Defect Detection of Multi-Die Mask

Abstract

Methods and systems for detecting defects on a mask are provided. One method includes generating a database reference image for a multi-die mask by simulation and detecting a first defect on the mask by comparing the database reference image to an image of the mask generated by an imaging subsystem for a first die of the plurality of dies. The method also includes generating a die reference image for a first die of the plurality of dies by applying one or more parameters of the imaging subsystem learned by generating the database reference image to an image generated by the imaging subsystem of one or more of the plurality of dies other than the first plurality of dies. In addition, the method includes detecting a second defect on the mask by comparing the die reference image to an image of the mask generated by the imaging subsystem for the first die of the plurality of dies.

Description

Technical Field

[0001] The present invention generally relates to the field of mask inspection. More specifically, the present invention relates to methods and systems for detecting defects on a multi-die mask.

Background Art

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

[0003] Manufacturing semiconductor devices such as logic devices and memory devices typically involves processing a substrate such as a semiconductor wafer using a number of semiconductor manufacturing processes to form various features and multiple levels of the semiconductor device. For example, lithography is a semiconductor manufacturing process that involves transferring a pattern from a photomask to a resist disposed on a semiconductor wafer. Further examples of semiconductor manufacturing processes include, but are not limited to, chemical mechanical polishing (CMP), etching, deposition, and ion implantation. Multiple semiconductor devices can be manufactured and disposed on a single semiconductor wafer and then separated into individual semiconductor devices.

[0004] As the performance of 193nm immersion lithography reaches its limit, and due to the substantially high costs and yield issues associated with multi-patterning lithography processes, extreme ultraviolet (EUV) lithography is under extensive development. The main candidate for next-generation lithography (NGL) technology is to extend Moore's law and make computer chips smaller, faster, and more efficient.

[0005] Detecting lithographically critical defects on a mask is one of the most important process control steps in the semiconductor industry. Today, mask inspection can be performed using either die-to-database (DB) mode or die-to-die (DD) mode, depending on the mask layout. Especially after the 5nm technology node, EUV technology has been widely adopted. However, since the chemically amplified EUV mask inspection tool is not yet mature, it is difficult to reliably expand the resolution-limited inspection tool for EUV defect detection.

[0006] The currently available methods and systems for mask inspection have several other drawbacks. For example, the DB inspection performance is affected by the rendering quality since its reference image is rendered from the design database. In more complex optical proximity correction (OPC) designs on patterns with relatively narrow pitches on the mask, the rendering errors are more prominent, limiting the usable sensitivity of DB inspection. DD inspection detects defects by comparing the optical images between one die and another. It cannot detect systematic writer errors, and its performance is affected by the photomask quality and focus control of the inspection tool. When the mask pattern is substantially sensitive to focus, such as in the contact layer and metal layer, it is very difficult to have reliable detection results. In addition, EUV photomasks may have flare corrections that cause OPC variations on different dies. This can also limit the usable sensitivity of DD inspection.

Prior Art Documents

Patent Documents

[0007]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0008] Therefore, it would be advantageous to develop a method and / or system for detecting defects on a mask that do not have one or more of the aforementioned drawbacks.

Means for Solving the Problem

[0009] The following description of various embodiments should in no way be construed as limiting the subject matter of the appended patent claims.

[0010] One embodiment relates to a system configured to detect defects on a mask. The system includes an imaging subsystem configured to generate an image of a mask on which a plurality of dies are formed. The system also includes a computer subsystem configured to generate a database reference image of the mask by simulating the physical version of the mask from the mask design and simulating an image of the physical version of the mask generated by the imaging subsystem from the simulated physical version of the mask. The computer subsystem is also configured to detect a first defect on the mask by comparing the database reference image with the image of the mask generated by the imaging subsystem for a first die among the plurality of dies. In addition, the computer subsystem is configured to generate a die reference image for the first die among the plurality of dies by applying one or more parameters of the imaging subsystem learned by generating the database reference image to an image generated by one or more imaging subsystems among the plurality of dies other than the first die. The computer subsystem is further configured to detect a second defect on the mask by comparing the die reference image with the image of the mask generated by the imaging subsystem for the first die among the plurality of dies. The system may further be configured as described herein.

[0011] Another embodiment relates to a computer-implemented method for detecting defects on a mask. The method includes generating a database reference image, detecting a first defect, generating a die reference image, and detecting the second defect described above, and these steps are executed by a computer subsystem coupled to the imaging subsystem described above. The method may be implemented as further described herein. The method may include any other steps of any other method described herein. The method may be implemented by any of the systems described herein.

[0012] Another embodiment relates to a non-transitory computer-readable medium storing program instructions executable on a computer system for executing a computer-implemented method for detecting defects on a mask. 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 for which the program instructions are executable may include any other steps of any other method described herein.

Brief Description of the Drawings

[0013] Other objects and advantages of the present invention will become apparent from reading the following detailed description and referring to the accompanying drawings.

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

[0014] Although the present invention is susceptible to various modifications and alternative forms, specific embodiments thereof are shown by way of example in the drawings and will be described in detail herein. However, the drawings and the detailed description thereof are not intended to limit the invention to the particular forms disclosed, but on the contrary, are intended to cover all modifications, equivalents, and alternatives falling within the spirit and scope of the invention as defined by the appended claims.

[0015] The terms "design", "design data", and "design information" as used interchangeably herein generally refer to the physical design (layout) of an IC or other semiconductor device and data derived from the physical design through complex simulations or simple geometric and Boolean operations. The design can include any other design data or design data proxy described in the jointly-owned U.S. Patents 7,570,796 (issued March 9, 2010 to Kulkarni et al., inventors Zafar et al. on August 4, 2009) and 7,676,077, both of which are incorporated herein by reference as if fully set forth herein. Additionally, design data can be standard cell library data, integrated layout data, design data for one or more layers, derivatives of design data, and complete or partial chip design data. Further, the "design", "design data", and "design information" described herein refer to information and data generated by a semiconductor device designer in the design process and are thus available for use in the embodiments described herein before printing the design on any physical specimen such as a reticle and a wafer.

[0016] With reference now to the drawings, it should be noted that the figures are not drawn to scale. In particular, the scale of some of the elements in the figures is 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 shown in multiple figures that may be similarly configured are shown using the same reference numerals. Unless otherwise noted herein, any of the elements described and illustrated may include any suitable commercially available elements.

[0017] The embodiments described herein generally relate to systems and methods for detecting defects on a mask. As further described herein, the embodiments are a novel and unique method for defect detection on a photomask (also referred to herein as a "mask") with a multi-die layout, combining certain aspects of die-to-die and die-to-database inspection methods. Additionally, the embodiments described herein can advantageously be used to reliably detect lithographically critical defects on a mask with a multi-die layout.

[0018] One embodiment relates to a system configured to detect defects on a mask. An embodiment of such a system is shown in FIG. 1. The system includes an imaging subsystem configured to generate an image of a mask on which a plurality of dies are formed. The imaging subsystem described herein can generate an image by directing energy at the mask and detecting the energy from the mask. The imaging subsystem described herein may use various possible energy types to generate an image.

[0019] In one embodiment, the mask is configured for use at one or more extreme ultraviolet (EUV) wavelengths of light. Such a mask can be referred to as an EUV mask. In other words, the mask can be configured for use in a lithography process that transfers a pattern from the mask to a wafer using light at one or more EUV wavelengths (e.g., reflecting EUV light from the mask to the wafer). In this way, the actinic wavelength of the mask (i.e., the wavelength of light used to transfer a pattern from the mask to the wafer and thereby cause a photochemical reaction in one or more materials on the wafer, such as photoresist) can be different from the optical wavelength or the type of energy used for imaging the mask can be different according to the embodiments described herein. The embodiments described herein are particularly useful for defect detection on EUV masks, but the embodiments may also be useful for defect detection on other types of multi-die masks, such as masks designed for use at one or more deep ultraviolet (DUV) wavelengths of light (i.e., multi-die, optical, or DUV masks).

[0020] In one embodiment, the imaging subsystem is configured to generate an image of the mask by irradiating the mask with light having a wavelength of 193 nm. In another embodiment, the imaging subsystem is configured to generate an image of the mask by irradiating the mask with light having one or more wavelengths in the range of 193 nm to 257 nm. In a further embodiment, the imaging subsystem is configured to generate an image of the mask by irradiating the mask with light having a wavelength of 13.5 nm (or another EUV wavelength of light). In a further embodiment, the imaging subsystem is configured to generate an image of the mask by illuminating the mask with electrons. In yet another embodiment, the imaging subsystem is configured to generate an image of the mask by irradiating the mask with ions.

[0021] As shown in FIG. 1, the system can include an imaging subsystem 100 configured to generate an image of the mask using light having one or more of the wavelengths described above, and / or an imaging subsystem 122 configured to generate an image of the mask using charged particles such as electrons or ions. Thus, a system such as that shown in FIG. 1 can include only one imaging subsystem, or one or more imaging subsystems, including one or more of the imaging subsystems shown in FIG. 1 and / or the imaging subsystems shown in FIG. 2.

[0022] As further shown in FIG. 1, the imaging subsystem 100 includes a light source 102. The light source 102 can include any suitable light source known in the art, such as a laser. The light source 102 is configured to direct light toward a beam splitter 104, and the beam splitter is configured to reflect the light from the light source 102 to a refractive optical element 106. The refractive optical element 106 is configured to focus the light from the beam splitter 104 onto the mask 108. The beam splitter 104 may include any suitable beam splitter, such as a 50 / 50 beam splitter. The refractive optical element 106 may include any suitable refractive optical element, and although the refractive optical element 106 is shown in FIG. 1 as a single refractive optical element, it may be replaced with one or more refractive optical elements and / or more reflective optical elements.

[0023] Thus, the light source 102, the beam splitter 104, and the refractive optical element 106 can form an illumination channel for the imaging subsystem. The illumination channel may include one or more polarization components, diffraction optical elements (DOEs), and any other suitable elements such as one or more filters (not shown in FIG. 1) such as spectral filters. As shown in FIG. 1, the light source, the beam splitter, and the refractive optical element are configured such that the light is directed toward the mask at a vertical or substantially vertical angle of incidence. The imaging subsystem may be configured to scan the light across the mask in any suitable manner.

[0024] The light reflected from the mask 108 due to illumination can be collected by the refractive optical element 106 and directed through the beam splitter 104 towards the detector 110. Thus, the refractive optical element, the beam splitter, and the detector can form a detection channel of the imaging subsystem. The detector can include any suitable imaging detector known in the art, such as a charge-coupled device (CCD) or a time delay integration (TDI) device. This detection channel may also include one or more additional components (not shown in FIG. 1) such as one or more polarization components, one or more spatial filters, one or more spectral filters, etc. The detector 110 is configured to generate an output in response to the reflected light detected by the detector. The output may include a signal, signal data, an image, image data, and any other suitable output.

[0025] In some embodiments, the imaging subsystem may additionally or alternatively generate an image of the mask by detecting the light transmitted by the mask. For example, the light transmitted by the mask 108 due to illumination can be collected by the refractive optical element 112 and directed towards the detector 114. Thus, the refractive optical element 112 and the detector 114 can form another detection channel of the imaging subsystem. The detector 114 may include any of the above-described imaging detectors and may further be configured as described above. This detection channel may also include one or more additional components (not shown in FIG. 1) including any of the above-described ones.

[0026] The imaging subsystem may be configured to have multiple modes in any suitable manner. In some cases, the imaging subsystem can have multiple modes successively (e.g., by changing one or more parameters of the imaging lens of the imaging subsystem such as the numerical aperture (NA) between scans of the mask). Further, the imaging subsystem can scan the mask in several modes and other modes simultaneously. The system can be configured to control the optical mode used for any scan of any mask in any suitable way.

[0027] The system may include several other components not shown in FIG. 1. For example, the system can include a loading module, an alignment module, a handler such as a robotic transfer arm, and an environmental control module, and can include any such components known in the art.

[0028] As shown in FIG. 1, the system includes a computer subsystem 116 coupled to the imaging subsystem 100. For example, the computer subsystem may be coupled to detectors of the imaging subsystem (e.g., one or more transmission media shown by the dashed lines in FIG. 1, which may include any suitable transmission media known in the art), such as detectors 110 and 114. The computer subsystem can be coupled to the detectors in any suitable manner such that the computer subsystem can send the output (e.g., an image) of the mask generated by the imaging subsystem and any other information to the computer subsystem, and optionally, the computer subsystem can send instructions for the computer subsystem to perform one or more steps to the imaging subsystem.

[0029] This computer 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 take 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. Generally, the term "computer system" may be broadly defined to include any device having one or more processors that execute instructions from a memory medium. A computer subsystem or system may also include any suitable processor known in the art, such as a parallel processor. In addition, a computer subsystem or system may include a computer platform having high-speed processing and software, either as a stand-alone tool or a network tool.

[0030] If the system includes multiple computer subsystems, different computer subsystems can be coupled to each other so that images, data, information, instructions, etc. can be transmitted between the computer subsystems, as further described herein. For example, computer subsystem 116 can be coupled to computer subsystem 118 (as shown by the dashed line in FIG. 1) by any suitable transmission medium that 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 a shared computer-readable storage medium (not shown).

[0031] The system can additionally or alternatively include a charged particle beam imaging subsystem configured to generate a charged particle beam image of a mask. In one embodiment, the charged particle beam imaging subsystem is configured as an electron beam subsystem. For example, as shown in FIG. 1, the electron beam subsystem can include an electron column 122 coupled to a computer subsystem 134.

[0032] The electron column includes an electron beam source 124 configured to generate electrons that are focused onto a mask 128 by one or more elements 126. The electron beam source may include, for example, a cathode source or an emitter tip, and the one or more elements 126 may include, for example, gun lenses, anodes, beam limiting apertures, gate valves, beam current selection apertures, objective lenses, and a scanning subsystem, all of which may include any such suitable elements known in the art.

[0033] Electrons returned from the mask (e.g., secondary electrons) may be focused onto a detector 132 by one or more elements 130. The one or more elements 130 may include a scanning subsystem that can be, for example, the same scanning subsystem included in the element 126. In some embodiments, the electron column may further or alternatively be configured to detect electrons transmitted by the mask. For example, the imaging subsystem may include another detector (not shown in FIG. 1) disposed under the mask to detect electrons transmitted by the mask. In this way, the imaging subsystem can be configured to generate an image of the mask by detecting electrons reflected from the mask and / or electrons transmitted by the mask.

[0034] The electron column may include any other suitable elements known in the art. Further, the electron column is incorporated as if fully defined herein by reference, and is described in U.S. Patent No. 8,664,594 issued to Kojima et al. on April 8, 2014 (Jiang et al. on April 4, 2014), 8,692,204, 8,698,093 issued to Gubbens et al. on April 15, 2014, and 8,716,662 issued to MacDonald et al. on May 6, 2014. In FIG. 1, the electron column is shown as being configured such that electrons are directed at the mask at an oblique angle of incidence and scattered from the mask at another oblique angle, but the electron beam may be directed at the mask and scattered from the mask at any suitable angle. Further, the electron beam subsystem can be configured to use multiple modes to generate an image of the mask (e.g., having different illumination angles, collection angles, etc.). The multiple modes of the electron beam subsystem can differ in any image generation parameter of the subsystem.

[0035] In another embodiment, the charged particle beam subsystem is configured as an ion beam subsystem. For example, in the electron column shown in FIG. 1, the electron beam source can be replaced with another charged particle beam source, such as an ion beam source, and the ion beam source can include any suitable ion beam source known in the art. Additionally, the charged particle beam subsystem may have any other suitable ion beam tool configuration, such as those included in commercially available focused ion beam (FIB) systems, helium ion microscope (HIM) systems, and secondary ion mass spectrometry (SIMS) systems.

[0036] Computer subsystem 134 may be coupled to detector 132 as described above. The detector can detect charged particles returning from the surface of the mask, thereby forming a charged particle beam image of the mask. The charged particle beam image may include any suitable charged particle beam image. Computer subsystem 134 can be configured to perform any of the functions described herein using the output of the detector and / or the charged particle beam image. Computer subsystem 134 may be configured to perform any additional steps described herein. Computer subsystem 134 may also be further configured as described herein.

[0037] The charged particle beam subsystem may be configured as a scanning electron microscope (SEM), and the charged particle beam image may be referred to as an "SEM image". However, the charged particle beam subsystem can have any suitable configuration for generating charged particle beam images including SEMs and other types of electron beam tools, such as transmission electron microscopes (TEMs). Additionally, the charged particle beam subsystem included in the system can include commercially available electron beam tools such as the Mask DR-SEM E5600 series and Mask MVM-SEM (registered trademark) E3600 series tools commercially available from Advantest America, Inc., San Jose, California.

[0038] The computer subsystems included in the system may also include computer subsystems such as computer subsystem 118 that are not coupled to any of the imaging subsystems. In this way, one of the computer subsystems can be a stand-alone computer subsystem that can be coupled to other computer subsystems shown in FIG. 1. For example, the stand-alone computer subsystem may be configured to acquire the images described herein and perform other steps described herein. In one such example, computer subsystems 116 and 134 may be configured to store the images received from the coupled imaging subsystems and store the images in a storage medium 120 that may be further configured as described herein. The stand-alone computer subsystem may then acquire the images from the storage medium and perform one or more of the steps described herein using any or all of the images.

[0039] As described above, the system may include a plurality of imaging subsystems. The imaging subsystem shown in FIG. 2 may be used in combination with or instead of any of the imaging subsystems shown in FIG. 1. As shown in FIG. 2, the imaging subsystem 200 includes an illumination subsystem and a collection subsystem. The illumination subsystem includes a light source 202. The light source 202 may be a coherent light source such as a laser. The light source can be configured to emit monochromatic light having a wavelength of 248 nm, 193 nm, and / or another DUV or EUV wavelength described herein. Alternatively, the light source may be configured to emit light having a range of wavelengths and may be coupled to a spectral filter (not shown). Examples of broadband light sources include, but are not limited to, a He-Xe arc lamp that generates light in the DUV wavelength region. In this way, the light source and the filter can emit monochromatic light having a wavelength as described above. The light source can be configured to emit light continuously or in pulses at various time intervals.

[0040] The illumination subsystem may also include several optical components coupled to the light source. For example, the light from light source 202 may first pass through homogenizer 204. Homogenizer 204 may be configured to reduce the speckle of the light from the light source. The illumination subsystem may also include aperture 206. Aperture 206 may have an adjustable NA. For example, the aperture may be coupled to a control mechanism configured to mechanically change the aperture in response to a control signal received from a user or a program instruction executed on the system from a program recipe. In this way, the light may have various partial coherence factors o. For example, aperture 206 may be changed to adjust the pupil of condenser lens 208. The pupil of the condenser lens controls the NA of the system. As the pupil of the condenser lens is reduced, the coherence of the illumination increases, thereby reducing the value of δ. The value of α can be expressed as the ratio of the NA of the condenser lens to the NA of the objective lens. The exposure system can have a value of G in the range of about 0.3 to about 0.9. Accordingly, aperture 206 may be changed such that the imaging subsystem has a value of δ in the range of about 0.3 to about 0.9. The value of δ can be changed according to the features on the mask. For example, when the mask includes lines and spaces, a higher value of δ can be used than when the mask includes contact holes. The control mechanism may also be configured to change the aperture to provide annular or off-axis illumination. The aperture may also be configured to provide other types of illumination such as quadrupole or dipole illumination. The aperture may further be configured to change the shape of the light beam. For example, the aperture may be a diffractive optical element or an apodization aperture.

[0041] The illumination subsystem may also include several additional optical components (not shown). For example, the illumination subsystem may also include a telescope configured to change the beam diameter of the light. Further, the illumination subsystem can include additional lenses such as one or more relay lenses, field lenses, folding mirrors, additional apertures, and beam splitters.

[0042] The illumination subsystem may also include a condenser lens 208. The condenser lens 208 may be configured to change the diameter of the light in the object (mask) plane to approximately or greater than the field of view of the subsystem. The light exiting the condenser lens can be directed towards a beam splitter 209, which directs the light from the condenser lens towards the mask 210, thereby illuminating the mask 210 supported on the stage 212. The beam splitter 209 may include any suitable beam splitter known in the art. The stage is configured to support the mask by contacting the outer side edges of the mask. The stage 212 can be configured to move the mask so that the alignment of the mask can be changed and light can scan across the mask. Alternatively, the illumination subsystem may include a scanning element (not shown), such as an acousto-optic deflector or a mechanical scanning assembly, so that the mask can remain substantially stationary while the light is scanned across the mask. The stage 212 may also be configured to move the mask through the focus, thereby changing the focus setting of the imaging subsystem. The stage may also be coupled to an autofocus device (not shown) configured to change the position of the stage and thereby change the position of the mask to maintain the focus setting of the imaging subsystem during imaging. Alternatively, the autofocus device can be coupled to the objective lens to change the position of the objective lens and maintain the focus setting during imaging.

[0043] The imaging subsystem may also include several optical components arranged to form a collection subsystem. For example, the collection subsystem includes an objective lens 214. The light reflected by the mask is collected by the objective lens 214. The collection subsystem also includes an aperture 216 having an adjustable NA. The NA of the aperture 216 may also be selected such that the light exiting the aperture has a selected magnification. The aperture 216 is positioned between the objective lens 214 and a lens 218 that can be configured as a tube lens. The light from the lens 218 can be directed towards a beam splitter 220. The beam splitter 220 may be configured to direct the light towards three detectors 222, 224, and 226. The collection subsystem may also include several additional optical components (not shown) such as a magnifying lens. The magnifying lens can be disposed between the lens 218 and the beam splitter 220.

[0044] The detectors 222, 224, and 226 can be configured to form an image of the light reflected by the illuminated portion of the mask. Such an image may be referred to as an "aerial image". The detectors may include, for example, a CCD or TD1 camera. The detectors may also have a one-dimensional or two-dimensional array of pixels. Each of the three detectors may have a different focus setting. In this way, the three detectors can form an image of the mask at three different focus settings substantially simultaneously. For example, one detector is substantially in focus, and the other two detectors are out of focus in opposite directions relative to the in-focus state. Additionally, the imaging subsystem may include any number of such things depending on the mechanical or physical constraints of the imaging subsystem.

[0045] Alternatively, the imaging subsystem can include only one detector configured to form an image of the mask. The detector can have a focus setting that is approximately equal to the focus setting of the exposure system. By changing the focus setting of the detector after each image is formed, images of the mask at different focus settings can be formed. In such an embodiment, the beam splitter 220 need not divide the light among a plurality of detectors.

[0046] The imaging subsystem may also include several optical components configured to generate an image of the light transmitted by the mask. For example, the collection subsystem may include an objective lens 228. The light transmitted by the mask is collected by the objective lens 228 and directed towards the detector 230. Each of these components can be configured as further described above. The transmitted light detection channel formed by the objective lens 228 and the detector 230 may also include any other optical components described herein. The imaging subsystem shown in FIG. 2 may also be configured to generate an image of the mask using a plurality of modes (e.g., using two or more different values of at least one of the imaging parameters of the imaging subsystem) as further described herein.

[0047] Computer subsystem 232 may be coupled to imaging subsystem 200. For example, the computer subsystem may be coupled to detectors, such as detectors 222, 224, 226, and 230 of an imaging subsystem (e.g., by one or more transmission media indicated by the dashed lines in FIG. 2, which may include any suitable transmission media known in the art). The computer subsystem can be coupled to the detectors in any suitable manner. The computer subsystem may be coupled to the imaging subsystem in any other suitable manner such that an image and any other information regarding a mask generated by the imaging subsystem can be transmitted to the computer subsystem and, optionally, such that the computer subsystem can transmit instructions for performing one or more of the steps described herein to the imaging subsystem.

[0048] Note that FIGS. 1 and 2 are provided herein to schematically illustrate some of the configurations of an imaging subsystem that may be included in the embodiments described herein. Clearly, the configurations of the imaging subsystem described herein may be modified to optimize the performance of the system, as is typically done when designing a commercial imaging system. Further, the systems described herein may be implemented using existing imaging subsystems, such as mask inspection tools commercially available from KLA Corp., Milpitas, Calif. (e.g., by adding the functionality described herein to an existing imaging system). For some such systems, the embodiments described herein may be provided as optional functionality of the system (e.g., in addition to other functions of the system). Alternatively, the systems described herein may be designed "from scratch" to provide a completely new system.

[0049] Only one charged particle imaging subsystem is shown in the figure, but the system may include two charged particle beam subsystems (for example, one for inspection imaging and the other for defect review imaging). The two charged particle beam subsystems may have different one or more parameters such that one of the subsystems is particularly suitable for inspection imaging and the other is particularly suitable for another process such as defect review or metrology. For example, the two subsystems may have different resolution capabilities (thus, the subsystem used for inspection has a lower resolution capability than that used for defect review or metrology). However, if the system includes only one charged particle beam imaging subsystem, that subsystem can be configured to have variable parameters so that it can be used for both inspection and defect review, or for another relatively high-resolution process such as metrology. One or more parameters of the subsystem can be changed between inspection and other processes so that the subsystem can be used for both tasks. For example, the charged particle beam subsystem shown in FIG. 1 can be configured to have a higher resolution when used for defect review or metrology instead of inspection. In other words, the embodiment of the charged particle beam subsystem shown in FIG. 1 can be adjusted in several ways that will be apparent to those skilled in the art to produce subsystems with different imaging capabilities that are more or less suitable for different applications. Some general and various configurations for the charged particle beam subsystem are described. The light-based imaging subsystem included in the system may be configured in a manner similar to that described above. As described above, the imaging subsystem may be configured to direct energy (e.g., light, electrons, ions) at a physical version of the mask and / or scan the energy across the physical version of the mask, thereby generating an actual image of the physical version of the mask. Thus, the imaging subsystem may be configured as an "actual" imaging system rather than a "virtual" system. The storage medium (such as storage medium 120) and computer subsystem 118 shown in FIG. 1 can be configured as a "virtual" system.Systems and methods configured as a "virtual" system are described in U.S. Patent Nos. 8,126,255 (issued February 28, 2012, to Bhaskar et al.) and 9,222,895, both assigned to the same assignee and issued on August 29, 2015, to Duffy et al., and are incorporated herein by reference as if fully set forth herein. Embodiments described herein may be further configured as described in these patents.

[0050] A typical EUV photomask stack is shown in FIG. 3. The mask substrate (not shown) is covered with 40 pairs of molybdenum (Mo) zirconium (Zr) multilayer (ML) thin films capped with a relatively thin layer of ruthenium (Ru), shown collectively as thin film 300 in FIG. 3. A two-layer film of tantalum boron nitride (TaBN) 302 and tantalum boron oxide (TaBO) 304 serves as the absorber and is selectively etched to form the mask pattern.

[0051] FIG. 4 shows an example of a mask pattern 400 without defects, where the colored regions represent the absorber and the non-colored regions represent the ML. FIG. 5 shows some examples of different types of defects that may be present in the mask pattern and may be detected and classified as further described herein. If there are no significant defects and only local line edge roughness (LER) contributes to the inspection signal or image detected, the detected defect is called LER as shown by defect example 500. When a defect such as a protrusion or intrusion is connected to the absorber, the detected defect is called a hard defect as shown by defect example 502. When the absorber has a pinhole and the bottom ML is exposed by the pinhole, the detected defect is a pinhole defect as shown in defect example 504. When part or all of a particle is located on the ML, the particle or particle portion has an impact on printability during wafer exposure and is classified as particle-on-ML as shown by defect example 506. When the particle is completely on the absorber, the particle has no impact on printability during wafer exposure and may fall into a different bin, which is particle-absorber as defect example 508.

[0052] The embodiments described herein have a number of novel and advantageous components aimed at solving the drawbacks of currently used die-to-database (DB) or die-to-die (DD) inspection methods and systems. One of these components is a DB detection flow that obtains corresponding images from a first die of a DB design and a multi-die mask. For example, as shown in FIG. 6, mask 600 may have a plurality of dies formed thereon. In this example, mask 600 is shown with a first die 602, a second die 604, ..., an Mth die 606. Thus, this mask example includes at least three dies, but the embodiments described herein can be used for masks having any number of dies (i.e., two or more dies) formed thereon in any arrangement on the mask. Further, the dies on the mask can include any type of die for any type of semiconductor device that can print features on a wafer or other sample using the mask. The DB detection flow may include obtaining a database (DB) and an optical (or other) image from the first die, as shown in step 608. The design can be obtained by a computer subsystem in any suitable manner known in the art. The image of the first die may also be obtained using one of the imaging subsystems described herein or from a storage medium in which the image is stored by a computer subsystem coupled to one of the imaging subsystems.

[0053] The computer subsystem is configured to generate a database (DB) reference image of the mask by simulating the physical version of the mask from the mask design and then simulating an image of the physical version of the mask generated by the imaging subsystem from the simulated physical version of the mask. The DB detection flow can perform these simulations in the DB rendering step 610. In this way, the computer subsystem can simulate a physical mask having the design formed thereon and then generate a DB reference image based on how the mask appears in the image generated by the imaging subsystem.

[0054] In one embodiment, simulating the physical version of the mask includes near-field modeling, and simulating an image of the physical version of the mask includes partially coherent illumination modeling. In this way, the DB reference image can be rendered from a DB design that combines near-field modeling and partially coherent illumination modeling for capturing mask pattern information. These modelings can be performed as described in U.S. Patent No. 9,478,019 (Sezginer et al., October 25, 2016), which is incorporated herein by reference as if fully set forth herein. The embodiments described herein may further be configured as described in this patent. These modeling steps may additionally or alternatively be performed in any other suitable manner known in the art.

[0055] In another embodiment, simulating an image of a physical version of a mask involves calibrating one or more parameters of the imaging subsystem in a model used to simulate the image of the physical version of the mask for focus error and aberration error of the imaging subsystem. For example, partially coherent illumination modeling can calibrate system-specific parameters to compensate for focus error and aberration error. In one such example, an image from a first die on the mask (or one of the dies on the mask) can be input into the DB rendering step. The design of the mask may be input into the simulation as described above, and a simulated image that simulates how the mask is imaged by the imaging subsystem may be compared to the image generated by the imaging subsystem for the first die (or one of the dies). One or more parameters of the model of the imaging subsystem may then be changed until the simulated image substantially matches the actual image generated by the imaging subsystem, which may be done in any suitable manner known in the art. In this way, the model can be calibrated for the specific parameters of the imaging subsystem (including any aberrations and all aberrations) used to generate an image of a die on the mask.

[0056] This step may be performed using the actual images of dies other than the die being used to detect systematic mask writer errors on the mask via the DB detection flow. Additionally, an image of another mask may be used to calibrate the parameters of the imaging subsystem within the model. If so, for calibration purposes, the input to the model will be the design of the other mask and any other relevant information regarding that mask. In this way, the model can be calibrated for focus error and aberration error in the imaging subsystem using a different mask (other than the mask on which the defect is detected). If different masks are used for the calibration parameters of the model, the actual images of the other masks used for calibration are preferably generated using the same parameters of the imaging subsystem as are used for the mask being inspected. In some such cases, the images of the other masks used for calibration may be generated specifically for calibration if the other masks have not yet been imaged using the parameters that would be used to inspect the mask. The other mask used for calibration may be any other available mask, but if multiple possible masks are available for this use, it may be preferable to use the mask with the design most similar to the mask being inspected, even if the only similarity between the two masks is that they are designed for use at the same wavelength.

[0057] The computer subsystem is also configured to detect a first defect on the mask by comparing the DB reference image with an image of the mask generated by the imaging subsystem for the first die of the plurality of dies. For example, the image generated by the imaging subsystem for the first die may be aligned with the DB and compared with the DB reference image. Aligning the first die image with the DB may be performed in any suitable manner, and comparing the first die image with the DB reference image may include subtracting the DB reference die image from the first die image, thereby generating a difference image. Any prominent differences are flagged as potential defect candidates. In one such example, any differences between the image of the first die and the DB reference image may be compared to a difference threshold. Differences having values above the threshold may be identified by the computer subsystem as defects, potential defects, or defect candidates. Differences not having values above the threshold may not be identified as defects, potential defects, or defect candidates. Of course, this is particularly useful and perhaps the simplest defect detection method that can be used in this defect detection step and other defect detection steps described herein, but the embodiments described herein are not limited to defect detection methods that can be used in any of the defect detection steps described herein.

[0058] The step of detecting the first defect on the mask can include the step of comparing the DB reference image with only the image of the first die. However, DB detection may include comparing the DB reference image with all the images of the dies. For example, DB detection is used herein to detect defects that repeatedly occur in multiple dies, and thus, due to the very high die-to-die nature of such detection, it cannot be detected by DD inspection, which is always the case for defects that repeat in every die and sometimes for defects that repeat in fewer dies than all dies. Thus, in many DB inspections, it can include comparing the DB reference image with only one of the multiple dies, and the "first die" can include the die on the mask that is scanned first. However, the "first die" may include any other die on the mask. In other words, generally, the DB inspection can be performed using the DB reference image and the image generated for any one die on the mask, and the user or computer subsystem can select any of the dies for such defect detection.

[0059] In one embodiment, the first defect includes a systematic mask writer error. For example, the purpose of the DB detection flow is to detect systematic writer errors. The term "systematic writer error" as used herein is defined as an error on the mask that systematically occurs (in each die or in two or more of the dies) due to the mask writer itself. As shown in step 612, the computer subsystem can perform systematic error detection. Since the rendered mask image is used to detect systematic writer errors (not necessarily for the other defect detection steps described herein), the rendering error budget is more relaxed and does not limit the available sensitivity. Thus, one advantage of the embodiments described herein is to provide the ability to detect systematic writer errors.

[0060] Another major component of the embodiments described herein that provides advantages over currently used mask inspection methods and systems is the main detection flow that combines aspects of both the DD and DB detection flows. One advantage of the embodiments described herein is to use a mask inspection tool to combine aspects of the DD and DB detection flows for a multi-die layout mask with substantially high usable sensitivity. Thus, the embodiments described herein provide a new type of inspection mode that combines aspects of both the DD detection flow and the DB detection flow.

[0061] The computer subsystem is further configured to generate a die reference image for a first die of the plurality of dies by applying one or more parameters of the imaging subsystem learned by generating a DB reference image to an image generated by one or more imaging subsystems of the plurality of dies other than the first die of the plurality of dies. This die reference image is also referred to herein as the "m-die reference image". Prior to generating the m-die reference image, images of all (or at least some) of the dies may be acquired and aligned to a DB design or some other common reference. For example, as shown in FIG. 6, optical (or other) images from all dies are acquired (step 614). Acquiring images from all dies may be done as further described herein. Aligning the images to a DB design or some other common criterion can be performed by any suitable method known in the art.

[0062] For the image from the first die, the m-die reference image is rendered using all (or at least some) of the images of the remaining dies and the system-specific parameters from the DB detection flow. When generating an m-die reference image for one of the dies, it may be preferable to use all of the images of the other dies on the mask. However, this is not necessary in all cases. For example, if there are more than three dies on the mask, when generating an m-die reference image for one of the dies, images for two or more (but not necessarily all) of the other dies may be used. Generally, the more die images used to generate the m-die reference image, the better the m-die reference image is for defect detection (using more die images for m-die reference image generation will better average out the die-to-die differences, thereby generating a more "defect-free" die reference image). However, depending on the mask configuration / layout and practical considerations, the m-die reference image can be generated from as few die images or as few die images as the user prefers.

[0063] As shown in step 616, die images from the die (all or at least some) can be aligned to the DB (design information). This alignment step can be performed in any suitable manner known in the art. It may be preferable to align each of the die images to the design, but the alignment can be performed with respect to any other common reference. As shown in step 618, m-die reference image generation can be performed using all or at least some of the die images that are aligned to the design or another common reference and thus to each other.

[0064] Applying one or more parameters of an imaging subsystem learned by generating a DB reference image to an image generated by an imaging subsystem of one or more dies among a plurality of dies other than the first die among the plurality of dies essentially overcomes the focus compensation limitation in conventional DD inspection by utilizing DB rendering in DD inspection. For example, in conventional DD inspection, in addition to pattern line edge roughness (LER) noise that can be reduced by m-die averaging, the focal distance between the reference die and the test die is a major contributing factor in its performance. In the case of an optical mask having both transmitted light (TL) and reflection (RL), physics-based methods can improve the alignment between the reference die and the test die. However, in total reflection systems such as some of the EUV mask inspection platforms described herein, conventional DD inspection has been plagued by the limitations of empirical models in focus compensation. The use of DB rendering for the purpose of focus compensation in the new DD approach can improve the focus alignment between the images for the reference die and the test die while overcoming common DB rendering noise.

[0065] In one embodiment, one or more parameters of the imaging subsystem learned by generating a DB reference image include a base image representing the focus error and aberration error of the imaging subsystem, and generating a die reference image includes synthesizing the die reference image as a linear combination of an image generated by one or more imaging subsystems of a plurality of dies other than the first die among the plurality of dies and the base image. Theoretically, the m-die reference image for each die can be synthesized as a linear combination of an image from other dies and a base image representing focus and aberration from the DB detection flow. The base image is from a DB rendering from the design data. The base image includes tool-specific information such as focus and aberration so that it can be used to overcome the focus error in the m-die reference image. After solving the linear optimization equation, it is expected that in the synthesized m-die reference image, the error between different dies on the mask is averaged and the focus-related error is compensated. Thus, the embodiments described herein can use DB modeling information to handle tool-related errors in traditional DD detection, can use the m-die reference image to handle plate errors and avoid modeling noise. In this way, the embodiments described herein can maximize the usable sensitivity by reducing rendering errors, photomask defects, and tool focus variations. Thus, one advantage of the embodiments described herein is to generate a substantially high-quality m-die reference image by utilizing multi-die averaging and DB focus compensation.

[0066] The computer subsystem is also configured to detect a second defect on the mask by comparing the die reference image with an image of the mask generated by the imaging subsystem for a first die among the plurality of dies. For example, as shown in step 620, the computer subsystem can perform defect detection using the m-die reference. This step may be performed for every die, but is not necessarily required, and the second defect detection may be performed for any or all of the dies on the mask. The detection of the second defect is otherwise performed as further described herein. In one embodiment, the computer subsystem applies one or more parameters of the imaging subsystem learned by generating a DB reference image to an image generated by one or more imaging subsystems among the plurality of dies other than a second die among the plurality of dies to generate an additional die reference image for the second die among the plurality of dies. Detecting a second defect on the mask includes comparing the additional die reference image with an image of the mask generated by the imaging subsystem for the second die among the plurality of dies. For example, for every image from each die, the m-die reference image can be rendered using, respectively, the images of the remaining dies and all of the system-specific parameters from the DB detection flow. In this way, each die may have its own m-die reference image, and DD detection may be performed for each die, thereby ensuring detection of the actual defect at a substantially accurate location.

[0067] In another embodiment, the computer subsystem determines the characteristics of one or more features of the mask within the inspection region of an image of a first die out of a plurality of dies on the mask, and based on the determined characteristics, selects a DB reference image or a die reference image to detect a second defect within the inspection region. In this way, the image of any die can be compared with its corresponding m-die reference image and the DB-rendered reference image from the DB detection flow. When the inspection region mainly includes relatively complex 2D patterns with generally low DB-rendered reference image quality, the m-die reference image is used. In contrast, when the inspection region mainly includes relatively dense repeating patterns with generally high DB-rendered reference image quality, the DB-rendered reference image is used as the final criterion for checking potential defects. In this smart reference image selection strategy, the system always selects defects with a better signal-to-noise ratio (SNR), and the available sensitivity is maximized. Thus, one advantage of the embodiments described herein is to provide a reference image selection strategy for maximizing defect detection with a better SNR.

[0068] In some embodiments, the computer subsystem obtains information about one or more patterns closest to the first and second defects on the mask, and is configured to classify the first and second defects respectively based on the information about one or more patterns closest to the first and second defects on the mask. All defect candidates, including defect candidates from the DB detection flow, can be further classified based on the surrounding pattern shape. For example, as shown in step 626 of FIG. 6, the computer subsystem can perform defect classification on defects detected by both the DB detection flow and the DD detection flow. A DB design with a substantially high resolution may be used in this step so that the pattern geometry is recognized substantially accurately based on relatively advanced anatomical rules.

[0069] A computer subsystem can determine in several different ways which pattern on the mask the detected defect is closest to and how close the detected defect is to those patterns. For example, by aligning the images used in the DB and DD inspection flows with the design, the defect location within the image can be determined relative to the design (e.g., by a simple coordinate transformation from image coordinates to design coordinates). Next, using the design coordinates of the defect, the spatial information of the detected defect relative to the design pattern on the mask can be identified using the design coordinates of the design pattern.

[0070] Determining information about the pattern closest to the defect can include identifying which pattern the detected defect overlaps with, which can be the case if the defect is located within one or more patterns, or at least partially overlaps with one or more patterns, or if the defect is in which pattern. Determining information about the pattern closest to the defect also includes determining the position of the detected defect relative to the pattern(s) (regardless of whether the detected defect overlaps with the patterned feature itself or is within the patterned feature itself). For example, if the detected defect spatially coincides (at least partially) with a patterned feature, determining information about the pattern closest to the defect can include determining the position of the detected defect relative to the outer perimeter or outer boundary of the patterned feature. If the detected defect does not spatially coincide with the patterned feature, determining information about the pattern closest to the defect can include identifying the patterned feature closest to the defect and then determining how close the defect is to the patterned feature. Determining information about the pattern closest to the defect also includes determining how close the detected defect is (in some length dimensions) to a particular part of the patterned feature (e.g., a corner, side, end, etc.). The space between the detected defect and the closest patterned feature can be represented as a single value, a range of values, an average, a function, or in any other suitable manner.

[0071] In addition, the user can pay particular attention to how close the detected defect is to a subset of the patterned features of the mask. For example, determining information about the pattern closest to the defect can be performed such that the detected defect position relative to the closest patterned feature is determined, and / or such that the detected defect position relative to any one type of patterned feature is determined. The relative position of the detected defect with respect to patterned features that are unlikely to be the closest can be determined in other ways as described herein.

[0072] In a further embodiment, the computer subsystem is configured to determine classification categories for first and second defects and assign scores to the first and second defects based on the determined classification categories, the scores indicating the severity of the first and second defects with respect to a lithography process performed using the mask. Thus, defect candidates having different classification categories can be assigned different scores indicating their severity so that the user can focus on lithographically important defects. In some embodiments, the system described herein can provide a user interface configured such that the user can specify the severity of different defect types.

[0073] The determination of the classification category of the detected defect can be performed in any suitable manner. For example, a computer subsystem can determine one or more characteristics of the detected defect based on an image of the defect (e.g., a test image) generated by an imaging subsystem, a difference image of the defect generated by either or both of the detection flows, a reference image used to detect the defect, or some combination thereof. These characteristics may include, for example, size, shape, orientation, location, location relative to any neighboring patterned features, texture, etc. Any determined features and optionally any of the above-described images may be input to a defect classifier by the computer subsystem. The defect classifier can be configured to determine the classification category (class, type, bin, etc.) of the detected defect based on one or more of their determined characteristics and / or images. The defect classifier can be any suitable defect classification method or algorithm known in the art. An example of such a defect classifier is a relatively simple decision tree in which different types of defects are separated by applying different cut lines to defect features that separate different types of defects from each other. Other examples of suitable defect classifiers are machine learning-based defect classifiers, some examples of which are incorporated by reference as if fully set forth herein and are described in U.S. Patent Application Publication No. 2018 / 0107928 (April 19, 2018, Zhang et al.) and 2019 / 0073568, published on March 7, 2019, by He et al. The embodiments described herein can be further configured as described in these publications.

[0074] Determining the classification category of a detected defect further or alternatively includes determining the printability of a detected defect on a wafer (or other sample) when the wafer is printed with a mask in a lithography process. The determination of printability preferably involves simulating the wafer printing (lithography) process performed using the mask, thereby predicting how the defect will affect the pattern printed on the wafer. For example, a computer subsystem can determine whether each detected defect is critical and causes a relatively large error in wafer printing or causes a new nuisance and a hardly observable error in wafer printing. Instead of performing this manually and using human judgment, an automated and substantially accurate method of modeling the EUV lithography process is provided by the embodiments described herein, thereby enabling a substantially accurate prediction of the printability of mask absorber defects starting from the detected defect image.

[0075] A computer subsystem can input one or more images generated for a detected defect into a model of a lithography process, thereby generating a simulated wafer image indicating how the defect affects one or more patterns printed on a wafer in the lithography process. Determining printability predicts the impact of defect printing by wafer exposure condition simulation. This simulation can include mask proximity simulations, optics, and resist exposure and development models that can use any of a compact approximation model such as Kirchoff approximation, a rigorous finite difference time domain (FDTD) solver, rigorous coupled wave analysis (RCWA), or a defect printability simulation (DPS) mask model (an example of which is described in "Compensation Method Using a New Model for Embedded Defects in Extreme Ultraviolet Lithography Masks" by Clifford et al., Photomask Technology 2010, Proceedings of SPIE Vol. 7823, 78230V, September 24, 2010, 10 pages, which is incorporated by reference as if fully set forth herein), and any other suitable such model known in the art. If the wafer process has not yet been fixed or the mask proximity model has not been well determined, printability can be directly checked on the mask binary layout without wafer exposure condition simulation for a rough estimate. Embodiments described herein may be configured as described in the above-mentioned papers. Determining printability may also be performed as described in U.S. Pat. Nos. 9,002,497 (issued Sep. 25, 2016 to Sezginer et al., filed Apr. 7, 2015, Marella et al.) and 9,478,019, which are incorporated by reference as if fully set forth herein. Embodiments described herein may be further configured as described in these patents.

[0076] In another embodiment, determining the classification category of the detected defect includes determining the printability of the detected defect on the wafer when the wafer is printed with a mask in the lithography process, and determining the printability includes simulating a spatial image for the detected defect and for the defect image generated for the corresponding reference image (either or both of the DB and m-die reference images). Then, based on the simulated aerial image, determine how the detected defect affects one or more patterns printed on the wafer in the lithography process. For example, the mask model can be applied to both the test image and the reference image of the defect with the relevant scanner exposure conditions to generate test and reference (EUV or other) aerial images. The mask model can be applied to the image as further described herein. In this way, the computer subsystem can perform a printability simulation using the scanner optical conditions. Then, a spatial image analyzer (AIA) can be executed to calculate the printability of all features within the field of view (FOV) of the imaging subsystem where the defect is located. For example, the test aerial image and the reference aerial image can be input to the AIA, and the AIA can generate a simulated image showing how the defect moves the features of the mask within the aerial image projected onto the wafer during the lithography process. An example of an AIA that can be used in the embodiments described herein is described in "Printability and actinic AIMS review of programmed mask blank defects" (Verduijn et al., SPIE Proceedings Vol. 10143, Extreme Ultraviolet (EUV) Lithography VIII, 101430K, March 24, 2017, page 13), which is incorporated herein by reference as if fully set forth herein. The embodiments described herein can be further configured as described in this publication.

[0077] The embodiments described herein can also be configured to determine the lithographic significance of defects detected on a mask, as described in U.S. Patent No. 7,873,204 (Wihl et al., January 18, 2011), which is incorporated herein by reference as if fully set forth herein. The embodiments described herein may further be configured as described in this patent.

[0078] In another embodiment, the characteristics of optical proximity correction (OPC) features in two or more of a plurality of dies are designed to be different for flare correction in a lithography process performed using a mask, and generating a DB reference image includes generating two or more different DB reference images of two or more of the plurality of dies on the mask, using the designs of two or more of the plurality of dies respectively. In this additional major new component, a dedicated but optional procedure is provided when flare correction causes significant variations in the OPC patterns on different dies. Theoretically, if the OPC has significant differences on multiple dies, this is not a typical DD-able inspection area. However, the embodiments described herein provide the ability to handle flare correction. The DB reference image can be generated for each die on the mask, as further described herein.

[0079] In an additional embodiment, the characteristics of OPC features in two or more of a plurality of dies are designed to be different for flare correction in a lithography process performed using a mask, and the computer subsystem is configured to determine the sensitivity at which a second defect is detected in two or more of the plurality of dies based on the positions of the OPC features in two or more of the plurality of dies. By comparing the DBs from all dies, the computer subsystem can easily detect the flare correction position, which can be treated as a special defect classification category for available sensitivity control. Thus, one advantage of the embodiments is to provide the ability to handle OPC variations caused by flare correction.

[0080] One of the patterning distortions that can occur in EUV lithography is flare caused by scattered light in the lithography optical system, which can cause variations in the dimensions of patterns on the wafer from the designed dimensions. One way flare is currently being addressed is through flare compensation in the OPC function. For example, flare can be modeled by inputting information about the lithography process and mask into an appropriate model for the process and mask. The simulated flare information can then be used to modify one or more characteristics of the OPC features on the mask with the desire to reduce (mitigate, or even eliminate) that flare. Thus, if different dies contain the same OPC feature, they can be modified in different ways to compensate for flare. For example, an OPC feature within one die may not need to be modified for flare, while the same OPC feature within another die may need to be adjusted (e.g., in size or shape) to reduce light scattering from the patterned feature to which the OPC feature is applied. Thus, when the images of that OPC feature are compared in DD detection, the DD detection may detect differences in the images of the OPC features in different dies and flag that difference as a defect or potential defect, even though the difference is caused by an intentional and designed difference between the OPC features in different dies. Such non-defect detections are clearly problematic for several reasons, and addressing such detections can cause other problems (e.g., when reducing sensitivity and causing the loss of other defects on the mask by avoiding such non-defect detections). However, the embodiments described herein can effectively address such differences in designs for different dies without any such adverse effects.

[0081] In the flare correction flow, the computer subsystem can obtain the DB (design) from all dies, as shown in step 622. In other words, this step may include obtaining the DB (design) for each die. Ideally, the DB designs for all dies are the same, and thus for efficiency, only the DB of one die is obtained and can be used for all dies. However, in some cases, slightly different DB designs may be required for different dies, which, as further explained above, can cause the new nuisance defect problem for the conventional DD inspection. In other words, the slight differences between the designs for different dies can cause corresponding slight differences between the dies formed on the mask. Therefore, these differences in the design, although intentional, can accidentally remove defects.

[0082] Next, as shown in step 624, the designs of different dies can be compared with each other for flare detection. In this step, the differences between the designs of different dies (and especially the designs of OPC features) can indicate differences intentionally made in the design to address the flare problem of the mask in the lithography process. Therefore, the differences between the images of the dies due to the design differences for intentional flare correction may not be defects even if detected by the DD defect detection process. Therefore, one way to prevent these design differences from being detected as defects is to identify the positions of the design differences as described above and then control the defect detection process so that these differences are not detected as defects, or to remove the defects detected at these positions from the final defect report. For example, as shown in FIG. 6, the results of flare detection 624 are input into the final defect report 628 to remove any defects detected at intentionally different positions in dies with different designs, or to filter out any defects detected at such positions where one or more characteristics (e.g., size) are below a predetermined limit. In this way, the computer subsystem can change the sensitivity of the inspection by filtering the defects detected by DD detection based on the identified flare correction positions.

[0083] In some embodiments, the computer subsystem is configured to generate inspection results for a mask that includes information about the first and second defects. For example, as shown in FIG. 6, both the defect detected by the systematic error detection step 612 in the DB detection flow and the defect detected by the defect detection step 620 in the DD detection flow can be input into the defect classification step 626, and then the results of all three steps (possibly corrected by the flare detection flow) can be used to generate the final defect report 628. In this way, the final defect report can include information about both the systematic mask writer errors detected by the DB detection flow and the defects detected by the DD detection. The final defect report can include any of the information generated by any of the steps described herein (e.g., defect ID, defect location, defect classification, etc.). The final defect report may have any suitable file or format known in the art and may be output by the computer subsystem in any suitable manner. In one embodiment, the image of the mask that is generated by imaging and used for the detection of the first defect, the generation of the die reference image, and the detection of the second defect is selected only from the images of the die equivalent regions of a plurality of dies, and the die equivalent regions are smaller than a plurality of dies. In such an embodiment, the die equivalent region has a one-dimensional size equal to the width of the swath scanned on the mask by the imaging subsystem. In another such embodiment, the plurality of dies includes dies formed in a two-dimensional array on the mask.

[0084] In the above-described embodiments, the inspection swath map may be set to scan the die equivalent regions in order. In this way, the equivalent swaths are scanned for each separate row of the die. For example, as shown in FIG. 7, there are three rows of die 702: row A, row B, and row C are formed on mask 700. Each row of the die contains four die, and the rows and columns of die on the mask form a two-dimensional array of die on the mask. Although FIG. 7 shows a 3-row by 4-column die, the masks described herein may include any suitable number of die arranged in both rows and columns, and any suitable number of rows and columns of die. A plurality of swaths, for example, swaths 1 through N, may be scanned in order from row A to row C, or vice versa, across each row of the die. The swath 704 shown is the Nth swath (swath N) in each of these rows of the die. The patch 706 shown is a die equivalent region that contains exactly the same pattern within each die. As shown in FIG. 7, each die equivalent region indicated by patch 706 is smaller than each of the die. Additionally, as shown in FIG. 7, each of the die equivalent regions indicated by patch 706 has a size in one dimension equal to the width of swath N.

[0085] This concept of die equivalent regions can be advantageous for certain steps described herein, and the die equivalent regions can be used in place of the die shown in FIG. 6. In other words, the steps described herein that are performed with respect to the die can be performed in the same manner with respect to the die equivalent regions. In this way, in any step described herein as being performed on the die, the same functionality can be performed on the die equivalent regions. Thus, one advantage of the embodiments described herein is that a die equivalent region setup can be used to cover all kinds of multi-die layout masks. Additionally, one of the new features of the embodiments described herein is that it can handle all kinds of multi-die layout masks. All of the embodiments described herein may include storing the results of further steps of the embodiments on a computer-readable storage medium. The results may include 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 results are stored, the results can be accessed within the storage medium and formatted for display to a user, used by another software module, method, or system used by any of the methods or system embodiments described herein, etc.

[0086] The computer subsystem can be configured to place detected defects based on any of the information generated by the embodiments described herein. As used herein, the term "placement" is defined as determining additional information about the detected defect that can be used to make a final determination as to how to appropriately handle the detected defect, e.g., whether the detected defect should be repaired or whether the detected defect is a nuisance and can be ignored. There may be no need to repair the detected defect, but the wafer printed with the mask should be monitored for the effects of the detected defect on the wafer pattern formed by the mask, etc.

[0087] The computer subsystem can send information regarding the detected defects to a mask review tool. For example, mask secondary charged particle beam (e-beam or ion beam) imaging provides a higher resolution than inspections that enable much better review of detected defects on the mask. Thus, the charged particle beam image can be used to more accurately (with higher resolution) determine information about the detected defects than the inspection image of the detected defects, thereby enabling additional information to be determined from the charged particle beam image that can be used to more effectively make (placement) decisions regarding the detected defects. Recent advances in multi-beam and multi-column electron beam imaging also enable large-scale SEM data collection. In this way, after the inspection is complete, at each detected defect location (or selected subset of detected defect locations) determined from the inspection, a test SEM image can be captured on a mask review, CD-SEM, or electron beam inspection tool (alternatively, an ion beam image can be captured using one of the ion beam tools described herein). Each test charged particle beam image may then be arranged from one or more of the following aspects, defect grades, and printabilities further described herein.

[0088] A computer subsystem or another system or method may use the results of any of the steps described herein to control and / or modify a mask repair or cleaning process aimed at removing or repairing one or more mask defects, thereby preventing such mask defects from being printed on a wafer. In one such example, information regarding defects detected on a mask may be used by a computer subsystem, another system (e.g., a mask repair system), or another method (e.g., a mask repair method) described herein to determine where on the mask the repair or cleaning process should be performed. Information about defects detected on a mask may also be used to determine one or more parameters of such a process, such as the area on which the process should be performed, the length of time for which the process should be performed, and which chemical substances or gases should be used in the process. The modified mask can then be used to print dies on the mask on a wafer in a lithography process. The modified mask may also be re-inspected using the embodiments described herein at another time after being introduced or re-introduced into the lithography process.

[0089] In one embodiment, the computer subsystem is configured to send any of the results described herein for any of the detected defects to a mask repair tool, and the mask repair tool uses the results for the detected defects in a repair process executed on the mask. In this way, defect information can be used by the repair tool when it is necessary to fix the mask. In one such example, as shown in FIG. 1, the computer subsystem 118 (or any other computer subsystem of the system) can be configured to send information about the detected defect to the mask repair tool 140, and the mask repair tool may or may not be part of the system. The mask repair tool can have any suitable configuration known in the art. Further, the mask repair tool may be a commercially available mask repair tool such as the MeRiT neXT system commercially available from Carl Zeiss SBE, LLC, Thornwood, NY. The mask repair tool uses the defect information to determine one or more parameters of the repair process, such as the repair location and area, for the purpose of repairing the detected defect without changing the features correctly formed in the vicinity of the detected defect. The embodiments described herein have several advantages over methods and systems for mask inspection in addition to the advantages already described herein. For example, lithographically critical defects on a mask can have a devastating impact on yield, leading to time and cost losses for the wafer manufacturer, and in extreme cases, a complete loss of profitability, especially for EUV masks. It is important to push the usable sensitivity of resolution-limited inspection tools before chemical ray photomask inspection tools for EUV masks mature. By combining the DD detection flow and the DB detection flow, the user can maximize the usable sensitivity and utilize the advantages of both the DD detection flow and the DB detection flow. The user is more suitable for reliably detecting lithographically critical defects on a mask having a multi-die layout.

[0090] Each of the embodiments of the above-described system can be combined together into a single embodiment.

[0091] Another embodiment relates to a computer-implemented method for detecting defects on a mask. This method includes generating a DB reference image of a mask with a plurality of dies formed thereon by simulating the physical version of the mask from the mask design and simulating an image of the physical version of the mask generated by an imaging subsystem from the simulated physical version of the mask. The method also includes detecting a first defect on the mask by comparing the DB reference image with an image of the mask generated by the imaging subsystem for a first die among the plurality of dies. Additionally, the method includes generating a die reference image for a first die among the plurality of dies by applying one or more parameters of the imaging subsystem learned by generating the DB reference image to an image generated by one or more imaging subsystems among the plurality of dies other than the first die. The method further includes detecting a second defect on the mask by comparing the die reference image with an image of the mask generated by the imaging subsystem for the first die among the plurality of dies. Generation of the DB reference image, detection of the first defect, generation of the die reference image, and detection of the second defect are performed by a computer subsystem coupled to the imaging subsystem.

[0092] Each of the steps of the method can be performed as further described herein. The method may also include any other steps that may be performed by an imaging subsystem and / or a computer subsystem or system configured according to any of the embodiments described herein, or may be implemented by such. Additionally, the method described above can be performed by any of the embodiments of the system described herein.

[0093] A further 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 mask. Such an embodiment is shown in FIG. 8. In particular, as shown in FIG. 8, the non - transitory computer - readable medium 800 includes program instructions 802 executable on a computer system 804. The computer - implemented method may include any step of any method described herein.

[0094] The program instructions 802 implementing a method such as those described herein may be stored on the computer - readable medium 800. The computer - readable medium can be a storage medium such as a magnetic or optical disk, magnetic tape, or any other suitable non - transitory computer - readable medium known in the art.

[0095] The program instructions can be implemented in any of a variety of ways, including, among other things, procedure - based techniques, component - based techniques, and / or object - oriented techniques. For example, the program instructions may be implemented using ActiveX controls, C objects, JavaBeans, Microsoft Foundation Classes ("MFC"), SSE (Streaming SIMD Extension), or other technologies or methodologies, as desired.

[0096] The computer system 804 can be configured according to any of the embodiments described herein.

[0097] Further modifications and alternative embodiments of various aspects of the present invention will be apparent to those skilled in the art in view of this description. For example, methods and systems for detecting defects on a mask are provided. Accordingly, this description should be construed as illustrative only and for the purpose of teaching those skilled in the art a general manner of implementing the present invention. It is to be understood that the forms of the invention shown and described herein are to be taken as presently preferred embodiments. Elements and materials may be substituted for those illustrated and described herein, parts and processes may be reversed, and certain features of the invention may be utilized independently, all of which will become apparent to those skilled in the art after having the benefit of this description of the invention. Changes may be made to the elements described herein without departing from the spirit and scope of the invention as set forth in the following claims.

Claims

1. A system configured to detect defects on a mask, comprising: an imaging subsystem configured to generate an image of the mask on which a plurality of dies are formed; a computer subsystem, simulating a physical version of the mask from a design of the mask, and simulating an image of the physical version of the mask generated by the imaging subsystem from the simulated physical version of the mask, to generate a database reference image of the mask; detecting a first defect on the mask by comparing the database reference image with the image of the physical version of the mask generated by the imaging subsystem for a first die among the plurality of dies; applying one or more parameters of the imaging subsystem learned by generating the database reference image to an image generated by the imaging subsystem for one or more of the plurality of dies other than the first die among the plurality of dies, to generate a die reference image for the first die among the plurality of dies; detecting a second defect on the mask by comparing the die reference image with the image of the physical version of the mask generated by the imaging subsystem for the first die among the plurality of dies; a computer subsystem configured as such; and a system comprising the same.

2. The system according to claim 1, wherein the mask is configured to be used at one or more extreme ultraviolet wavelengths of light.

3. The system according to claim 1, wherein the imaging subsystem is further configured to generate an image of the mask by irradiating the mask with light having a wavelength of 193 nm.

4. The system according to claim 1, wherein the imaging subsystem is further configured to generate an image of the mask by irradiating the mask with light having one or more wavelengths in the range of 193 nm to 257 nm.

5. The system according to claim 1, wherein the imaging subsystem is further configured to generate an image of the mask by irradiating the mask with light having a wavelength of 13.5 nm.

6. The system according to claim 1, wherein the imaging subsystem is further configured to generate an image of the mask by electronically illuminating the mask.

7. The system according to claim 1, wherein the imaging subsystem is further configured to generate an image of the mask by ionically illuminating the mask.

8. The system according to claim 1, wherein simulating the physical version of the mask includes near-field modeling, and simulating an image of the physical version of the mask includes partially coherent illumination modeling.

9. The system according to claim 1, wherein simulating an image of the physical version of the mask includes calibrating one or more parameters of the imaging subsystem in a model used to simulate the image of the physical version of the mask for focus error and aberration error of the imaging subsystem.

10. One or more parameters of the imaging subsystem learned by generating the database reference image include basis images representing focus error and aberration error of the imaging subsystem, and the step of generating the die reference image includes synthesizing the die reference image as a linear combination of an image generated by the imaging subsystem of one or more of the plurality of dies other than the first die among the plurality of dies and the basis image. The system according to claim 1, characterized in that it comprises.

11. The computer subsystem is determining the characteristics of one or more features of the mask within an inspection area of an image of the first die among the plurality of dies on the mask, selecting the database reference image or the die reference image for detecting the second defect within the inspection area based on the determined characteristics, The system according to claim 1, further configured as follows.

12. The computer subsystem is further configured to obtain information regarding one or more patterns closest to the first and second defects on the mask, and classify the first and second defects based on information regarding one or more patterns closest to the first and second defects on the mask respectively. The system according to claim 1, characterized in that.

13. The computer subsystem is further configured to determine classification categories for the first and second defects, and assign scores to the first and second defects based on the determined classification categories, wherein the scores indicate the severity of the first and second defects with respect to a lithography process performed using the mask. The system according to claim 1, characterized in that.

14. The computer subsystem is further configured to generate an inspection result for the mask including information about the first and second defects. The system according to claim 1, characterized in that.

15. The characteristics of the optical proximity effect correction features in two or more of the plurality of dies are designed to be different for flare correction in a lithography process performed using the mask, and generating the database reference image includes using designs of two or more of the plurality of dies respectively to generate two or more different database reference images of the plurality of dies on the mask. The system according to claim 1, characterized in that.

16. The characteristics of the optical proximity effect correction features in two or more of the plurality of dies are designed to be different for flare correction in a lithography process performed using the mask, and the computer subsystem is further configured to determine the sensitivity at which the second defect is detected in two or more of the plurality of dies based on the positions of the optical proximity effect correction features in two or more of the plurality of dies. The system according to claim 1, characterized in that.

17. The computer subsystem is further configured to apply one or more parameters of the imaging subsystem learned by generating the database reference image to an image generated by one or more of the plurality of dies other than the second die among the plurality of dies, thereby generating an additional die reference image for the second die among the plurality of dies, and detecting the second defect on the mask includes comparing the additional die reference image with an image of the mask generated by the imaging subsystem for the second die among m dies. The system according to claim 1, characterized in that it comprises.

18. The system according to claim 1, characterized in that the first defect includes a systematic mask writer error.

19. The image of the mask generated by the imaging subsystem, used to detect the first defect, generate the die reference image, and detect the second defect, is selected only from images of die-equivalent regions of the plurality of dies, and the die-equivalent regions are smaller than the plurality of dies. The system according to claim 1, characterized in that it is.

20. The system according to claim 19, characterized in that the die-equivalent region has a one-dimensional size equal to the width of the swath scanned on the mask by the imaging subsystem.

21. The system according to claim 19, characterized in that the plurality of dies include dies formed in a two-dimensional array on the mask.

22. A computer-readable medium storing program instructions executable on a computer system for executing a computer-implemented method for detecting defects on a mask, the computer-implemented method comprising: generating a database reference image for the mask on which a plurality of dies are formed by simulating the physical version of the mask from the design of the mask and simulating an image of the physical version of the mask generated by the imaging subsystem from the simulated physical version of the mask; detecting a first defect on the mask by comparing the database reference image with an image of the mask generated by the imaging subsystem for a first die among the plurality of dies; By applying one or more parameters of the imaging subsystem learned by generating the database reference image to an image generated by one or more of the imaging subsystems of a plurality of dies other than the first die among the plurality of dies, generating a die reference image for the first die among the plurality of dies; Detecting a second defect on the mask by comparing the die reference image with an image of the mask generated by the imaging subsystem for the first die among the plurality of dies; A computer-readable medium comprising.

23. A computer-implemented method for detecting a defect on a mask, Simulating a physical version of the mask from the design of the mask, and generating a database reference image for the mask on which a plurality of dies are formed by simulating an image of the physical version of the mask generated by an imaging subsystem from the simulated physical version of the mask; Detecting a first defect on the mask by comparing the database reference image with an image of the mask generated by the imaging subsystem for a first die among the plurality of dies; By applying one or more parameters of the imaging subsystem learned by generating the database reference image to an image generated by one or more of the imaging subsystems of a plurality of dies other than the first die among the plurality of dies, generating a die reference image for the first die among the plurality of dies; Detecting a second defect on the mask by comparing the die reference image with an image of the mask generated by the imaging subsystem for the first die among the plurality of dies; Including, the step of generating the database reference image, the step of detecting the first defect, the step of generating the die reference image, and the step of detecting the second defect are executed by a computer subsystem coupled to the imaging subsystem.

Images