Detecting defects on a sample

Abstract

Methods and systems for detecting defects on a sample are provided. One system performs dual detection in which at least one of the reference images compared to a test image is a calculated reference image generated from a plurality of images corresponding to the test image. Another reference image may or may not be calculated from more than one of the plurality of images. This calculated reference image may also be a median-based calculated reference generated from a multi-median image generated from different subsets of in-job images of images generated by the inspection subsystem for the sample. The system may also group images for rows of dies on a sample into different jobs based on color, such that the different jobs have different ranges of color values. This grouping may also be performed such that each of the jobs includes a number of images greater than a predetermined minimum job size.

Description

Technical Field

[0001] The present invention generally relates to methods and systems for detecting defects on a sample in the semiconductor field. Background Art

[0002] The following description and examples are not considered to be background art by virtue of their inclusion in this section.

[0003] Fabricating semiconductor devices such as logic and memory devices typically includes processing a sample such as a semiconductor wafer using a number of semiconductor fabrication processes to form various features and multiple levels of the semiconductor device. For example, photolithography is a semiconductor fabrication process that typically involves transferring a pattern to a resist disposed on a semiconductor wafer. Additional examples of semiconductor fabrication processes include, but are not limited to, chemical mechanical polishing, etching, deposition, and ion implantation. Multiple semiconductor devices may be fabricated on a single semiconductor wafer in an arrangement and then separated into individual semiconductor devices.

[0004] In the semiconductor industry, inspection using optical or electron beam imaging is an important technology for debugging the semiconductor manufacturing process, monitoring process variations, and improving production yield. With the continuous reduction in the scale of modern integrated circuits (ICs) and the continuous increase in the complexity of the manufacturing process, inspection has become increasingly difficult.

[0005] In each processing step performed on a semiconductor wafer, the same circuit pattern is printed in each die. Most wafer inspection systems take advantage of this fact and use relatively simple die-to-die comparisons to detect defects on the wafer. However, the printed circuits in each die may include many areas with patterned features repeated in the x or y direction, such as areas of DRAM, SRAM, or FLASH. This type of area is usually called an "array area" (the remaining areas are called "random areas" or "logic areas"). In order to obtain better sensitivity, advanced inspection systems use different strategies to inspect array areas and random areas or logic areas.

[0006] One problem with inspection in the logic area is that the patterns are often repeated at relatively large spacings, e.g., from die to die, rather than within a die as is the case with the array area. For inspection, this means that for a test image, a suitable reference image may be from an adjacent die that differs significantly from the test image. Those differences between dies may not be due to defects, but rather to non-defect noise sources on the sample (e.g., thickness and color variations) and / or differences in the processing of the die that are not a problem for the device itself but cause the images produced by the die to differ (e.g., variations from one lithography exposure to another). However, those differences may be falsely detected as defects, which obviously creates several other problems with the usefulness of the inspection results.

[0007] Therefore, several different methods have been developed to obtain or generate suitable reference images for die-to-die type comparisons. Some of the initial attempts to generate suitable reference images involved simulating reference images from design data for the sample. However, such simulations often have their own problems, such as difficulty accounting for the actual pattern on the sample and non-defect related variations in the images generated from die-to-die and / or sample-to-sample (when the same reference image is used for multiple samples). Therefore, other attempts involved generating reference images from actual images of the sample.

[0008] One way to generate a reference image from an actual sample image includes computing a reference image from images of multiple instances of the same patterned area on the sample (e.g., images of multiple dies on the sample). Such a reference is often referred to as a "computed reference" or CR. The theory behind combining multiple actual sample images to generate a CR is that any defect signal in any of the combined images will be mitigated by the combining process, thereby generating a reference image that preferably represents a "defect-free" instance of a repeating patterned area.

[0009] An example of a currently used CR method is to process 8 adjacent frames from the same die row in one job. In CR, for each target test frame (Test), all other test frames in the job are selected as base frames to be combined into a single reference frame (Ref) with minimal differences between Test and Ref. Defects can then be detected in the comparison: Test-Ref.

[0010] The currently used CR methods and systems for defect detection do have several important disadvantages. For example, the currently used CR methods and systems require the working frames to have substantially similar colors (grayscale). If the working frames have relatively large color changes, the generated Ref frame is difficult to match with the Test, which results in relatively low defect detection sensitivity and substantially high noise rate. In addition, defects on the CR base frame can cause artifacts in the currently used CR methods and systems.

[0011] Therefore, it would be advantageous to develop methods and systems for sample testing that do not suffer from one or more of the disadvantages described above. Summary of the invention

[0012] The following description of various embodiments should in no way be construed as limiting the subject matter of the appended claims.

[0013] One embodiment relates to a system configured for detecting defects on a sample. The system includes an inspection subsystem configured to generate an image on the sample, the image including a test image, two or more first images corresponding to the test image, and one or more second images corresponding to the test image. The system also includes a computer subsystem configured to calculate a first reference image and a second reference image of the test image from the two or more first images and the one or more second images, respectively. One or more characteristics of the first reference image and the second reference image are different. The computer subsystem is configured to generate a first difference image and a second difference image by separately comparing the test image with the first reference image and the second reference image, respectively. In addition, the computer subsystem is configured to detect defects in the test image by applying a defect detection method to the first difference image and the second difference image. The defect detection method determines that the defect exists at a location in the test image only if the defect detection method detects a defect at a corresponding location in the first difference image and the second difference image. The system may be further configured as described herein.

[0014] Another embodiment relates to a computer-implemented method for detecting defects on a sample. The method includes acquiring an image generated by an inspection subsystem for a sample, the image including a test image, two or more first images corresponding to the test image, and one or more second images corresponding to the test image. The method also includes the calculation, generation, and detection steps as described above. The acquisition, calculation, generation, and detection are performed by a computer subsystem coupled to the inspection subsystem. Each of the steps of the method described above may be performed as further described herein. In addition, the embodiment of the method described above may include any other steps of any other method described herein. The method described above may be performed by any of the systems described herein.

[0015] Another embodiment relates to a non-transitory computer-readable medium storing program instructions executable on a computer system for performing a computer-implemented method for detecting defects on a sample. The computer-implemented method includes the steps of the method described above. The computer-readable medium may be further configured as described herein. The steps of the computer-implemented method may be executed as further described herein. Additionally, the computer-implemented method may include any other steps of any other method described herein, the program instructions being executable to achieve the computer-implemented method.

[0016] Yet another embodiment relates to another system configured for detecting defects on a sample. This system includes an inspection subsystem configured for generating images of the sample. The system also includes a computer subsystem configured for grouping the images generated by the inspection subsystem for rows of dies on the sample into different jobs based on color, such that the images in a first job of the different jobs have a first color value range, and the images in a second job of the different jobs have a second color value range different from the first color value range. The computer subsystem is also configured for detecting defects in the images in the first job of the different jobs by applying a defect detection method only to the images in the first job of the different jobs. Additionally, the computer subsystem is configured for detecting defects in the images in the second job of the different jobs by applying the defect detection method only to the images in the second job of the different jobs. The system may be further configured as described herein.

[0017] Additional embodiments relate to a different system configured for detecting defects on a sample. This system includes an inspection subsystem configured for generating an image of the sample. This system also includes a computer subsystem configured for computing a multiple median image from different subsets of images in a job of the images generated by the inspection subsystem for the sample. The computer subsystem is also configured for computing a median-based calculated reference from the multiple median images. Additionally, the computer subsystem is configured for generating a difference image by comparing a test image in the job to the median-based calculated reference, and detecting defects in the test image by applying a defect detection method to the difference image. The system may be further configured as described herein. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Further advantages of the present invention will become apparent to those skilled in the art upon having the benefit of the following detailed description of the preferred embodiments and upon reference to the accompanying drawings, in which:

[0019] Figure 1 and 1a is a schematic diagram illustrating a side view of an embodiment of a system configured as described herein;

[0020] Figure 2 is a schematic diagram illustrating a plan view of an example of a job layout currently used on a sample;

[0021] Figure 3 is a schematic diagram illustrating a plan view of one embodiment of a job layout having similarity groupings within a row of dies on a sample;

[0022] Figure 4 is a flow chart illustrating one embodiment of steps that may be performed to detect defects on a sample;

[0023] Figure 5 is a schematic diagram illustrating an embodiment of steps that may be performed to detect defects on a sample;

[0024] Figures 6 to 7 is a schematic diagram illustrating an embodiment of a plan view of an image generated by an inspection subsystem for a sample and how the image can be used to calculate a reference image; and

[0025] Figure 8 is a block diagram illustrating one embodiment of a non-transitory computer-readable medium storing program instructions for causing a computer system to perform the computer-implemented methods described herein.

[0026] Although the invention is susceptible to various modifications and alternative forms, specific embodiments thereof are shown in the drawings by way of example and will be described in detail herein. The drawings may not be drawn to scale. However, it should be understood that the drawings and detailed description of the drawings are not intended to limit the invention to the specific forms disclosed, but on the contrary, the invention is intended to cover all modifications, equivalent forms and alternative forms within the spirit and scope of the invention as defined by the appended claims. DETAILED DESCRIPTION

[0027] Turning now to the drawings, it should be noted that the drawings are not drawn to scale. In particular, the proportions of some of the elements in the drawings are greatly exaggerated to emphasize the characteristics of the elements. It should also be noted that the drawings are not drawn to the same scale. Elements that may have similar configurations shown in more than one figure have been indicated using the same element numbers. Unless otherwise mentioned herein, any of the elements described and shown may include any suitable commercially available elements.

[0028] In general, the embodiments described herein are configured for detecting defects on a sample. The embodiments described herein are particularly useful for defect detection using color-adaptive thresholding (for separating defects from non-defects), such as the multi-color adaptive thresholding (MCAT) algorithm used by some inspection tools commercially available from KLA Corp., Milpitas, Calif. However, the embodiments described herein provide several improvements to such an MCAT algorithm, and so may be referred to as MCAT2. Such improvements include a new computed reference (CR) generation method, CR performed with dual detection, and grouping jobs by color.

[0029] CR is a powerful method for detecting random defects in logic areas of wafers. Current CR requires that all test frames in a job have substantially similar colors, which cannot always be met in reality, resulting in relatively low sensitivity and substantially high noise rates. In addition, defects on the CR base frame will introduce artifacts in the final reference image generated. The embodiments described herein have several new features that can be implemented individually or in combination (e.g., a new CR generation method plus dual detection), which can: mitigate color variations through intra-job similarity grouping and larger job sizes; further reduce noise and artifacts through dual detection; and use median-based CR for at least one of the references used for dual detection.

[0030] As used herein, the term "noise" (which is sometimes used interchangeably with "noise defect") is generally defined as an event detected on a sample, but is not a true actual defect on the sample. Noise that is not actually a defect may be detected as an event due to non-defect noise sources on the sample (e.g., particles in metal lines on the sample, signals from underlying layers or materials on the sample, line edge roughness (LER), relatively small critical dimension (CD) variations in patterned features, thickness variations, etc.) and / or due to marginalities in the inspection system itself or its configuration for inspection.

[0031] In some embodiments, the sample is a wafer. The wafer may include any wafer known in the semiconductor art. Although some embodiments may be described herein with respect to one or more wafers, the embodiments are not limited to the samples with which they may be used. For example, the embodiments described herein may be used with samples such as photomasks, flat panels, personal computer (PC) boards, and other semiconductor samples.

[0032] Figure 1 One embodiment of a system configured for detecting defects on a sample is shown in FIG. The system includes an inspection subsystem 100. Figure 1 In the embodiment shown in , the inspection subsystem is configured as a light-based inspection subsystem. However, in other embodiments described herein, the inspection subsystem is configured as an electron beam or charged particle beam-based inspection subsystem.

[0033] The inspection subsystem is configured to generate an output in response to patterned features formed in an array area on a sample. In general, the inspection subsystem described herein includes at least one energy source and a detector. The energy source is configured to generate energy directed to the sample. The detector is configured to detect energy from the sample and generate an output in response to the detected energy.

[0034] In a light-based inspection subsystem, the energy directed to the sample comprises light, and the energy detected from the sample comprises light. Figure 1 In an embodiment of the system shown in , the inspection subsystem includes an illumination subsystem configured to direct light to the sample 14. The illumination subsystem includes at least one light source. For example, as in Figure 1 , the illumination subsystem includes a light source 16. In one embodiment, the illumination subsystem is configured to direct light to the sample at one or more incident angles (which may include one or more oblique angles and / or one or more normal angles). Figure 1 As shown, light from light source 16 is directed through optical element 18 and then through lens 20 and reaches sample 14 at an oblique angle of incidence. The oblique angle of incidence may include any suitable oblique angle of incidence, which may vary depending on, for example, the characteristics of the sample and the defects to be detected on the sample.

[0035] The illumination subsystem may be configured to direct light to the sample at different incident angles at different times. For example, the inspection subsystem may be configured to alter one or more characteristics of one or more elements of the illumination subsystem so that different Figure 1 The incident angles shown direct light to the sample. > In one such example, the inspection subsystem may be configured to move light source 16, optical element 18, and lens 20 so that light is directed to the sample at different oblique angles of incidence or normal (or near normal) angles of incidence.

[0036] In some examples, the inspection subsystem may be configured to direct light to the sample at more than one incident angle at the same time. For example, the illumination subsystem may include more than one illumination channel, one of which may include Figure 1 , and another of the illumination channels (not shown) may include similar elements that may be configured differently or identically, or may include at least one light source and possibly one or more other components (such as those described further herein). If this light is directed to the sample at the same time as other light, one or more characteristics (e.g., wavelength, polarization, etc.) of the light directed to the sample at different incident angles may be different, so that light resulting from illumination of the sample at different incident angles may be distinguished from each other at the detector.

[0037] In another example, the illumination subsystem may include only one light source (e.g., Figure 1), and the light from the light source may be separated into different optical paths (e.g., based on wavelength, polarization, etc.) by one or more optical elements (not shown) of the illumination subsystem. The light in each of the different optical paths may then be directed to the sample. Multiple illumination channels may be configured to direct light to the sample at the same time or at different times (e.g., when different illumination channels are used to sequentially illuminate the sample). In another example, the same illumination channel may be configured to direct light with different characteristics to the sample at different times. For example, in some examples, the optical element 18 may be configured as a spectral filter, and the properties of the spectral filter may be changed in a variety of different ways (e.g., by swapping one spectral filter for another spectral filter) so that light of different wavelengths may be directed to the sample at different times. The illumination subsystem may have any other suitable configuration known in the art for directing light with different or the same characteristics to the sample sequentially or simultaneously at different or the same angles of incidence.

[0038] The light source 16 may include a broadband plasma (BBP) light source. In this way, the light generated by the light source and directed to the sample may include broadband light. However, the light source may include any other suitable light source, such as a laser. The laser may include any suitable laser known in the art, and may be configured to generate light at any suitable wavelength known in the art. In addition, the laser may be configured to generate monochromatic or nearly monochromatic light. In this way, the laser may be a narrow-band laser. The light source may also include a polychromatic light source that generates light at multiple discrete wavelengths or bands.

[0039] Light can be focused from optical element 18 onto sample 14 by lens 20. Figure 1 Lens 20 is shown in FIG. 2 as a single refractive optical element, but it is understood that, in reality, lens 20 may include several refractive and / or reflective optical elements that focus light from the optical element in combination to the sample. Figure 1 The illumination subsystem shown in and described herein may include any other suitable optical elements (not shown). Examples of such optical elements include, but are not limited to, polarizing components, spectral filters, spatial filters, reflective optical elements, apodizers, beam splitters, stops, and the like, which may include any such suitable optical elements known in the art. Additionally, the system may be configured to alter one or more of the elements of the illumination subsystem based on the type of illumination to be used for the inspection.

[0040] The inspection subsystem may also include a scanning subsystem configured to change the location on the sample to which light is directed and from which it is detected and may cause the light to scan the sample. For example, the inspection subsystem may include a stage 22 on which the sample 14 is disposed during inspection. The scanning subsystem may include any suitable mechanical and / or robotic assembly (including the stage 22) that may be configured to move the sample so that light can be directed to and detected from different locations on the sample. Additionally, or alternatively, the inspection subsystem may be configured so that one or more optical elements of the inspection subsystem perform some scanning of the sample with light so that light can be directed to and detected from different locations on the sample. In examples where light is scanned over the sample, the light may scan the sample in any suitable manner, such as in a serpentine path or in a spiral path.

[0041] The inspection subsystem further includes one or more detection channels. At least one of the detection channels includes a detector configured to detect light from the sample due to illumination of the sample by the subsystem and to generate an output in response to the detected light. For example, Figure 1 The inspection subsystem shown in FIG. 1 includes two detection channels, one formed by a light collector 24, an element 26, and a detector 28, and the other formed by a light collector 30, an element 32, and a detector 34. Figure 1 , the two detection channels are configured to focus and detect light at different focus angles. In some examples, the two detection channels are configured to detect scattered light, and the detection channels are configured to detect light scattered from the sample at different angles. However, two or more of the detection channels may be configured to detect the same type of light (e.g., reflected light) from the sample.

[0042] like Figure 1 As further shown in , both detection channels are shown as being located in the plane of the paper, and the illumination subsystem is also shown as being located in the plane of the paper. Thus, in this embodiment, both detection channels are located in the plane of incidence (e.g., centered about the plane of incidence). However, one or more of the detection channels may be located outside the plane of incidence. For example, the detection channel formed by the light collector 30, the element 32, and the detector 34 may be configured to collect and detect light scattered out of the plane of incidence. Thus, this detection channel may generally be referred to as a "side" channel, and this side channel may be centered on a plane substantially perpendicular to the plane of incidence.

[0043] although Figure 1An embodiment of an inspection subsystem including two detection channels is shown, but the inspection subsystem may include a different number of detection channels (e.g., only one detection channel or two or more detection channels). In one such example, the detection channel formed by the light collector 30, element 32, and detector 34 may form a side channel as described above, and the inspection subsystem may include an additional detection channel (not shown) formed as another side channel located on the opposite side of the incident plane. Thus, the inspection subsystem may include a detection channel that includes the light collector 24, element 26, and detector 28 and is centered on the incident plane and is configured to gather and detect light at a scattering angle that is at or close to normal to the sample surface. Thus, this detection channel may generally be referred to as a "top" channel, and the inspection subsystem may also include two or more side channels configured as described above. As such, the inspection subsystem may include at least three channels (ie, one top channel and two side channels), and each of the at least three channels has its own collector, each of which is configured to collect light at a different scattering angle than each of the other collectors.

[0044] As further described above, each of the detection channels included in the inspection subsystem can be configured to detect scattered light. Figure 1 The inspection subsystem shown in can be configured for dark field (DF) inspection of the sample. However, the inspection subsystem can also or alternatively include a detection channel configured for bright field (BF) inspection of the sample. In other words, the inspection subsystem can include at least one detection channel configured to detect light specularly reflected from the sample. Thus, the inspection subsystem described herein can be configured for DF only, BF only, or both DF and BF inspection. Although in Figure 1 Each of the light collectors is shown in as a single refractive optical element, but it is understood that each of the light collectors may include one or more refractive optical elements and / or one or more reflective optical elements.

[0045] One or more detection channels may include any suitable detector known in the art. For example, the detector may include a photomultiplier tube (PMT), a charge coupled device (CCD), and a time delay integration (TDI) camera. The detector may also include a non-imaging detector or an imaging detector. If the detector is a non-imaging detector, each of the detectors may be configured to detect some characteristics (such as intensity) of the scattered light, but may not be configured to detect such characteristics based on the position in the imaging plane. In this way, the output generated by each of the detectors included in each of the detection channels may be a signal or data, but not an image signal or image data. In such examples, a computer subsystem (such as a computer subsystem 36 that forms part of an inspection system having an inspection subsystem) may be configured to generate an image of the sample from the non-imaging output of the detector. However, in other examples, the detector may be configured as an imaging detector configured to generate an imaging signal or image data. Therefore, the inspection subsystem may be configured to generate an image in a number of ways.

[0046] It should be noted that this article provides Figure 1 The configuration of an inspection subsystem that may be included in an embodiment of the system described herein is generally illustrated. Obviously, the inspection subsystem configuration described herein may be altered to optimize the performance of the inspection subsystem, as is typically performed when designing a commercial inspection system. In addition, the system described herein may be implemented using an existing inspection system such as the 29xx / 39xx series of tools commercially available from KLA Corporation of Milpitas, California (e.g., by adding the functionality described herein to the existing inspection system). For some such systems, the methods described herein may be provided as optional functionality of the inspection system (e.g., in addition to other functionality of the inspection system). Alternatively, the inspection system described herein may be designed "from scratch" to provide an entirely new inspection system.

[0047] The computer subsystem 36 may be coupled to the detector of the inspection subsystem in any suitable manner (e.g., via one or more transmission media, which may include "wired" and / or "wireless" transmission media) so that the computer subsystem can receive the output generated by the detector. The computer subsystem 36 may be configured as further described herein to use the output of the detector to perform a number of functions. The computer subsystem coupled to the inspection subsystem may be further configured as described herein.

[0048] The computer subsystem coupled to the inspection subsystem (as well as other computer subsystems described herein) may also be referred to herein as a computer system. Each of the computer subsystems or systems described herein may be in various forms, including a personal computer system, an image computer, a mainframe computer system, a workstation, a network appliance, an Internet appliance, or other devices. In general, the term "computer system" may be broadly defined to encompass any device having one or more processors that execute instructions from a memory medium. The computer subsystem or system may also include any suitable processor known in the art, such as a parallel processor. In addition, as a stand-alone tool or a network-connected tool, the computer subsystem or system may include a computer platform with high-speed processing and software.

[0049] If the system includes more than one computer subsystem, the different computer subsystems may be coupled to each other so that images, data, information, instructions, etc. may be sent between the computer subsystems. For example, computer subsystem 36 may be coupled to computer system 102 (e.g., a computer subsystem 36 may be coupled to computer system 102 via any suitable transmission medium. Figure 1 The transmission medium may include any suitable wired and / or wireless transmission medium known in the art. Two or more of such computer subsystems may also be effectively coupled by sharing a computer-readable storage medium (not shown).

[0050] Although the inspection subsystem is described above as an optical or light-based inspection subsystem, in another embodiment, the inspection subsystem is configured as an electron beam-based inspection subsystem. In an electron beam-type inspection subsystem, the energy directed to the sample comprises electrons, and the energy detected from the sample comprises electrons. Figure 1a In one such embodiment shown in FIG. 1 , the inspection subsystem includes an electron column 122, and the system includes a computer subsystem 124 coupled to the inspection subsystem. The computer subsystem 124 may be configured as described above. In addition, this inspection subsystem may be configured as described above and Figure 1 The same manner as shown in is coupled to one or more other computer subsystems.

[0051] Also like Figure 1a , the electron column includes an electron beam source 126 configured to produce electrons that are focused to a sample 128 by one or more elements 130. The electron beam source may include, for example, a cathode source or emitter tip, and the one or more elements 130 may include, for example, a gun lens, an anode, a beam limiting aperture, a gate valve, a beam current selection aperture, an objective lens, and a scanning subsystem, all of which may include any such suitable elements known in the art.

[0052] Electrons returning from the sample (eg, secondary electrons) may be focused by one or more elements 132 to a detector 134. One or more elements 132 may include, for example, a scanning subsystem, which may be the same scanning subsystem included in element 130.

[0053] The electron column may include any other suitable elements known in the art. In addition, the electron column may be further configured as described in U.S. Pat. No. 8,664,594 issued to Jiang et al. on April 4, 2014, U.S. Pat. No. 8,692,204 issued to Kojima et al. on April 8, 2014, U.S. Pat. No. 8,698,093 issued to Gubbens et al. on April 15, 2014, and U.S. Pat. No. 8,716,662 issued to MacDonald et al. on May 6, 2014, which are incorporated by reference as if fully set forth herein.

[0054] Although the electron column Figure 1a , but it should be understood that the electron beam can be directed to the sample and scattered from the sample at any suitable angle. In addition, the electron beam inspection subsystem can be configured as further described herein to use multiple modes to produce output for the sample (e.g., at different illumination angles, focus angles, etc.). The multiple modes of the electron beam inspection subsystem can differ in any output generation parameters of the inspection subsystem.

[0055] The computer subsystem 124 may be coupled to the detector 134 as described above. The detector may detect electrons returning from the surface of the sample, thereby forming an electron beam image (or other output) of the sample. The electron beam image may include any suitable electron beam image. The computer subsystem 124 may be configured to perform any of the steps described herein. Figure 1a The system of inspection subsystems shown in may be further configured as described herein.

[0056] It should be noted that this article provides Figure 1aThe configuration of an electron beam inspection subsystem that may be included in the embodiments described herein is generally illustrated. As with the optical inspection subsystem described above, the electron beam inspection subsystem described herein may be modified to optimize the performance of the inspection subsystem, as is typically performed when designing a commercial inspection system. In addition, the systems described herein may be implemented using existing inspection systems, such as tools commercially available from KLA (e.g., by adding the functionality described herein to the existing inspection system). For some such systems, the methods described herein may be provided as optional functionality of the system (e.g., in addition to other functionality of the system). Alternatively, the systems described herein may be designed "from scratch" to provide an entirely new system.

[0057] Although the inspection subsystem is described above as being an optical or electron beam inspection subsystem, the inspection subsystem may be an ion beam inspection subsystem. This inspection subsystem may be configured as follows, except that the electron beam source may be replaced with any suitable ion beam source known in the art. Figure 1a Additionally, the inspection subsystem may include any other suitable ion beam tool, such as included in commercially available focused ion beam (FIB) systems, helium ion microscope (HIM) systems, and secondary ion mass spectrometry (SIMS) systems.

[0058] The inspection subsystem described herein may be configured to generate outputs, such as images of samples, in a variety of modes. In general, a "mode" is defined by the parameter values ​​of the inspection subsystem for generating outputs and / or images of the sample (or outputs for generating images of the sample). Therefore, for at least one of the parameters of the inspection subsystem, the mode may differ in value (except for the position on the sample at which the output is generated). For example, in the optical subsystem, different modes may be illuminated using light of different wavelengths. The modes may differ in illumination wavelength, as further described herein (e.g., by using different light sources, different spectral filters, etc. for different modes). In another example, different modes may use different illumination channels of the optical subsystem. For example, as described above, the optical subsystem may include more than one illumination channel. Thus, different illumination channels may be used for different modes. The modes may also or alternatively differ in one or more aggregation / detection parameters of the optical subsystem. The modes may differ in any one or more variable parameters of the inspection subsystem (e.g., illumination polarization, angle, wavelength, etc., detection polarization, angle, wavelength, etc.). The inspection subsystem may be configured to scan the sample with different modes in the same scan or in different scans, for example, depending on the ability to scan the sample simultaneously using multiple modes.

[0059] In a similar manner, the output generated by the electron beam subsystem may include output, e.g., images, generated by the electron beam subsystem at two or more different values ​​of parameters of the electron beam subsystem. Multiple modes of the electron beam subsystem may be defined by the values ​​of the parameters of the electron beam subsystem used to generate the output and / or image of the sample. Thus, the modes may differ in the value of at least one of the electron beam parameters of the electron beam subsystem. For example, different modes may use different angles of incidence for illumination.

[0060] As described above, the inspection subsystem is configured to scan a physical version of a sample with energy (e.g., light, electrons, etc.), thereby generating an output of a physical version of the sample. In this way, the inspection subsystem can be configured as a "real" subsystem rather than a "virtual" subsystem. However, Figure 1 The storage media (not shown) and computer subsystem 102 shown in the can be configured as a "virtual" system. In particular, the storage media and computer subsystem can be configured as a "virtual" inspection system as described in commonly assigned U.S. Patent No. 8,126,255 issued on February 28, 2012 to Bhaskar et al. and U.S. Patent No. 9,222,895 issued on December 29, 2015 to Duffy et al., both of which are incorporated by reference as if fully set forth herein. The embodiments described herein can be further configured as described in these patents.

[0061] When there are relatively large color variations among the input test frames, the reference (Ref) images generated by the currently used CR cannot match the test images well, which may be especially the case for edge die frames, thereby resulting in relatively low detection sensitivity and substantially high noise rates for inspection. In other words, the color variations in the image may be attributed to some non-defect noise sources on the sample, which include some of the noise sources further described herein (e.g., non-defect related thickness variations). Unless properly handled or resolved, color differences between images may be falsely detected as defects, which increases noisy detection and may reduce the achievable inspection sensitivity. Some embodiments described herein propose new job layouts to mitigate problems caused by color variations within a test frame (i.e., from test frame to test frame). One such embodiment of a system configured for detecting defects on a sample includes an inspection subsystem configured for generating an image of the sample, which may include any of the inspection subsystems described herein.

[0062] The terms "image" and "image frame" are used interchangeably herein. For example, the terms reference image and reference image frame are used interchangeably herein. In general, an "image frame" is defined as a collection of pixels in an image that are processed together for purposes such as defect detection. Therefore, the size of an image or image frame may vary depending on some characteristics of the inspection subsystem or computer subsystem. In some inspection use cases, an image may actually be composed of multiple image frames, but this is generally not the case for the embodiments described herein. In other words, for the embodiments described herein, a reference image is the same as a reference image frame, a base image is the same as a base image frame or non-test image frame, a test image is the same as a test image frame, and so on. However, where appropriate, the embodiments described herein may also be applied to images composed of multiple frames.

[0063] This system also includes a computer subsystem, which may include any of the computer subsystems described herein, configured to group images generated by the inspection subsystem for a row of dies on a sample into different jobs based on color, such that images in a first job in the different jobs have a first color value range, and images in a second job in the different jobs have a second color value range different from the first color value range. In this way, within each row of dies, frames with similar colors will be grouped into one job. Thus, the embodiments described herein may be configured for a new job layout with color similarity grouping within a row of dies. With color similarity grouping, the chances of the generated Ref matching the test well are higher, resulting in better defect detection sensitivity and noise reduction.

[0064] Figure 2 An example of a currently used CR job layout is illustrated. In this example, a job layout 202 is shown that spans a row of dies (not shown) on a wafer 200. Figure 2 As shown in FIG. 1 , a job layout may include job 1, job 2, ..., job N-1, and job N. The number of jobs in a job layout may vary depending on, for example, where the job layout is located on the wafer, how many frames are selected in each job, etc. Each of the jobs includes the same number of image frames. For example, typically, each job has about 8 frames.

[0065] As used herein, the term "job" is defined as a number of image frames that are processed together by a computer subsystem for detecting defects on a sample. Typically, the entirety of images generated during the inspection process cannot all be processed together (e.g., even if possible, it is usually not time or cost effective). Therefore, the entirety of images is separated into jobs of image frames that can be processed together for defect detection in a relatively inexpensive and timely manner.

[0066] In currently used recipes, the job layout is typically set before the sample is scanned and does not change based on any characteristics of the image itself. In other words, the job layout currently used is typically static and predetermined, and does not change on the fly or based on the image generated by the inspection subsystem. In contrast, the embodiments described herein may advantageously be configured for new job layouts with similarity groupings. Specifically, the computer subsystem may compare the color values ​​of different image frames in a row of dies and group the image frames into jobs such that image frames with similar colors are included in one job, while image frames with different colors are included in different jobs. In other words, the computer subsystem may group image frames with a first range of color values ​​into Job 1, image frames with a second range of color values ​​into Job 2, and so on.

[0067] Color "similarity" as used by the embodiments described herein may be defined in different ways depending on the sample, expected variations in color, user preferences, etc. In general, how similar the colors of two image frames must be to be grouped into one job may be predetermined in some way so that color similarity grouping can be easily performed during the inspection process.

[0068] In one example, a user may set up an inspection recipe so that in the new job layout described herein, each color value range for each job is a different single value, e.g., job 1 includes an image frame with color value 1, job 2 includes an image frame with color value 2, and so on. However, in general, having a different job for each color value may not be practical or useful. Instead, the computer subsystem may determine the color value of each image frame and the entire range of color values ​​for all image frames in a row of dies. Then, in one example, the entire range of color values ​​may be divided into a predetermined number of sub-ranges, and the image frames may be assigned to jobs associated with the sub-ranges in which their color values ​​are located. For example, a user may set up a recipe to divide all image frames in a row of dies into 6 jobs, and then the entire range of color values ​​for the image frames may be divided into 6 sub-ranges, one sub-range for each job. Then, the image frames may be assigned to jobs having a sub-range in which the color values ​​of the image frames are contained.

[0069] Other ways of defining a target number of jobs in a die row or how similar the color values ​​of image frames need to be to be included in a single job are obviously possible. For example, a user may predefine that for any image frame, any other image frames having color values ​​within ±5% of the image frame's color value are grouped together into a job. After those image frames are grouped, another image frame may be selected and grouped in the same manner based on color value percentage similarity. In this way, grouping may continue until all image frames have been analyzed and / or grouped.

[0070] The color values ​​used in the embodiments described herein may be grayscale values ​​of an image frame. Since each frame may contain different grayscale values ​​(i.e., different values ​​for different pixels in a frame), the grayscale values ​​used for similarity grouping may be determined from some or all grayscale values ​​of an image frame. For example, the color values ​​used in the embodiments described herein may be the average grayscale value, median grayscale value, grayscale distribution, some properties of the grayscale distribution, etc. of the pixels in a frame. In the embodiments described herein, when the goal is to reduce the grayscale difference between a reference image and a test image, similarity grouping may be performed based on grayscale distribution. However, when the goal of similarity grouping changes, the values ​​of the image frames used for grouping may also change. For example, similarity grouping based on geometric features may be used in the embodiments described herein to separate image frames into jobs that are processed separately for defect detection.

[0071] Unlike the job layouts currently used, the new job layouts described herein may include jobs with different numbers of image frames within them. In other words, not all jobs in the new job layout have (or need to have) the same number of image frames. For example, each input job may have a different number of frames, and the embodiments described herein support such dynamic job size scenarios. In one such example, Figure 3 As shown in the new job layout embodiment of FIG. 302, jobs 1, 2, 3, and 4 in layout 302 may all include different numbers of image frames. Although all jobs in layout 302 are shown as including different numbers of image frames (as indicated by their different relative areas on the sample), one or more of the jobs may include the same number of image frames. In other words, compared to currently used job layouts, the embodiments described herein are less concerned with the number of image frames in each job and more concerned with image frames having similar color values ​​in each job.

[0072] Furthermore, although four jobs are shown in layout 302, the number of jobs into which image frames in any die row swath are grouped may be different than four, and may include at least two jobs without any upper limit on the number of jobs. However, the number of jobs generated may be limited to some extent based on a predetermined minimum job size and the number of image frames in a die row, which predetermined minimum job size may be determined as further described herein. In general, any one image frame may be included in only one job. For example, image frames included in one job may be mutually exclusive of image frames included in all other jobs.

[0073] Another way that the job layout embodiments differ from the job layouts currently used is that all image frames included in one job do not have to be located in one continuous area on the sample. For example, in one embodiment, at least one of the first job and the second job in different jobs includes two or more of the images generated at non-adjacent areas in the row of dies on the sample. One such embodiment is shown in Figure 3 In job layout 302 shown across a row of dies (not shown) on wafer 300, job 1 includes image frames in two regions spaced apart from each other on the wafer. Jobs 2, 3, and 4 in layout 302 include images generated in regions between the regions in which the image frames included in job 1 were generated.

[0074] In another embodiment, at least one of the first job and the second job in the different jobs includes two or more of the images generated in a first region of the row of dies located closer to the edge of the sample than to the center of the sample, and at least an additional one of the first group and the second group in the different groups includes another two or more images generated in a second region of the row of dies located closer to the center of the sample than to the edge of the sample. Figure 3 As shown in job layout 302 in FIG. 3 , job 1 may include edge die frames (i.e., image frames generated closer to the edge of the sample than to the center of the sample) with similar colors, which results in those image frames being grouped into one job. Additionally, edge die frames located on both ends of a row of dies and spaced apart from each other may be grouped into one job. In contrast, jobs 2, 3, and 4 in layout 302 are spaced apart from the edge of the sample and may be located closer to the center of the sample than to the edge of the sample.

[0075] Thus, one advantage of the flexible job layout and grouping of image frames into jobs based on color as described herein is that some image frames generated on a sample have image features specific to a sample area that are not caused by defects on the sample. For example, some samples, such as wafers, may have variations in the characteristics of a repeating patterned area on the sample that are not related to defects, which may result in corresponding variations in the image, which may then be erroneously detected as defects. However, by grouping image frames into jobs as described herein based on image characteristics (such as color), image frames with similar image characteristics will be processed together for defect detection. In other words, comparing (or at least using) image frames with similar image characteristics for defect detection will reduce the number of noise detected in the image and may increase the sensitivity of the inspection.

[0076] In some embodiments, the computer subsystem is configured to group the images into different jobs based on a predetermined minimum job size, such that each of the different jobs contains a number of images greater than the predetermined minimum job size. For example, the embodiments described herein may only set a minimum group size for each job, rather than requiring each of the jobs to contain the same number of image frames as currently set. Each of the jobs may then contain at least that predetermined number of image frames, which may vary from job to job based on the similarity of the image frames and how the similarity grouping is set, as described herein.

[0077] There may be no upper limit to the number of image frames that may be included in any one job, but this may be useful, for example, when there is a practical limit on how many image frames may be processed together. For example, if the image frames in a row of dies are separated by color similarity into one edge frame job and one center frame job, then the center frame job may contain so many image frames that it may be useful or practical to separate those center frames into more than one job for ease of processing. In one such example, the center frames in a row of dies may be so similar in color that Figure 3 Jobs 2, 3, and 4 shown in layout 302 of FIG. 302 may all have similar color value ranges. The computer subsystem may initially separate those image frames into a single job (not shown) separate from the image frames in job 1 shown in layout 302. Then, those central image frames may be separated again based on how many image frames are actual for any one job. Thus, two or more of the jobs in the new layout may actually have the same or similar color value ranges.

[0078] In one such embodiment, the predetermined minimum job size is 20 images. For example, the frame count in the new job may be 20 or even 30, both of which are significantly larger than the typical image frame count (8) currently used for CR type inspection jobs. In one such example, by comparing Figure 2 The layout of the assignment 202 and Figure 3As can be seen from the job layout 302 in the new job layout, the job layout 202 includes many more jobs than the job layout 302, and each of the jobs in the job layout 302 is much larger than each of the jobs in the job layout 202. With similarity grouping and larger job frame counts, the chances of the generated Ref matching the test well are higher, resulting in better defect detection sensitivity and noise reduction. For example, one or more of the jobs in the new job layout may be relatively small, or much smaller than all or some other jobs. However, by setting a predetermined minimum job size that is much larger than the job size currently used for each of the jobs, the probability of generating a reference that substantially matches the test image for each of the test images in each of the jobs is much greater than the job size currently used. In other words, in order to increase the chance that each frame within each job has a similar matching frame for defect detection purposes, the job size is increased. Therefore, the number of noise detected in each test image with references generated from other frames in the same job as the test image will be lower, and the inspection sensitivity can be higher.

[0079] The computer subsystem is also configured to detect defects in images in a first job in different jobs by applying the defect detection method only to images in a first job in the different jobs, and to detect defects in images in a second job in the different jobs by applying the defect detection method only to images in the second job in the different jobs. In this way, the detection step is performed separately for each of the different jobs. Other steps described herein (such as calculating a reference image and generating a difference image) may also be performed separately for each of the different jobs. In this way, images in one job may be processed together for reference calculation, difference image generation, and defect detection, and processed separately from images in other jobs. However, defects detected in each job may be reported together with defects detected in other jobs.

[0080] Defects on the base frame (i.e., non-test images or images other than test images) introduce artifacts on the Ref image. Some embodiments are configured for new types of dual detection to further improve defect detection sensitivity (e.g., MCAT algorithm sensitivity) and reduce noise. For example, another advantageous embodiment is a dual detection performed within a die row using at least one CR. Therefore, the steps performed by these embodiments may be referred to as a CR dual detection process. One such embodiment of a system configured for detecting defects on a sample includes an inspection subsystem configured to generate an image of the sample, the image including a test image, two or more first images corresponding to the test image, and one or more second images corresponding to the test image. The inspection subsystem may be configured as further described herein. The first image and the second image may be any of the images in the job other than the test image in the job. As further described herein, which of the basic images in the job are selected to be used as the first image and the second image. In other words, the terms "first" and "second" images used herein are only for the convenience of referring to different basic images, but those terms are not meant to have any other meanings for the embodiments described herein.

[0081] Figure 4 An embodiment of the steps that can be performed for the CR double detection process is shown. As shown in step 400, the computer subsystem can obtain a test frame Test i . Other images described above may also be acquired in this step. The image may be responsive to patterned features formed in an area on the sample, such as a logic area. Acquiring an image may include generating an image with an inspection subsystem configured as described herein. This image acquisition may be performed when the computer subsystem and the inspection subsystem are coupled in one tool and possibly when the defect detection described herein is performed on the tool and / or when the image is being generated. In other examples, the computer subsystem may acquire an image from another method, system, or storage medium. For example, the computer subsystem and the inspection subsystem may or may not be coupled to a single tool, and the inspection subsystem, the computer subsystem, or another computer subsystem may store an image generated by the inspection subsystem. Then, the computer subsystem may acquire an image from a storage medium in which the image is stored. This image acquisition may be performed when the computer subsystem performs the steps described herein offline and / or after all (or at least some) images have been generated. Each of the elements described above may be configured as further described and shown herein.

[0082] The system also includes a computer subsystem configured to calculate a first reference image and a second reference image of a test image from two or more first images and one or more second images, respectively. Figure 4As shown in steps 402 and 404 in FIG. 1 , the computer subsystem may respectively calculate Ref1 i and Ref2 i In one embodiment, Ref1 i can be calculated using the current CR method, and Ref2 i Can be calculated as a median-based CR, which is a new type of CR that can be generated as further described herein. In most examples, two reference images can be calculated for each test image. In other words, the reference image may not be used for more than one test image in a job. Instead, after grouping the images into jobs as further described herein, the computer subsystem will create two reference images for each test frame within each job. Both reference images are based on frames within the same job as the test image. The two reference images will be used to perform double detection to further suppress noise. In addition, the reference images will be job-specific, which means that for each job, the step of calculating the reference image should be performed separately. The reference images can be calculated as further described herein.

[0083] One or more characteristics of the first reference image and the second reference image are different. For example, ideally, the two reference images are not exactly the same, because this will make the two comparisons between the two reference images and the test image further described herein redundant. Instead, at least one characteristic of the first reference image should be different from the same characteristic of the second reference image. Creating two useful but different reference images can be done in several different ways described herein. One way is to make the two different reference images different types of reference images, like the currently used CR and the median-based CR, which will cause the reference images to be at least somewhat different. In addition, the images used to generate the two different reference images may be different. For example, as further described herein, the first reference image is generated from two or more first images, and the second reference image is generated from one or more second images. In some examples, the two or more first images may be mutually exclusive with the one or more second images, which means that any of the first images used to generate the first reference image is not used to generate the second reference image (and vice versa). However, one or more images can be used to generate the first reference image and the second reference image. For example, in one embodiment, when the second reference image is generated from a single reference image frame in the job, the first reference image can be generated from all image frames in the job except the test image frame. In this way, the image frame used to generate the second reference image can also be used to generate the first reference image. In another example, when the different reference images are different types of reference images (e.g., currently used CR and median-based CR), all image frames in the job except the test image frame can be used to generate two reference images. In this way, different methods can be used to generate different reference images from the same set of images.

[0084] The computer subsystem is also configured to generate a first difference image and a second difference image by separately comparing the test image with the first reference image and the second reference image, respectively. Figure 4 As shown in steps 406 and 408 of , the computer subsystem may respectively perform the following steps by subtracting the first reference image (ie, Test i -Ref1 i ) to generate a first difference image Diff1 i , and by subtracting the second reference image (i.e., Test i -Ref2 i ) to generate a second difference image Diff2 i Subtracting the reference image from the test image may be performed in any suitable way.

[0085] The computer subsystem is further configured to detect defects in the test image by applying a defect detection method to the first difference image and the second difference image. Figure 4 , the first difference image generated in step 406 may be input to a detection step 410, and the second difference image generated in step 408 may be input to a detection step 412. In perhaps the simplest implementation, detecting defects in steps 410 and 412 may include comparing the difference image pixels or signals to a threshold value, and determining that any pixel or signal having a value above the threshold value is a defect, and determining that any pixel or signal not having a value above the threshold value is not a defect. However, many more sophisticated defect detection methods have been developed in the art and may be used in steps 410 and 412. In other words, the difference image generated by the embodiments described herein may be used for defect detection in the same manner as any other difference image is used for defect detection. In addition, the defect detection method may include any defect detection method known in the art (e.g., the MCAT defect detection method), which is a defect detection algorithm used by some inspection systems commercially available from KLA, or another suitable commercially available defect detection method and / or algorithm.

[0086] The defect detection method determines that the defect exists at the position in the test image only if the defect detection method detects the defect at the corresponding position in the first difference image and the second difference image. Figure 4 As shown in step 414 of the embodiment, the defects detected by both the test steps 410 and 412 can be combined into two comparisons: Test i -Ref1 i and Test i -Ref2 i Then, a set of combined defects can be Figure 5 The plots shown in the figure were analyzed to produce Figure 4 The final defect result 416 shown in the figure is only included in Diff1 i and Diff2 i In this way, the defect detection method is preferably configured so that a defect is reported only when a defect is detected at the same position or the same pixel in the two difference images.

[0087] Figure 5 Various images that may be produced or used by the embodiments described herein are shown, but are only single-value grayscale images except for defects and noise. Figure 5 The images shown in are not meant to represent any particular kind of images that may be used or generated by the embodiments described herein. Rather, the images are shown only as single-value grayscale images to better illustrate the arbitration that may be performed in the CR dual detection described herein.

[0088] Test image 500 includes real defect 502. As further described herein, two different reference images 504 and 506 can be generated for test image 500. Because the two reference images have at least one different characteristic, the reference images can include different noise or artifacts. For example, reference image 504 includes noise or artifact 508, while reference image 506 does not include any noise or artifacts. The reference images can be used to generate different difference images for the test image. For example, difference image 512 can be obtained from Test image 500. i -Ref1 i Step 510 generates and the difference image 516 can be obtained from Test i -Ref2 i Step 514 occurs.

[0089] Because the reference images are different, the two different difference images will also be different in important respects. For example, difference image 512 includes true defect 502 from test image 500 and noise or artifact 508 from reference image 504 because true defect 502 and noise or artifact 508 are two differences between test image 500 and reference image 504. In contrast, difference image 516 includes only true defect 502 from test image 500 because true defect 502 is the only difference between test image 500 and reference image 506. A defect detection method applied to the difference images may detect true defect 502 and artifact 508 in difference image 512 as defects, and detect true defect 502 in difference image 516 as a defect. Therefore, Figure 5 The diagram illustrates how using only one difference image for defect detection can lead to disadvantageous and false detection of noise as a real defect. In other words, if only the reference image 504 is used for defect detection, noise or artifacts in this reference image may be falsely detected as a defect.

[0090] However, arbitration performed by the computer subsystem using defects detected in both comparisons can be used to separate real defects from artifacts. Specifically, the computer subsystem can compare the defects detected in the difference images, and any commonality among the defects will be designated as a real defect, while any difference between the defects will be designated as an artifact. For example, because real defect 502 is detected as a defect in both difference images 512 and 516, the computer subsystem can designate and report this detected defect as a real defect in step 520. In other words, the defect reported in both comparisons is reported in the final inspection results. In contrast, because artifact 508 is only detected in one of the difference images, the computer subsystem designates and reports this "detected defect" as an artifact in step 518. In this way, defects detected by only one of the comparisons will not be reported in the final inspection results. The computer subsystem may compare the results of different defect detection steps performed with different difference images in any suitable manner so that the results at corresponding locations in the different difference images may be compared to thereby determine whether the detected defect is a real defect or an artifact or noise.

[0091] In this way, dual detection can be performed with at least one CR type reference image to further reduce noise and / or any artifacts. In contrast, in the currently used CR, real, relatively large defects on one or more base frames used to generate the CR will produce artifacts on the final Ref created by the base frame and compared with the test frame. In the embodiment with dual detection described in this article, two Refs (Ref1 and Ref2) are generated for each target test frame (Test), and at least one of these Refs is a CR type Ref. Defects are only reported when defects are detected in both comparisons: Test-Ref1 and Test-Ref2. Therefore, at least one CR with dual detection can eliminate a large amount of noise and / or any artifacts caused by the reference generation process. In other words, CR dual detection can minimize the impact of unique noise or artifacts in the reference image on the defect detection result through arbitration.

[0092] This system embodiment may also be configured to group images into different jobs based on color, as further described herein. For example, in one embodiment, the computer subsystem is configured to group images generated by the inspection subsystem for rows of dies on a sample into different jobs based on color, such that images in a first job of the different jobs have a first range of color values, and images in a second job of the different jobs have a second range of color values ​​that is different from the first range of color values. In one such embodiment, the calculation, generation, and detection are performed separately for each of the different jobs. These embodiments may be further configured or performed as described herein.

[0093] In another embodiment, at least one of the first and second jobs in the different jobs includes two or more of the images generated at non-adjacent regions in the row of dies on the sample. In an additional embodiment, at least one of the first and second jobs in the different jobs includes two or more of the images generated in a first region in the row of dies that is located closer to the edge of the sample than to the center of the sample, and an additional one of the first and second jobs in the different jobs includes another two or more of the images generated in a second region in the row of dies that is located closer to the center of the sample than to the edge of the sample. These embodiments may be further configured as described herein.

[0094] In yet another embodiment, the computer subsystem is configured to group the images into different jobs based on a predetermined minimum job size such that each of the different jobs contains a number of images greater than the predetermined minimum job size. In some embodiments, the predetermined minimum job size is 20 images. These embodiments may be further configured as described herein.

[0095] In another embodiment, the test image is included in a job of images generated by the inspection subsystem for the sample and processed together by the computer subsystem, and two or more of the first images and one or more second images comprise other images in the job. For example, as further described herein, a "job" of images is generally defined in the present technology as a group of images that are processed together for purposes such as defect detection. A job may be defined via similarity grouping as further described herein. Thus, a first image and a second image used to detect defects in a test image may be included in the same job via similarity grouping. In other words, Figure 4 and 5 All images used in the defect detection shown in can be in the same job. Then, defect detection can be performed separately for each of the jobs, and the defects detected in each of the jobs can be combined into one defect detection result for the sample.

[0096] In some embodiments, the two or more first images include all other images in the job, and the one or more second images include only one of the other images that is adjacent to the test image. For example, the two or more first images may be used to generate a CR type reference using the currently used CR method or the new median-based CR method described herein. In some cases, the two or more first images may include all image frames in the job except the test image (although of course fewer than all base image frames may be used).

[0097] In this embodiment, the one or more second images may include only a single image frame adjacent to the test image. For example, the second reference image may be only one adjacent frame in the job, which may or may not be modified before being used as a reference image. In one such example, a single base frame adjacent to the test image may be high pass filtered to eliminate relatively large image signals, but unless this processing includes combining the image frame with another image frame in some way, this processing is not generally considered a CR method. Other image processing may also be used to modify the adjacent base frame without actually calculating the reference, as that term is commonly used in the art.

[0098] Then, in such embodiments, a single image frame used for the second reference image can also be used to calculate the first reference image. However, the groups and numbers of image frames used to create the different reference images are different. In addition, the methods used to generate or calculate the different reference images are different. In this way, the different reference images will have a high probability of not being exactly the same. In particular, it is highly unlikely that two reference images contain the same artifact at the same location. In this way, the different reference images can be successfully used for dual detection as described herein to reduce noise and increase sensitivity.

[0099] In yet another embodiment, the two or more first images comprise a first subset of the other images in the job, and the one or more second images comprise a second subset of the other images in the job that is different from the first subset. Depending on the number of non-test images in the job and even on the kind of reference images generated from each subset, the number of images included in each subset may vary greatly. For example, to generate a median-based CR as described herein, the minimum number of image frames required may be 3, but there is no upper limit beyond a practical number from a time and cost perspective.

[0100] The first subset and the second subset may differ in several important aspects. If the first subset includes all images in the job except the test image, and the second subset includes only a single image frame adjacent to the test image frame, then one approach may be as described above. However, when both of the reference images are calculated references, both of the subsets of images may include more than one image frame. If both reference images are the same type of CR, or the currently used CR, or the new median-based CR described herein, then preferably, the first subset of images and the second subset of images are not identical. In this way, different CRs are not identical either, which is advantageous as further described herein. However, when the two reference images are different types of CRs, then the first subset of images and the second subset of images may or may not be different. For example, different types of CRs generated from the same image may be sufficiently different so that the CRs can be successfully used for dual detection, as further described herein. However, to ensure that the different CRs are sufficiently different, the images used to create each of the reference images may be different. Such image subsets may also be mutually exclusive, although this is not necessary. For example, for one test image, the remaining base image frames in the job can be evenly divided (or approximately evenly divided) into different subsets of images for separately generating different reference images.

[0101] Figure 6 The diagram illustrates one embodiment of how the currently used CR types are generated for the embodiments described herein. Figure 6 As shown in , a job 600 of image frames includes test images T1, T2, T3, ..., TN. This job and the images included therein can be configured as further described herein. In this CR generation, all image frames in the job except the test frames are selected as base frames. For example, if a CR reference is being generated for test image T4, then T4 will be considered as the base image. Specifically, if Figure 6 As shown in , if Ref14 (602) is being generated for T4, the images used to generate Ref14 include T1, T2, T3, T5, ..., TN.

[0102] In one embodiment, computing the first reference image includes a minimization operation that reduces the difference between the test image and the first reference image. For example, in the CR embodiments described herein, the final Ref may be a combination of all base frames. The combination may be optimized in such a way as to minimize the difference between Test and the generated Ref. In one such example, for a frame where a CR reference image is being generated for T4 Figure 6In the embodiment shown in , the computer subsystem can combine other tests with the minimum difference (i.e., min||Test4-Ref14||). In this way, the reference image can be a linear combination of other frames in the same job. The linear combination weighting factors of each basis frame can be optimized in this way so that the difference between the final reference image and the expected test image is minimal. This optimization process pushes the generated reference image to match the test image as closely as possible, which is how systematic noise (as well as low-frequency random noise) is reduced. In this way, the minimization operation helps to reduce systematic and random noise in the difference image Test-Ref used in the detection step.

[0103] Generating CR in this manner can be used in combination with other features described herein to further reduce noise and improve sensitivity. For example, as further described herein, base image selection can be extended beyond the normal job size. In one such instance, the base image can come from the entire die row. With the help of similarity grouping, similar frames are grouped together for calculated reference generation. In this way, system noise can be further reduced because the embodiment now has a better chance to find a better matching reference frame for each individual test frame. This better match helps to improve sensitivity. In other words, the embodiments described herein can optimize how to generate reference images in two important ways. One way is by changing the base image set. Another way is by changing the weighting factor of each individual base image. The grouping further described herein enables changing the base image set. Changing the weighting factor is an optimization mathematical process. In addition, when the base image set is large enough, more than one reference image can be generated for each test image. Then, multiple difference image values ​​can be generated for each test image pixel. In this way, detection can be based on these multiple difference images. One way to suppress noise is to report a defect only when all difference images contain a defect. This concept is basically a dual detection scheme, where two reference images are generated for each test image.

[0104] In additional embodiments, two or more first images include a first subset of in-service images of images generated by the inspection subsystem for the sample, one or more second images include a second subset of in-service images, and the first reference image and the second reference image are different types of calculated reference images. In one example, the first reference image can be a non-median CR (or a CR calculated without using a median operation), and the other reference image can be a median CR calculated as further described herein. However, depending on the use case, different reference images may include other types of CRs and do not need to include median-based CRs. For example, if an array area is being inspected using the embodiments described herein, the median center reference image may not be required as one of the two reference images for dual detection. Instead, when the array pattern repeats in both the X and Y directions (which is typically the case for SRAM patterns), embodiments may create two reference images, one from images generated only for patterns repeated in the X direction, and the other from images generated only for patterns repeated in the Y direction. Then, those reference images can be used for dual detection in corresponding test images. In the embodiments described herein, the images used to create the reference image and the type of reference image generated may in fact be situation specific.Median-based CR is an advantageous way to create a second reference image, but it is obviously not the only way to create a new reference image.

[0105] As further described herein, defects on the base frame can cause artifacts on the Ref image. Some embodiments are configured for a new type of reference generation method with or without dual detection to further improve the sensitivity of the MCAT algorithm (or other types of defect detection) and reduce the noise rate. In one such embodiment, one of the first reference image and the second reference image is a calculated reference based on the median, and the other of the first reference image and the second reference image is not a calculated reference based on the median. For example, one of the reference images can be generated by the currently used CR method or any suitable CR method known in the art, and the other reference image can be generated as a median-based CR. In this way, a median-based CR can be used for at least one of the references in the above-mentioned dual detection. For example, the currently used CR method can be used to generate Ref1. Ref2 should have the same high quality as Ref1 to maintain the true defect signal, and Ref2 should be as different from Ref1 as possible to eliminate more unique noise. Although it is possible that only one of the reference images is a median-based CR, in some cases both of the reference images are median-based CR. In that case, the base images used to generate each reference image are preferably a different set of images.

[0106] In some embodiments, at least one of the first reference image and the second reference image is a median-based calculated reference, and the computer subsystem is configured to calculate a median-based calculated reference by calculating multiple median images from different subsets of images in a job of images generated by the inspection subsystem for the sample and calculating a median-based calculated reference from the multiple median images. For example, in a median-based CR, multiple median frames can be selected as base frames to form Ref2. Each median frame is the median of a subset of test frames. The median operation helps to reduce random noise and real defect signals on the base frame. This median-based Ref2 brings a more stable defect signal to the Test-Ref2 comparison. In addition, arbitration (double detection) eliminates unique noise and artifacts in other Test-Ref1 comparisons.

[0107] Figure 7 An embodiment of how to generate a median-based CR is shown. In this embodiment, a job 700 of input frames includes test images T1, T2, T3, ..., TN. The job and the images contained therein may be configured and generated as further described herein. The median operation may be performed by a computer subsystem to generate multiple median images F1 (702), F2 (704), F3 (706), ..., FN (708). Each of the different median images may be generated from a different subset of the images in the job. For example, in Figure 7 In the example, F1 is generated from T1 to T5, F2 is generated from T4 to T8, and so on. Figure 7 Each median image is shown to be generated from 5 image frames, but any suitable number of image frames can be used to generate each median image. In addition, although it may be easiest to use the same number of image frames to generate each median image, this is not necessary. In addition, although Figure 7 Some of the images shown for generating one of the median images may also be used to generate the other of the median images (e.g., T4 is shown for calculating both F1 and F2), but the subsets of images used to generate each median image may be mutually exclusive. Then, a median-based CR may be generated from multiple median images. For example, Ref2 i (710) can be generated by calculating the median of F1, F2, F3, ..., FN.

[0108] In some such embodiments, computing a median-based calculated reference from a multi-median image includes a minimization operation that reduces the difference between the test image and the median-based calculated reference. In one such example, for a test image where the median-based CR reference image is for T i to produce Figure 7 In the embodiment shown in FIG. 1 , the computer subsystems can be combined to have the minimum difference (ie, min|Test i –Ref2i These embodiments may be further configured as described herein.

[0109] The embodiments described above do not necessarily have to be performed using the median of the test image or a median or multiple median images. For example, instead of the median, a local specific representative value like Image_k=F(image_k, image_k+1, image_k+2, ..., image_k+m) can be used to calculate the CR, where F identifies the local representativeness, such as the mean, median or any other suitable function, as further described above. In particular, the purpose of the local median image is to provide a localized representative image. Individual images are extreme cases of localized representative images.

[0110] CR has two main goals: to remove low frequency noise (color noise) and to reduce high frequencies at the Ref images based on a linear combination of the Ref candidate images. Since the Ref candidates are not orthogonal basis functions, the linear combination may actually be biased towards a few specific (worst case, only a single) Ref candidates. Since individual images may have extreme color noise with high frequency noise, a method to keep a diversity (multiple candidates) of representative color noise and to minimize individual high frequency noise is useful. This is why using multiple median images and computing the median is advantageous for the embodiments described herein (it is easiest and has a relatively low computational cost).

[0111] At the same time, the "hotter" the scan is (i.e., using a much lower threshold than normal to capture relatively and / or substantially low signal defects), the greater the chance of hitting the noise floor. Hitting the noise floor (i.e., detecting an event (e.g., a possible defect) in or near the noise floor) is extremely problematic because a lot of noise is captured as defects. To alleviate these problems, a new dual detection using CR is proposed for the embodiments described herein, because each way Ref is generated has slightly different noise characteristics. In other words, by creating two CRs or using two reference images (one CR and one non-CR) for the embodiments described herein, the two reference images have at least one different characteristic (which can be caused by a variety of ways described herein) and performing dual detection with those two reference images, the embodiments described herein can suppress more noise and detect more defects with higher sensitivity, even in extremely hot scans. For example, different noise distributions in the reference images trigger different high-frequency noise, which can be used to eliminate unnecessary high-frequency noise by retaining dual detection of only common defects between the two detections.

[0112] In one embodiment, the test image, two or more first images, and one or more second images are generated in a logical area on a sample. For example, the embodiments described herein may be particularly suitable for detecting defects in a logical area on a sample. However, the embodiments are not limited to the types of areas for which they can be used. For example, as further described above, the embodiments may also be used for defect detection in a repeating storage area, and some steps of the embodiments may be modified depending on the use case.

[0113] Another embodiment of a system configured for detecting defects on a sample may be configured for median-based CR with or without other features of the embodiments described herein. For example, one such system includes an inspection subsystem configured to generate an image of the sample, which may be configured as further described herein. This embodiment also includes a computer subsystem configured to calculate multiple median images from different subsets of images in a job of images generated by the inspection subsystem for the sample. Calculating multiple median images may be performed as further described herein. The computer subsystem is also configured to calculate a median-based calculated reference from the multiple median images, which may be performed as further described herein. In addition, the computer subsystem is configured to generate a difference image by comparing the test image in the job with the median-based calculated reference. The computer subsystem is further configured to detect defects in the test image by applying a defect detection method to the difference image. Generating a difference image and detecting defects may be performed as further described herein. This embodiment may also be further configured according to any of the embodiments described herein.

[0114] Any of the computer subsystems described herein may generate an inspection result, which may include the results of any of the steps described herein. The inspection result may include information of the detected defect, such as a defect ID, the location of a bounding box of the detected defect, etc., size, detection score, information about defect classification (such as a class label or ID, etc.), or any such suitable information known in the art. The results of the defect may be generated by the computer subsystem in any suitable manner. The results of the defect may have any suitable form or format, such as a standard file type. The computer subsystem may generate the results and store the results so that the results can be used by the computer subsystem and / or another system or method to perform one or more functions on the sample or another sample of the same type.

[0115] The computer subsystem may be configured to store information of detected defects in any suitable computer-readable storage medium. The information may be stored with any of the results described herein and may be stored in any manner known in the art. The storage medium may include any storage medium described herein or any other suitable storage medium known in the art. After the information has been stored, the information may be accessed in the storage medium and used by any of the method or system embodiments described herein, formatted for display to a user, used by another software module, method or system, and the like.

[0116] The results and information generated by performing inspections on samples can be used in a variety of ways by the embodiments and / or other systems and methods described herein. Such functions include, but are not limited to, processes that change, for example, a manufacturing process or step that has been or will be performed on the inspected sample or another sample in a feedback or feedforward manner. For example, the computer subsystem may be configured to determine one or more changes to a process that has been or will be performed on a sample inspected as described herein based on the detected defects. Changes to the process may include any suitable changes to one or more parameters of the process. The computer subsystem preferably determines those changes so that defects on other samples on which correction processing is performed can be reduced or prevented, defects on the sample can be corrected or eliminated in another process performed on the sample, defects can be compensated in another process performed on the sample, and so on. The computer subsystem can determine such changes in any suitable manner known in the art.

[0117] Then, those changes may be sent to a semiconductor fabrication system (not shown) or a storage medium (not shown) accessible to the computer subsystem and the semiconductor fabrication system. The semiconductor fabrication system may or may not be part of the system embodiments described herein. For example, the computer subsystem and / or inspection subsystem described herein may be coupled to the semiconductor fabrication system, e.g., via one or more common elements (e.g., housing, power supply, sample handling device or mechanism, etc.). The semiconductor fabrication system may include any semiconductor fabrication system known in the art, such as a lithography tool, an etching tool, a chemical mechanical polishing (CMP) tool, a deposition tool, and the like.

[0118] The embodiments described herein have several advantages over other methods and systems for detecting defects on samples. For example, the embodiments described herein can provide better sensitivity due to better color variation handling through similarity grouping and the use of larger job sizes. The embodiments described herein can also further reduce noise and artifacts through dual detection where at least one reference image is CR. At least one of the reference images can also be median-based CR, which can also provide better sensitivity and reduce noise and artifacts than currently used dual detection methods and currently used CR methods.

[0119] The advantages described above are achieved through several important new features of the embodiments described herein. For example, the embodiments described herein can group frames within a die row and use a larger job size to mitigate color variations in the test frame image. In addition, the embodiments described herein can be configured for CR dual detection to further eliminate noise and artifacts. Furthermore, the median-based CR can be used as a reference image or one of the reference images for CR dual detection.

[0120] Although some of the embodiments may be described herein for "die rows," the embodiments described herein may not be limited to processing images from only one die row on a sample at a time. For example, the input to the embodiments described herein may include image frames from different die rows (not shown) on the sample. The similarity grouping described herein for grouping similar frames within a common die row may be performed in the same manner to group frames from different die rows. In that example, the input frames may be frames from the entire sample, including frames from different die rows. Whether to input image frames from only one die row or from multiple die rows into the similarity grouping may be determined based on throughput considerations. For example, grouping frames from multiple die rows requires additional operations from the hardware system side, which has throughput impacts. However, the tradeoff for throughput impacts may be better inspection performance.

[0121] Each of the embodiments of the system described above can be combined together into a single embodiment. In other words, unless otherwise stated herein, no one system embodiment is mutually exclusive with any other system embodiment.

[0122] Another embodiment relates to a computer-implemented method for detecting defects on a sample. The method includes acquiring an image generated by an inspection subsystem for the sample, the image including a test image, two or more first images corresponding to the test image, and one or more second images corresponding to the test image (e.g., Figure 4The method also includes calculating a first reference image and a second reference image of the test image from two or more first images and one or more second images, respectively (e.g., in step 400 of FIG. 1 ). Figure 4 The method further includes generating a first difference image and a second difference image by separately comparing the test image to the first reference image and the second reference image, respectively (e.g., Figure 4 In addition, the method includes detecting defects in the test image by applying the defect detection method to the first difference image and the second difference image (such as in Figure 4 The defect detection method determines that the defect exists at the position in the test image (e.g., at steps 410 and 412 shown in FIG. 1 ) only if the defect detection method detects a defect at corresponding positions in the first difference image and the second difference image. Figure 5 The acquiring, calculating, generating, and detecting steps are performed by a computer subsystem coupled to a testing subsystem, which may be configured according to any of the embodiments described herein.

[0123] Each of the steps of the method may be performed as further described herein. The method may also include any other steps that may be performed by the inspection subsystem and / or computer subsystem described herein. In addition, the method described above may be performed by any of the system embodiments described herein.

[0124] Additional embodiments are directed to a non-transitory computer readable medium storing program instructions executable on a computer system for performing a computer-implemented method for detecting defects on a sample. One such embodiment is shown in Figure 8 In particular, Figure 8 , non-transitory computer readable medium 800 includes program instructions 802 executable on a computer system 804. The computer-implemented method may include any of the steps of any of the methods described herein.

[0125] Program instructions 802 implementing methods such as those described herein may be stored on computer readable medium 800. The computer readable medium may be a storage medium such as a magnetic or optical disk, tape, or any other suitable non-transitory computer readable medium known in the art.

[0126] Program instructions may be implemented in any of a variety of ways, including process-based techniques, component-based techniques, and / or object-oriented techniques, etc. For example, program instructions may be implemented using ActiveX controls, C++ objects, JavaBeans, Microsoft Foundation Classes ("MFC"), SSE (Streaming SIMD Extensions), or other techniques or methods, as desired.

[0127] Computer system 804 may be configured according to any of the embodiments described herein.

[0128] In view of this description, other modifications and alternative embodiments of various aspects of the present invention will be apparent to those skilled in the art. For example, systems and methods for detecting defects on samples are provided. Therefore, this description should be regarded as merely illustrative and is for the purpose of teaching those skilled in the art to implement the general manner of the present invention. It should be understood that the forms of the present invention shown and described herein should be regarded as currently preferred embodiments. As will be apparent to those skilled in the art after having benefited from this description of the present invention, elements and materials may be substituted for those illustrated and described herein, parts and processes may be reversed, and specific features of the present invention may be utilized independently. Changes may be made to the elements described herein without departing from the spirit and scope of the present invention as described in the following claims.

Claims

1. A system configured to detect defects on a sample, comprising: an inspection subsystem configured to generate an image of a sample, the image comprising a test image, two or more first images corresponding to the test image, and one or more second images corresponding to the test image; as well as A computer subsystem configured to: calculating a first reference image and a second reference image of the test image from the two or more first images and the one or more second images, respectively, wherein one or more characteristics of the first reference image and the second reference image are different; generating a first difference image and a second difference image by separately comparing the test image with the first reference image and the second reference image, respectively; and The defect in the test image is detected by applying a defect detection method to the first difference image and the second difference image, wherein the defect detection method determines that the defect exists at a position in the test image only when the defect detection method detects the defect at corresponding positions in the first difference image and the second difference image. 2 . The system of claim 1 , wherein calculating the first reference image comprises a minimization operation that reduces a difference between the test image and the first reference image.

3. A system according to claim 1, wherein the test image is included in a job of the images generated by the inspection subsystem for the sample and processed jointly by the computer subsystem, and wherein the two or more first images and the one or more second images include other images in the job.

4. The system of claim 3, wherein the two or more first images include all of the other images in the job, and wherein the one or more second images include only one of the other images that is adjacent to the test image.

5. The system of claim 3, wherein the two or more first images comprise a first subset of the other images in the job, and wherein the one or more second images comprise a second subset of the other images in the job that is different from the first subset.

6. A system according to claim 1, wherein the computer subsystem is further configured to group the images generated by the inspection subsystem for the rows of die on the sample into different jobs based on color, so that the images in a first job among the different jobs have a first color value range, and the images in a second job among the different jobs have a second color value range different from the first color value range.

7. The system of claim 6, wherein the calculating, generating, and detecting are performed separately for each of the different operations.

8. The system of claim 6, wherein at least one of the first job and the second job of the different jobs includes two or more of the images generated at non-adjacent areas in the row of dies on the sample.

9. A system according to claim 6, wherein at least one of the first job and the second job in the different jobs includes two or more of the images generated in a first area in the row of dies that is closer to an edge of the sample than to a center of the sample, and wherein at least an additional one of the first job and the second job in the different jobs includes another two or more of the images generated in a second area in the row of dies that is closer to the center of the sample than to the edge of the sample.

10. The system of claim 6, wherein the computer subsystem is further configured for grouping the images into the different jobs based on a predetermined minimum job size, such that each of the different jobs includes a number of the images that is greater than the predetermined minimum job size.

11. The system of claim 10, wherein the predetermined minimum job size is 20 images.

12. A system according to claim 1, wherein the two or more first images comprise a first subset of images in a job of the images generated by the inspection subsystem for the sample, wherein the one or more second images comprise a second subset of the images in the job, and wherein the first reference image and the second reference image are different types of calculated reference images.

13. The system of claim 12, wherein one of the first reference image and the second reference image is a median-based calculated reference, and wherein the other of the first reference image and the second reference image is not a median-based calculated reference.

14. A system according to claim 1, wherein at least one of the first reference image and the second reference image is a median-based calculated reference, and wherein the computer subsystem is further configured to calculate the median-based calculated reference by calculating multiple median images from different subsets of images in a job of the images generated by the inspection subsystem for the sample and calculating the median-based calculated reference from the multiple median images.

15. The system of claim 14, wherein computing the median-based calculated reference from the multi-median image comprises a minimization operation that reduces a difference between the test image and the median-based calculated reference.

16. The system of claim 1, wherein the test image, the two or more first images, and the one or more second images are generated in logical areas on the sample.

17. The system of claim 1, wherein the sample comprises a wafer.

18. The system of claim 1, wherein the inspection subsystem is further configured as a light-based inspection subsystem.

19. The system of claim 1, wherein the inspection subsystem is further configured as an electron beam based inspection subsystem.

20. A non-transitory computer readable medium storing program instructions executable on a computer system for performing a computer implemented method for detecting defects on a sample, wherein the computer implemented method comprises: Acquire an image generated by the inspection subsystem for the sample, the image comprising a test image, two or more first images corresponding to the test image, and one or more second images corresponding to the test image; calculating a first reference image and a second reference image of the test image from the two or more first images and the one or more second images, respectively, wherein one or more characteristics of the first reference image and the second reference image are different; generating a first difference image and a second difference image by comparing the test image with the first reference image and the second reference image respectively; as well as The defect in the test image is detected by applying a defect detection method to the first difference image and the second difference image, wherein the defect detection method determines that the defect exists at a position in the test image only when the defect detection method detects the defect at corresponding positions in the first difference image and the second difference image.

21. A computer-implemented method for detecting defects on a sample, comprising: Acquire an image generated by the inspection subsystem for the sample, the image comprising a test image, two or more first images corresponding to the test image, and one or more second images corresponding to the test image; calculating a first reference image and a second reference image of the test image from the two or more first images and the one or more second images, respectively, wherein one or more characteristics of the first reference image and the second reference image are different; generating a first difference image and a second difference image by comparing the test image with the first reference image and the second reference image respectively; as well as The defect in the test image is detected by applying a defect detection method to the first difference image and the second difference image, wherein the defect detection method determines that the defect exists at the location in the test image only if the defect detection method detects the defect at the corresponding location in the first difference image and the second difference image, and wherein the acquiring, calculating, generating and detecting are performed by a computer subsystem coupled to the inspection subsystem.

22. A system configured to detect defects on a sample, comprising: an inspection subsystem configured to generate an image of the sample; and A computer subsystem configured to: grouping the images generated by the inspection subsystem for the rows of dies on the sample into different jobs based on color, such that the images in a first job of the different jobs have a first color value range and the images in a second job of the different jobs have a second color value range different from the first color value range; detecting defects in the images in the first of the different jobs by applying a defect detection method only to the images in the first of the different jobs; and Defects in the image in the second one of the different jobs are detected by applying the defect detection method only to the image in the second one of the different jobs.

23. A system configured to detect defects on a sample, comprising: an inspection subsystem configured to generate an image of the sample; and A computer subsystem configured to: computing a multi-median image from different subsets of images in a job of the images generated by the inspection subsystem for the sample; calculating a median-based calculated reference from the multiple median images; generating a difference image by comparing a test image in the job to the median-based calculated reference; and Defects in the test image are detected by applying a defect detection method to the difference image.

Images