than the first processing modules.

[0196] Item 18: The charged particle assessment system of Item 16, wherein the first processing module is configured to receive scanned image data from detectors associated with corresponding beams in a plurality of columns of the multi-column beam system.

[0197] Clause 19: A charged particle evaluation system as described in Clauses 16, 17 or 18, wherein the columns of the multi-column beam system are configured such that the same beam is used to scan corresponding portions of a plurality of pattern repeating regions, preferably, the columns are spaced apart by a distance corresponding to one of the distances between the plurality of pattern repeating regions, preferably, wherein the pattern repeating regions are different grains on the sample.

[0198] Clause 20: A method for detecting defects in a sample having a plurality of pattern repeating regions using a charged particle beam system, the method comprising, in sequence: a first scan of a first portion of a first region of the pattern repeating regions to generate first scan image data; a second scan of a first portion of a second region of the pattern repeating regions to generate second scan image data, the second region being spaced apart from the first region; a third scan of a first portion of a third region of the pattern repeating regions to generate third scan image data, the third region being spaced apart from the first and second regions; and a subsequent scan of a second portion of the first region to generate fourth scan image data. Desirably, the pattern repeating regions of the sample are regions containing patterns repeated elsewhere on the sample, such as one or more of the first, second, and third regions; desirably, all of the first, second, and third regions. Desirably, the pattern repeating regions are nominally identical. Desirably, the pattern repeating regions correspond to a grain or target portion of the sample. In one embodiment, the pattern repeating region may be smaller than the grain. In another embodiment, the pattern repeating region may correspond to a plurality of combined grains.

[0199] Item 21: The method of Item 20, wherein during each of the first scan, the second scan, the third scan and the fourth scan, the sample is scanned in a single direction relative to the charged particle beam system.

[0200] Item 22: The method of Item 20 or 21, wherein the charged particle beam system has a plurality of beams, and the first scan, the second scan and the third scan are performed such that the same beam is used to scan the corresponding portions of the respective pattern repeating regions.

[0201] Article 23: The method of Articles 20, 21 or 22, wherein it further comprises processing the first scan data, the second scan data and the third scan data to detect a defect in one of the first pattern repeating area, the second pattern repeating area or the third pattern repeating area.

[0202] Article 24: The method of Article 23, wherein processing the first scan data, the second scan data and the third scan data includes processing a subset of each of the first scan data, the second scan data and the third scan data representing corresponding regions of the first region, the second region and the third region.

[0203] Clause 25: The method of Clause 23 or 24, wherein the processing comprises: a first subroutine that applies a first defect detection test to select a subset of one of the first scan image data, the second scan image data and the third scan image data as first selected data; and a second subroutine that applies a second defect detection test to select a subset of the first selected data as second selected data.

[0204] Item 26: The method of Item 25, wherein the time taken by the first subroutine is less than or equal to the time interval between the start of the first scan and the start of the fourth scan.

[0205] Item 27: The method of any one of items 20 to 26, wherein each of the first scan image data, the second scan image data and the third scan image data comprises data from scans of a plurality of multi-beam fields of view on the sample.

[0206] Item 28: The method of Item 27, wherein the adjacent scans of the plurality of multi-beam fields of view are at least connected on the sample.

[0207] Item 29: As in Item 27 or 28, the plurality of multi-beam fields of view cover the respective sample regions corresponding to the first image data, the second image data and the third image data.

[0208] Clause 30: The method of any one of Clauses 27 to 29, wherein each of the plurality of multi-beam fields of view comprises a plurality of processing regions, each processing region corresponding to a field of view of one of the plurality of beams.

[0209] Clause 31: The method of Clause 30, wherein the first image data, the second image data and the third image data derived from the scan of at least one of the multi-beam fields of view are achieved by overscanning.

[0210] Clause 32: The method of Clause 31, wherein in the overscan, at least one of the plurality of beams associated with one of the multiple beam fields of view is overscanned by pitch overscan.

[0211] Item 33: The method of Item 32, wherein in the pitch overscan, one region of the sample scanned by the associated beam is larger than the processing region of the associated beam.

[0212] Clause 34: The method of Clause 32 or 33, wherein in the pitch overscan, one region of the sample scanned by the associated beam is at least connected to one region of the sample scanned by a beam adjacent to the associated beam of the plurality of beams (e.g., a processing region).

[0213] Clause 35: The method of any one of Clauses 32 to 34, wherein the pitch overscan comprises: extending multiple scans across the sample, so that preferably, the corresponding elongated scan region is longer than the width of the processing region; widening the width (or strip overscan) of one or more scans across the sample, so that preferably, the corresponding elongated scan includes an elongated region and an additional boundary, preferably, the additional boundary is connected at least to the region of the sample scanned in the adjacent elongated scan; and / or scanning one or more additional scans across the sample, so that preferably, the scanned region of the sample is larger than the processing region in the sum of the regions scanned by the scans across the sample.

[0214] Clause 36: The method of any one of Clauses 32 to 35, wherein the overscan is used, at least one of the plurality of beams ensures that similar beams of the plurality of beams are used to generate individual data of the first image data, the second image data and the third image data for processing to detect defects.

[0215] Clause 37: The method of any one of Clauses 20 to 36 above, wherein the subsequent scan scans at least one region of the sample that is connected to at least one of the first region, the second region and the third region.

[0216] Clause 38: The method of Clause 37, wherein overscanning is used to ensure that the area scanned for subsequent scans is connected to at least one of the first, second and third regions, preferably, the overscanning includes at least one of multi-beam field-of-view overscanning, pitch overscanning and strip scanning.

[0217] Item 39: The method of any one of items 20 to 38 further includes buffering the first scan image data at least until the second scan image data is generated.

[0218] Item 40: The method of any one of items 20 to 39 further includes buffering the first scan image data and the second scan image data at least until the third scan image data is generated.

[0219] Item 41: The method of any one of items 20 to 40 further includes buffering the first scan image data, the second scan image data and the third scan image data.

[0220] Clause 42: A method for detecting defects in a sample using a charged particle beam system with one charged particle multi-beam system comprising a plurality of sub-beams, the sample having a plurality of pattern repeating regions nominally being identical regions, the method comprising, in sequence: using the charged particle multi-beam system to perform a first scan of one of the first pattern repeating regions of the sample to generate a first scan image data; using the charged particle multi-beam system to perform a second scan of one of the second pattern repeating regions of the sample to generate a second scan image data, the second pattern repeating region being spaced apart from the first pattern repeating region; using the charged particle multi-beam system to perform a third scan of one of the third pattern repeating regions of the sample to generate a second scan image data. Three scan image data, wherein the third pattern repeating region is spaced apart from the first pattern repeating region and the second pattern repeating region; and the first scan image data, the second scan image data and the third scan image data are processed by comparing the first scan image data, the second scan image data and the third scan image data to identify a defect; wherein each sub-beam is assigned a portion of the first pattern repeating region, the second pattern repeating region and the third pattern repeating region such that each scan includes scanning the corresponding portion of the respective pattern repeating regions using the same sub-beam; wherein the processing includes comparing scan image data obtained from the corresponding portion of the respective pattern repeating regions by the same sub-beam.

[0221] Clause 43: A method for detecting defects in a sample using a charged particle beam system with one of a plurality of sub-beams of charged particles, the sample having a plurality of nominally identical pattern repeating regions, the method comprising: sequentially scanning at least three different pattern repeating regions of the sample using the charged particle multi-beam to generate different first sample image data sets; processing the different first sample image data sets by comparing the image data of each region to identify a defect; wherein each sub-beam is assigned a corresponding portion of each of the pattern repeating regions such that each scan comprises scanning the corresponding portion of the respective pattern repeating regions using the same sub-beam.

[0222] Item 44: A data processing method for detecting defects in sample image data generated by a charged particle evaluation system, the method comprising: receiving a sample image data stream from the charged particle evaluation system, the sample image data stream comprising a series of ordered data points representing an image of the sample; applying the first defect detection test as a localized test, the application of the first detection test comprising selecting a subset of the sample image data stream as first selected data, wherein the application of the first defect detection test is performed in parallel with receiving the sample image data stream; receiving the first selected data; applying a second defect detection test, the application of the second detection test comprising selecting a subset of the first selected data as second selected data.

[0223] Item 45: A data processing method for detecting defects in data (e.g., images) generated by an evaluation system for evaluating a sample, the method comprising: receiving data generated by the evaluation system, the data representing an image of a sample; applying a first test to select a subset of the data; and applying a second test to the subset to verify whether at least a portion of the subset of the data represents at least one defect.

[0224] Item 46: The data processing method as described in Item 45, wherein the data is received in a data stream.

[0225] Item 47: The data processing method of Item 46, wherein the first test is applied to the data in the data stream.

[0226] Item 48: The data processing method of any one of items 45 to 47, wherein the first test includes a simple algorithm for application to the data, such as a localization test.

[0227] Item 49: The data processing method of any of items 45 to 48, wherein the second test is used to apply to the data, and to apply as needed to one of the complex algorithms of the subset of the data.

[0228] Item 50: The data processing method of any one of items 45 to 49, wherein the application of the first test and the receiving of the data are performed in parallel.

[0229] Item 51: A data processing method for detecting defects in sample image data generated by an evaluation system, the method comprising: receiving a sample image data stream from the evaluation system, the sample image data stream comprising a series of ordered data points representing one image of the sample; applying the receiving of the sample image data stream in parallel with a first defect detection test as one of a localization test, the application of the first detection test comprising selecting a subset of the sample image data stream as first selected data; receiving the first selected data; and applying a second defect detection test to the first selected data and comprising selecting a subset of the first selected data as second selected data representing a defect.

[0230] Item 52: The data processing method of any of Items 45 to 51, wherein the evaluation system is a charged particle evaluation system.

[0231] Clause 53: A computer program comprising instructions configured to control a processor to perform the methods described in any of Clauses 20 to 52.

[0232] Although the invention has been described in conjunction with various embodiments, other embodiments of the invention will become apparent to those skilled in the art from consideration of this specification and practice of the invention disclosed herein. This specification and examples are intended to be illustrative only, wherein the true scope and spirit of the invention are indicated by the following claims. [Simplified Explanation of the Diagram]

[0012] The above and other aspects of the present invention will become more apparent from the description of the exemplary embodiments taken in conjunction with the accompanying drawings.

[0013] Figure 1 is a schematic diagram illustrating an exemplary charged particle beam detection system.

[0014] Figure 2 is a schematic diagram illustrating an exemplary multi-beam charged particle evaluation device, which is part of the exemplary charged particle beam detection system of Figure 1.

[0015] Figure 3 is a schematic diagram of an exemplary electron optical column containing a condenser lens array.

[0016] Figure 4 is a schematic diagram of an exemplary electron optical column including a giant collimator and a giant scanning deflector.

[0017] Figure 5 is a schematic diagram of an exemplary electron optical column including a beam splitter.

[0018] Figure 6 is a schematic cross-sectional view of the objective array of a charged particle evaluation system according to an embodiment.

[0019] Figure 7 is a bottom view of the modified objective array of Figure 7.

[0020] Figure 8 is a schematic diagram of an illustrative single-beam electron optical column.

[0021] Figure 9 is a schematic diagram of a data path according to an embodiment.

[0022] Figure 10 is a flowchart of a method according to an embodiment.

[0023] Figure 11 is a schematic diagram of a data processing apparatus according to an embodiment.

[0024] Figure 12 is a diagram of a uniform kernel in one embodiment.

[0025] Figure 13 is an example of a SEM image on which the method of the present invention can be performed.

[0026] Figures 14A, 14B and 14C are schematic diagrams of the field of view of the sub-beam, the sub-beam processing area and the field of view of the multi-beam in the charged particle evaluation system.

[0027] Figure 15 is a schematic diagram of the scanning area and pattern repetition area on the sample.

[0028] Figure 16 is a swimlane diagram of the method of the second embodiment, showing more details.

[0029] Figure 17 is a diagram illustrating the overscanning of each band.

[0030] Figure 18 is a diagram illustrating the overscanning of the repeating area of ​​each pattern.

[0031] Figure 19 is a diagram illustrating the overscanning of each field of view of a multi-beam electron optical column.

[0032] Figure 20 is a diagram illustrating the stepping and scanning sequences.

[0033] The schematic diagrams and views illustrate the components described below. However, the components depicted in the diagrams are not drawn to scale.

Claims

1. A method for detecting defects in a sample having a plurality of pattern repeating regions using a charged particle beam system, the method comprising, in sequence: a first scan of a first portion of a first region of the pattern repeating regions to generate first scan image data; a second scan of a first portion of a second region of the pattern repeating regions to generate second scan image data, the second region being spaced apart from the first region; a third scan of a first portion of a third region of the pattern repeating regions to generate third scan image data, the third region being spaced apart from the first region and the second region; and a subsequent scan of a second portion of the first region to generate fourth scan image data.

2. The method of claim 1, wherein one pattern repeating region of the sample is a region containing one pattern that repeats elsewhere on the sample.

3. The method of claim 1 or 2, wherein it further comprises processing the first scan data, the second scan data and the third scan data to detect defects in one of the first pattern repeating area, the second pattern repeating area or the third pattern repeating area.

4. The method of claim 3, wherein processing the first scan data, the second scan data, and the third scan data includes processing a subset of each of the first scan data, the second scan data, and the third scan data representing corresponding regions of the first region, the second region, and the third region.

5. The method of claim 3, wherein the processing comprises: a first subroutine that applies a first defect detection test to select a subset of one of the first scan image data, the second scan image data, and the third scan image data as first selected data; and a second subroutine that applies a second defect detection test to select a subset of the first selected data as second selected data.

6. The method of request 5, wherein the time taken by the first subroutine is less than or equal to the time interval between the start of the first scan and the start of the fourth scan.

7. The method of claim 1 or 2, wherein the charged particle beam system has a plurality of beams, and the first scan, the second scan, and the third scan are performed such that the same beam is used to scan corresponding portions of respective pattern repeating regions.

8. The method of claim 1 or 2, wherein each of the first scan image data, the second scan image data and the third scan image data comprises data from scans of a plurality of multi-beam fields of view on the sample.

9. The method of claim 8, wherein the adjacent scans of the plurality of multi-beam fields of view are at least connected on the sample.

10. The method of claim 8, wherein each of the plurality of multi-beam fields of view comprises a plurality of processing regions, each processing region corresponding to one field of view of one of the plurality of beams.

11. The method of claim 10, wherein the first image data, the second image data and the third image data derived from scanning at least one of the multi-beam fields of view are achieved by overscanning.

12. The method of claim 11, wherein in the overscan, at least one of the plurality of beams associated with one of the multi-beam fields of view is overscanned by pitch overscan.

13. The method of claim 11, wherein the overscan is used to ensure that at least one of the plurality of beams ensures that similar beams of the plurality of beams are used to generate individual data of the first image data, the second image data and the third image data for processing to detect defects.

14. The method of claim 1 or 2, wherein the subsequent scan scans a region of the sample that is connected to at least one of the first region, the second region and the third region.

15. The method of claim 1 or 2 further includes buffering the first scan image data at least until the second scan image data is generated.

Images