Method and control device for analyzing an image of a blank photomask for microlithography

Abstract

Method for analyzing an image (104) of a blank (100) of a photomask (102) for microlithography, wherein the image (104) has a two-dimensional arrangement (114) of pixels (116) and a two-dimensional intensity distribution (118) depending on the pixels (116), comprising the steps: a) Determining (S101) gradient values ​​(G) of the two-dimensional intensity distribution (118) of the image (104), and b) Determining (S102) a candidate (K) of a defect (D) of the blank (100) shown in the image (104) based on determined gradient values ​​(G) that are greater than a gradient threshold (SW1, SW2).

Description

[0001] The present invention relates to a method and a control device for analyzing an image of a blank photomask for microlithography.

[0002] Microlithography is used to manufacture microstructured components, such as integrated circuits. The microlithography process is carried out using a lithography system equipped with an illumination system and a projection system. The image of a mask (reticle) illuminated by the illumination system is projected by the projection system onto a substrate coated with a light-sensitive layer (photoresist) and arranged in the image plane of the projection system, for example, a silicon wafer, in order to transfer the mask structure to the light-sensitive coating of the substrate.

[0003] Driven by the pursuit of ever smaller structures in the production of integrated circuits, EUV lithography systems are currently being developed that use light with a wavelength in the range of 0.1 nm to 30 nm, in particular 13.5 nm.

[0004] The mask used in microlithography (i.e., lithography mask) is created from a mask blank. Microstructures are arranged on the mask blank in order to project the mask structure thus created onto the substrate on a reduced scale. The structures of the lithography mask are themselves very small, ranging in size from a few micrometers to a few nanometers. In order to produce microstructured components with high accuracy using microlithography, the structures of the lithography mask must also be manufactured very precisely and the lithography mask must be free of defects (e.g., faulty structures and contamination). A defect-free lithography mask is also very important because a lithography mask is usually used for a large number of exposures. For this reason, blanks for lithography masks and already structured lithography masks are examined with great effort for defects.

[0005] Against this background, it is an object of the present invention to provide an improved method and an improved control device for analyzing an image of a blank photomask for microlithography.

[0006] Accordingly, a method for analyzing an image of a photomask blank for microlithography is proposed. The image has a two-dimensional arrangement of pixels and a two-dimensional intensity distribution depending on the pixels. The method further comprises the steps: a) Determining gradient values ​​of the two-dimensional intensity distribution of the image, and b) Determining a candidate defect of the blank depicted in the image based on determined gradient values ​​that are greater than a gradient threshold.

[0007] By applying the gradient values, candidates for imaged defects on the blank that appear only as faint elements in the image can also be identified. In other words, candidates for defects on the blank that have a low signal-to-noise ratio can also be identified.

[0008] In particular, an intensity gradient of the image of the blank from a background intensity of the image to intensity values ​​in a region of a real imaged defect of the blank typically has a larger gradient value than an intensity gradient of the image from the background intensity to intensity values ​​of an artifact caused by image noise. Thus, based on the gradient values, an image of a real defect can be better distinguished from an image artifact.

[0009] Furthermore, the gradient values ​​are mathematical values ​​of determined gradients. Therefore, by applying the gradient values, both a decrease in intensity and an increase in intensity of the image can be determined. This allows the detection of so-called bright defects (brighter than the background intensity), so-called dark defects (darker than the background intensity), as well as defects that have both a bright component (brighter than the background intensity) and a dark component (darker than the background intensity).

[0010] The blank is a blank for a lithography mask. The blank is, for example, a blank for a transmissive lithography mask for DUV lithography (DUV: "deep ultra violet," working light wavelength in the range of 30–250 nm) or a blank for a reflective lithography mask for EUV lithography (EUV: "extreme ultra violet," working light wavelength in the range of 0.1–30 nm, especially 13.5 nm).

[0011] The blank comprises, for example, a substrate. The substrate comprises, for example, one or more layers. The substrate comprises, for example, silicon dioxide (SiO2), e.g., quartz glass, and / or an alternating sequence of molybdenum and silicon layers.

[0012] In particular, the blank does not yet have any structures (e.g., absorber structures) that are projected onto a substrate during the microlithographic process. One could also say that the blank is free of a structural pattern that is to be projected onto a substrate during the microlithographic process.

[0013] The blank, for example, has a main plane of extension. The image of the blank is captured, for example, with an image recording device whose line of sight is (e.g., substantially) perpendicular to the main plane of extension of the blank.

[0014] A defect in the blank is, for example, a particle located on a surface of the blank or between adjacent layers of the blank. The particle is, for example, a foreign body, such as a dust particle and / or a contaminant.

[0015] The image of the blank is, for example, a microscopic image of the blank. The image of the blank is, for example, captured using a microscope image acquisition unit (e.g., a light microscope). The image of the blank can also be captured using a scanning microscope, e.g., a scanning electron microscope and / or a scanning ion microscope. One advantage of capturing the image with a light microscope over one with a scanning microscope is that the light microscope has a larger field of view, allowing a larger area of ​​the blank to be captured. Another advantage of the light microscope is a less invasive imaging technique compared to a scanning microscope.

[0016] The image of the blank is, for example, an image of a portion of the blank. For example, a size of the blank in its main plane of extension is approximately 15 cm × 15 cm. Furthermore, an image section for capturing the image is merely exemplary, for example, 17.5 µm × 17.5 µm.

[0017] In particular, the image has intensity values ​​assigned to the pixels of the two-dimensional arrangement of pixels, which form the two-dimensional intensity distribution.

[0018] The determined gradient values ​​are, in particular, each an absolute value of a respective determined gradient (or a squared absolute value of a respective determined gradient). The gradients of the two-dimensional intensity distribution of the image correspond, in particular, to a respective first derivative of the intensity distribution.

[0019] Determining the gradient values ​​includes, for example, determining a gradient of the two-dimensional intensity distribution at several or all pixels of the image. A gradient for a specific pixel is determined, for example, based on the evaluation of a predetermined number of pixels surrounding that pixel. The predetermined number includes, for example, the pixels arranged around the central pixel in a 3 × 3 pixel, 5 × 5 pixel, 7 × 7 pixel, 9 × 9 pixel, and / or 11 × 11 pixel square. The individual pixels can be included in the gradient calculation with different predetermined weightings.

[0020] Determining the gradient values ​​of the image involves, for example, determining a matrix of gradients (e.g., a gradient image) from the original image (i.e., the two-dimensional intensity distribution) or a portion of the original image (i.e., a portion of the two-dimensional intensity distribution). Contours of the candidate defect detected in the image are located at the pixels where the intensity (brightness) of the original image changes most significantly, and thus the gradient image exhibits the highest intensities. In other words, a contour of an imaged defect corresponds to a region of large gradients in the intensity distribution.

[0021] For example, step a) of the method may optionally include determining (e.g., calculating) squares of the determined gradient values ​​(i.e., calculating squared gradient values). This can achieve an enhancement of the gradient values ​​corresponding to the contours of an imaged defect.

[0022] For example, in step a), gradient values ​​can be determined for the entire image (i.e., each pixel of the image). Furthermore, in step b), gradient values ​​can be evaluated for the entire image or only for a region of interest in the image, i.e., compared with the gradient threshold.

[0023] The region of interest (ROI) of the image, for example, is a pre-determined area of ​​the image that potentially contains a defect. For example, the size of the ROI of the image can be just a few micrometers (e.g., 2 µm × 2 µm, 3 µm × 3 µm, 4 µm × 4 µm, or 5 µm × 5 µm).

[0024] The gradient threshold is, in particular, a pre-determined gradient threshold. The gradient threshold is determined, for example, based on the determined gradient values ​​of the two-dimensional intensity distribution of the region of interest in the image.

[0025] According to one embodiment, in step b), a pixel-dependent location and / or a pixel-dependent property of the candidate of the defect depicted in the image is determined, and / or the pixel-dependent location of the candidate of the defect depicted in the image has a center of gravity, a contour and / or a region of the candidate, and / or the pixel-dependent property of the candidate of the defect depicted in the image has an area and / or a geometric shape of the candidate.

[0026] This allows the method to locate a found candidate defect on the blank in the image.

[0027] The pixel-dependent location of the defect candidate is, in particular, a location in the image that depends on the pixels of the image. The location in the image corresponds, in particular, to a location on the blank.

[0028] The pixel-dependent location of the defect candidate may, for example, include multiple pixels and / or subpixel positions of the image.

[0029] The pixel-dependent location of the candidate defect is, for example, a location (e.g., one or more pixels and / or subpixel positions) of the image and / or the region of interest of the image for which the determined gradient values ​​are equal to and / or greater than the gradient threshold.

[0030] A centroid of the defect candidate is, for example, a center point and / or a geometric centroid of the candidate (e.g., a geometric shape of the candidate).

[0031] A contour of the candidate defect is, for example, a contour and / or outer contour that encompasses (e.g. completely) the candidate of the imaged defect.

[0032] For example, a region of the defect candidate has pixel and / or subpixel positions of the image that are determined as pixel and / or subpixel positions of the candidate, e.g., because they lie on and / or within a contour of the candidate.

[0033] An area of ​​the candidate defect is, for example, a two-dimensional area in the image encompassed by a contour of the candidate (e.g., completely).

[0034] A geometric shape of the candidate defect is, for example, a two-dimensional geometric shape of the candidate that is encompassed by the contour of the candidate (e.g., completely).

[0035] According to a further embodiment, the gradient threshold is 70% or more, 80% or more and / or 90% or more of a maximum value of the determined gradient values.

[0036] The maximum value of the determined gradient values ​​is, for example, a maximum value determined for the entire image (i.e., a maximum value of all gradient values ​​determined for the entire image). However, the maximum value of the determined gradient values ​​can also, for example, be a maximum value determined for the region of interest of the image (i.e., a maximum value of all gradient values ​​determined for the region of interest of the image).

[0037] In the case where squares of the determined gradient values ​​(ie squared gradient values) are determined in step a), the gradient threshold may be 70% or more, 80% or more and / or 90% or more of a maximum value of the squared gradient values.

[0038] For example, a gradient threshold of 70% or more, 80% or more, and / or 90% or more of the maximum value of the determined gradient values ​​is a first gradient threshold applied to determine a rough position of a candidate defect depicted in the image. In a subsequent step, a second gradient threshold can then be applied to determine a fine position of the candidate defect depicted in the image using the determined rough position.

[0039] According to a further embodiment, in step b), several candidates for a corresponding number of defects of the blank depicted in the image and their respective pixel-dependent areas are determined. Furthermore, the one of the several determined candidates that has the largest area among the several candidates is determined as a real defect of the blank depicted in the image.

[0040] For example, the blank may have a defect that appears low-contrast in the image compared to the background intensity of the image. In this case, it may happen that several candidates for defects in the blank are found in step b) because one or more image artifacts are also identified as candidates. Furthermore, the image of a real defect usually occupies a larger area in the image than the area of ​​an artifact in the image caused by noise components. Therefore, the size of the determined area of ​​the identified candidates can be used to distinguish, among the several identified candidates, the candidate that corresponds (e.g., with high probability) to a real defect from candidates that correspond (e.g., with high probability) to an artifact.

[0041] According to a further embodiment, a gradient mean and a gradient maximum of the determined gradient values ​​are determined. Furthermore, the gradient threshold is greater than the gradient mean and less than the gradient maximum, and / or the gradient threshold is an average of the gradient mean and the gradient maximum.

[0042] The gradient mean (i.e., an average value, e.g., an arithmetic mean, of the determined gradients) approximately describes how the background intensity of the image varies, because the background intensity makes up the largest part of the image. The gradient maximum value (i.e., a maximum value of the determined gradients) captures the steepest increase or decrease in the intensity values ​​of the image. The steepest increase or decrease in the intensity values ​​of the image occurs in the region of the imaged defect. Consequently, the gradient maximum value essentially captures an edge of the imaged defect in the intensity image.

[0043] By determining the gradient threshold as a value that lies between the gradient mean and the gradient maximum and / or that is an average of the gradient mean and the gradient maximum, a threshold is applied that detects not only an edge of the imaged defect but also flatter areas of the imaged defect.

[0044] The gradient threshold determined in this embodiment may, for example, be a second gradient threshold that is subsequently applied to apply a first gradient threshold. In particular, the gradient threshold determined in this embodiment may be a second gradient threshold for determining a fine position of a candidate defect depicted in the image, wherein a rough position of the candidate was previously determined based on the first gradient threshold. For example, a rough position of a candidate defect depicted in the image may be determined based on a first gradient threshold that is 70% or more, 80% or more, and / or 90% or more of the maximum value of the gradient values ​​determined for the region of interest.Furthermore, for example, a fine position of the candidate defect depicted in the image may be determined based on a second gradient threshold that is greater than the gradient mean and less than the gradient maximum and / or an average of the gradient mean and the gradient maximum.

[0045] For example, the gradient mean and the gradient maximum are each determined for the entire image (ie, for the gradient values ​​of the entire image).

[0046] According to a further embodiment: a first gradient threshold is determined such that it is 70% or more, 80% or more and / or 90% or more of a maximum value of the determined gradient values, a second gradient threshold is determined such that it is greater than a gradient mean of the determined gradient values ​​and smaller than a gradient maximum of the determined gradient values ​​and / or that it is an average of the gradient mean and the gradient maximum, a rough pixel-dependent location of the candidate defect depicted in the image is determined based on the first gradient threshold, and a refined pixel-dependent location of the candidate of the defect depicted in the image is determined based on the determined coarse pixel-dependent location of the candidate and the second gradient threshold.

[0047] According to a further embodiment, the method comprises: Determining a characteristic of a background intensity of the two-dimensional intensity distribution of the image, wherein the characteristic has a background frequency distribution, a background mean value, a background median value, a background maximum value and / or a background minimum value of intensity values ​​of a background area of ​​the image, Determining that the candidate defect depicted in the image corresponds to a candidate bright defect if one or more intensity values ​​associated with the pixel-dependent location of the candidate are greater than the characteristic for the intensity values ​​of the background area, Determining that the candidate defect depicted in the image corresponds to a candidate for a dark defect if one or more intensity values ​​associated with the pixel-dependent location of the candidate are smaller than the characteristic for the intensity values ​​of the background area, Determining, for a region of interest of the image and / or for an enveloping region of the determined candidate of the defect depicted in the image, an upper intensity threshold of the two-dimensional intensity distribution in the case of a candidate of a bright defect and a lower intensity threshold of the two-dimensional intensity distribution in the case of a candidate of a dark defect, Determining a pixel-dependent contour of the candidate defect depicted in the image based on the determined upper intensity threshold in the case of a bright defect and based on the determined lower intensity threshold in the case of a dark defect, and / or Determining a centroid of the candidate defect depicted in the image based on the determined pixel-dependent contour of the candidate.

[0048] For example, the candidate defect depicted in the image is determined to be a bright defect candidate if one or more intensity values ​​associated with the pixel-dependent location of the candidate are greater than the background maximum value. For example, the candidate defect depicted in the image is determined to be a dark defect candidate if one or more intensity values ​​associated with the pixel-dependent location of the candidate are less than the background minimum value.

[0049] For example, the background area of ​​the image is the entire area of ​​the image outside the area of ​​interest.

[0050] However, the background area of ​​the image can also be, for example, the entire area of ​​the image including the area of ​​interest.

[0051] The enveloping area of ​​the identified candidate for the defect depicted in the image is, for example, a sub-area of ​​the region of interest. The enveloping area of ​​the identified candidate for the defect depicted in the image is, for example, a rectangle that encompasses the identified candidate such that a contour of the candidate is completely contained within the rectangle. The enveloping area of ​​the identified candidate for the defect depicted in the image is, for example, the smallest rectangle that encompasses the identified candidate.

[0052] According to a further embodiment: an upper and lower reference intensity threshold of the area of ​​interest and the envelope area are determined, a difference between the upper and lower intensity reference threshold is determined, the upper intensity threshold is determined in such a way that it is smaller than the upper reference intensity threshold by a pre-determined fraction of the difference, and the lower intensity threshold is determined in such a way that it is greater than the lower reference intensity threshold by a pre-determined fraction of the difference.

[0053] This determines the upper and lower intensity thresholds based on the determined upper and lower reference intensity thresholds of the region of interest and the envelope region, respectively.

[0054] For example, the pre-determined fraction for determining the upper intensity threshold has a value in the range of 5% to 30% (e.g., 10%, 15%, or 20%). The pre-determined fraction for determining the lower intensity threshold has a value in the range of 5% to 30% (e.g., 10%, 15%, or 20%). For example, the pre-determined fraction for determining the upper intensity threshold and the pre-determined fraction for determining the lower intensity threshold have the same value.

[0055] By selecting the upper intensity threshold for a bright defect to be lower than the upper reference intensity threshold by the pre-determined fraction of the difference, it is possible to consider not only the maximum intensity value of the candidate defect, but also intensity values ​​of the candidate that are slightly below the maximum intensity value. This allows a contour of the candidate defect to be determined.

[0056] The same applies to the case of a dark defect and the lower intensity threshold, which is selected to be greater than the lower reference intensity threshold by the pre-determined fraction of the difference.

[0057] According to a further embodiment: the upper reference intensity threshold corresponds to a maximum intensity value of the region of interest or the envelope, and the lower reference intensity threshold corresponds to a minimum intensity value of the region of interest or the envelope, or the gradient threshold is greater than a gradient mean of the determined gradient values ​​and less than a gradient maximum of the determined gradient values, and / or the gradient threshold is an average of the gradient mean and the gradient maximum, and large gradient areas in the image are identified based on detected gradient values ​​that are larger than the gradient threshold, the upper reference intensity threshold in the case of a candidate for a bright defect is determined based on an average of intensity values ​​of the large gradient range, the lower reference intensity threshold in the case of a candidate for a dark defect is determined based on an average of intensity values ​​of the large gradient range, and the lower reference intensity threshold in the case of a candidate for a bright defect and the upper reference intensity threshold in the case of a candidate for a dark defect are each determined based on the background intensity parameter.

[0058] For example, the upper reference intensity threshold in the case of a candidate for a bright defect is determined such that it is equal to the mean of the intensity values ​​of the large gradient range. For example, the lower reference intensity threshold in the case of a candidate for a dark defect is determined such that it is equal to the mean of the intensity values ​​of the large gradient range.

[0059] According to a further embodiment, the method determines a first and a second candidate for a defect depicted in the image. Furthermore, a pixel-dependent first contour of the first candidate, which comprises a first pixel-dependent region, a first pixel-dependent center of gravity of the first candidate, a pixel-dependent second contour of the second candidate, which comprises a second pixel-dependent region, and a second pixel-dependent center of gravity of the second candidate are determined. The method further comprises: Determining whether the first and second areas intersect, the first area forms a subset of the second area and / or a minimum distance between the first area and the second area is less than a distance threshold, Determining, if the first and second regions intersect, the first region forms a subset of the second region, and / or a minimum distance between the first and second regions is less than the distance threshold, that the first and second regions of the first and second candidates are regions of a single defect depicted in the image, and Determining a common centroid of the first and second regions based on the first and second pixel-dependent centroids of the first and second candidates.

[0060] This makes it possible to determine that two seemingly separate candidates actually correspond to an image of one and the same defect.

[0061] For example, a common contour of the first and second areas can also be determined based on the first and second contours.

[0062] The minimum distance between the first region and the second region is, for example, a minimum orthonormal distance between outer contours of the first and second candidate.

[0063] The distance threshold is specified as a multiple of a number of pixels. For example, the distance threshold is 30 pixels or less, 20 pixels or less, and / or 10 pixels or less.

[0064] According to a further embodiment: the image of the photomask blank is taken using an illuminating light with a wavelength in the UV range, DUV range, less than 300 nm, less than 250 nm, less than 200 nm and / or less than 100 nm, and / or the method comprises a step of taking one or more images of the blank of the photomask, wherein the blank is illuminated with an illumination light having a wavelength in the UV range, DUV range, less than 300 nm, less than 250 nm, less than 200 nm and / or less than 100 nm, an illumination light reflected back from the blank is detected, and the two-dimensional intensity distribution of the image is based on the reflected back illumination light.

[0065] The acquisition of the one or more images of the photomask blank takes place, in particular, before step a). The acquisition of the one or more images of the photomask blank takes place, for example, according to an image section of an image acquisition unit used. In particular, one or more images of a part of the blank corresponding to the image section are acquired. The image section includes, in particular, the region of interest.

[0066] For example, several images of the blank are taken and averaged, and the two-dimensional intensity distribution of the image is based on the averaged image.

[0067] For example, the illumination light has a wavelength between 190 nm and 200 nm, e.g. 193 nm.

[0068] According to a further embodiment, the method comprises: Determining a defect probability for a determined candidate of a defect depicted in the image, wherein a first pixel-dependent area of ​​the candidate is determined based on a first intensity distance to a background mean value of a background intensity of the image, a second pixel-dependent area of ​​the candidate is determined based on a second intensity distance to the background mean, the first intensity distance is greater than the second intensity distance, and the defect probability is determined based on a quotient of the first pixel-dependent area and the second pixel-dependent area.

[0069] By determining the defect probability for a detected candidate of a defect depicted in the image, the significance of a detected candidate can be better assessed with regard to whether it corresponds to a real defect or an image artifact.

[0070] In particular, the defect probability is determined by applying two intensity distances of different sizes. The two intensity distances can also be viewed as applying two criteria of different rigor for determining a candidate defect.

[0071] In particular, the first intensity distance, i.e., the larger of the two intensity distances, corresponds to a more stringent criterion for identifying a candidate defect. Applying the first intensity distance reduces the number of artifacts among the identified defect candidates.

[0072] The second intensity distance, i.e., the smaller of the two intensity distances, corresponds to a less stringent criterion for determining a candidate defect. By applying the second intensity distance, the number of artifacts among the identified defect candidates can be increased. For example, one can say that the larger the number of artifacts among the identified defect candidates, the larger the determined second area and the lower the determined defect probability.

[0073] In embodiments, a defect probability for a determined candidate defect depicted in the image can also be determined as follows—instead of based on the above-described first and second intensity distances: The first area is determined based on the above-described upper intensity threshold in the case of a bright defect and based on the above-described lower intensity threshold in the case of a dark defect. Furthermore, the second area is calculated as the total area of ​​the image minus the first area. The defect probability is then determined based on the quotient of the first and second areas.

[0074] According to a further embodiment, the method comprises: Determining a frequency distribution and a standard deviation of the frequency distribution for intensity values ​​of a background area of ​​the image, and Determine the first and second intensity distance based on the standard deviation of the frequency distribution.

[0075] This allows the first and second intensity distances (ie the two different stringency criteria for identifying a candidate defect) to be determined based on a statistical evaluation of the background intensity.

[0076] For example, the first and second intensity distances are each determined based on the standard deviation of the frequency distribution, for example, as a multiple of the standard deviation. For example only, the first intensity distance is a first multiple (e.g., four) of the standard deviation of the frequency distribution, and the second intensity distance is a second multiple (e.g., three) of the standard deviation of the frequency distribution.

[0077] According to a further embodiment, the method comprises: Determining that the identified candidate corresponds to a real defect depicted in the image if the defect probability is greater than a pre-determined probability threshold, and Determine that the identified candidate is an artifact if the defect probability is equal to or less than the pre-determined probability threshold.

[0078] According to a further embodiment, the method produces a photomask for microlithography, and the method comprises, after step b): Arranging microstructures on the blank based on the determined pixel-dependent location of the candidate defect depicted in the image.

[0079] This allows, for example, locations and / or areas of the blank where one or more candidates for defects and / or one or more defects have been detected to be avoided when arranging the microstructures. Consequently, the accuracy of the produced photomask (lithography mask) can be improved.

[0080] The arrangement of microstructures on the blank can be carried out - in addition to based on the determined pixel-dependent location of the candidate defect depicted in the image - also based on a determined defect probability of the candidate defect depicted in the image.

[0081] Consequently, the method according to this embodiment is a method for manufacturing a photomask for microlithography.

[0082] According to a further aspect, a computer program product is proposed. The computer program product comprises instructions which, when the program is executed by at least one computer, cause the computer to execute a method as described above (e.g., also one or more of the embodiments of the method described above).

[0083] A computer program product, such as a computer program means, can be provided or delivered, for example, as a storage medium, such as a memory card, USB stick, CD-ROM, DVD, or in the form of a downloadable file from a server in a network. This can be done, for example, in a wireless communications network by transmitting a corresponding file with the computer program product or the computer program means.

[0084] According to a further aspect, a control device for analyzing an image of a photomask blank for microlithography is proposed. The image has a two-dimensional arrangement of pixels and a two-dimensional intensity distribution depending on the pixels. Furthermore, the control device comprises: a first determination device for determining gradient values ​​of the two-dimensional intensity distribution of the image, and a second determining device for determining a candidate defect of the blank depicted in the image based on gradient values ​​that are greater than a gradient threshold.

[0085] The control device is in particular designed to carry out a method as described above (e.g. also one or more of the embodiments of the method described above).

[0086] The control device may also comprise one or more further determination devices to carry out one or more steps of the method described above according to one or more described embodiments. For example, a third determination device may be provided, which is configured to determine a defect probability for a candidate defect in the blank determined by the second determination device.

[0087] The respective unit, for example, the control device and each of the detection devices described herein, can be implemented in hardware and / or software. In a hardware implementation, the respective unit can be configured as a device or as part of a device, for example, as a computer or a microprocessor or as a control computer of a vehicle. In a software implementation, the respective unit can be configured as a computer program product, as a function, as a routine, as part of a program code, or as an executable object.

[0088] "One" in this case is not necessarily limited to a single element. Rather, multiple elements, such as two, three, or more, may also be included. Any other counting term used here should not be understood as implying a limitation to the exact number of elements stated. Rather, numerical deviations upwards and downwards are possible, unless otherwise stated.

[0089] The embodiments and features described for the method apply accordingly to the computer program product and the proposed control device and vice versa.

[0090] Further possible implementations of the invention also include combinations of features or embodiments described above or below with respect to the exemplary embodiments that are not explicitly mentioned. In this case, the person skilled in the art will also add individual aspects as improvements or additions to the respective basic form of the invention.

[0091] Further advantageous embodiments and aspects of the invention are the subject of the dependent claims and the exemplary embodiments of the invention described below. The invention is explained in more detail below using preferred embodiments with reference to the accompanying figures. Fig. 1 shows a blank for a photomask for microlithography in a plan view according to one embodiment; Fig. 2 shows the blank in a cross-sectional view along line II-II in Fig. 1; Fig. 3 shows a photomask for microlithography with the blank made of Fig. 1 according to one embodiment; Fig. 4 shows a picture of the blank from Fig. 1, taken with an image recording device, according to an embodiment; Fig. Figure 5 shows a flowchart of a method for analyzing the image of the blank from Fig. 4 according to one embodiment; Fig. 6 shows a one-dimensional intensity distribution of the image from Fig. 4 according to one embodiment; Fig. 7 shows gradient values ​​of the intensity distribution from Fig. 6 and a first gradient threshold according to an embodiment; Fig. 8 shows an area of ​​interest of the image Fig. 4 with a pictured candidate for a defect of the blank from Fig. 1 according to one embodiment; Fig. 9 shows an area of ​​interest of the image Fig. 4 with several candidates for a defect in the blank shown Fig. 1 according to one embodiment; Fig. 10 shows a view similar Fig. 7, illustrating a second gradient threshold according to an embodiment; Fig. 11 shows another embodiment of an intensity distribution of the image from Fig. 4; Fig. 12 shows gradient values ​​of the intensity distribution from Fig. 11 according to one embodiment; Fig. 13 to 15 each show an area of ​​interest in the image Fig. 4 with two candidates for a defect of the blank shown in each case according to embodiments; Fig. 16 shows a view similar Fig. 6, wherein a first and second intensity distance to a background intensity of the image are characterized according to an embodiment; Fig. 17 shows an intensity distribution of a background intensity of the image from Fig. 4 according to one embodiment; Fig. 18 shows another embodiment of an intensity distribution of the image from Fig. 4, where compared to Fig. 16 a low-contrast defect of the blank Fig. 1 is shown; Fig. 19 illustrates a threshold for a defect probability according to one embodiment; Fig. 20 shows a flow diagram of a method for manufacturing a photomask for microlithography according to an embodiment; Fig. 21 shows a flowchart of a method for analyzing the image of the blank from Fig. 4 according to another embodiment; Fig. 22 shows in the lower part an enlarged section of the intensity distribution from Fig. 6 and in the upper part an enlarged section of the gradient distribution from Fig. 10 according to one embodiment; Fig. 23 shows another embodiment of an intensity distribution of the image from Fig. 4 according to one embodiment; Fig. 24 shows an area of ​​interest of the image Fig. 4 with five candidates for a defect of the blank according to an embodiment; and Fig. 25 shows a control device for analyzing the image of the blank from Fig. 4 according to one embodiment.

[0092] In the figures, identical or functionally equivalent elements are provided with the same reference numerals unless otherwise indicated. Furthermore, it should be noted that the representations in the figures are not necessarily to scale.

[0093] In the following, with reference to the Fig. 1 to 19, a method for examining a blank 100 of a photomask 102 for microlithography according to one embodiment is described. In particular, the method analyzes an image 104 of the blank 100.

[0094] Fig. 1 shows the blank 100 in a plan view and Fig. 2 in a cross-sectional view along line II-II in Fig. 1. The blank 100 has, for example, a cuboid shape with a plane parallel to the xy plane in Fig. 1 arranged main extension plane E.

[0095] The blank 100 may have a defect D. The defect D is, for example, a particle, such as a dust particle. The defect D is, for example, as shown in Fig. 2, is arranged between layers 106 of the blank 100. However, in other examples, the defect D may also be arranged on a surface 108 of the blank 100.

[0096] The defect D can impair the quality of a photomask 102 produced using the blank 100. Fig. 3 shows a photomask 102 produced from the blank 100. For example, during the production of the photomask 102, microstructures 110 are microlithographically arranged (e.g., generated) on the blank 100, the pattern of which is projected onto a wafer in the lithography process.

[0097] Since a defect D in the blank 100 can impair the quality of the photomask 102 produced using the blank 100, it is desirable to examine the blank 100 with the defect D more closely. For example, a location of the defect D on the blank 100 can be determined.

[0098] For this purpose, one or more images 104 of the blank 100 of the photomask 102 are provided and / or recorded. The one or more images 104 are, for example, microscopic images. The one or more images 104 are recorded, for example, using a light microscope. The one or more images 104 are recorded, for example, using an illumination light with a wavelength in the UV range, DUV range, less than 300 nm, less than 250 nm, less than 200 nm, and / or less than 100 nm. For example, one wavelength of the illumination light is 193 nm.

[0099] Fig. 4 shows an exemplary image 104 of the blank 100 from Fig. 1. An exemplary image section 112 captured in the image 104 is shown in Fig. 1. The image section 112 is in Fig. 1 is exaggerated for illustrative purposes. For example, the size of the blank 100 in its main extension plane E is approximately 15 cm × 15 cm. Furthermore, the image section 112 for recording the image 104 is, for example, 17.5 µm × 17.5 µm.

[0100] For example, several images (e.g. 15 images or a different number of images) of the blank 100 with the image section 112 are taken and averaged. For example, the Fig. 4, the image 104 shown is an averaged image of the image section 112 of the blank 100.

[0101] The image 104 has a two-dimensional arrangement 114 of pixels 116. A pixel size is, for example, a few to a few nanometers (e.g., 17.5 nm × 17.5 nm). The image 104 also has a two-dimensional intensity distribution 118 depending on the pixels 116. In particular, each of the pixels 116 is assigned a corresponding intensity value I iassigned so that the i-th pixel 116 has the intensity value I i has.

[0102] Optionally, a region of interest 120 (ROI) of the image 104 can be taken into account, in which the defect D is detected. The region of interest 120 can, for example, be determined by another examination device (not shown), with which the blank 100 was previously examined for defects D (e.g., with lower spatial resolution). In addition to or instead of a region of interest 120, an enveloping region 122 of a defect D depicted in the image 104 can also be determined during the process, which region is indicated by dashed lines in Fig. 4 is marked.

[0103] Fig. 5 shows a flow diagram of a method for analyzing the image 104 of the blank 100 according to one embodiment.

[0104] In a first step S101 of the method, gradient values ​​G of the two-dimensional intensity distribution 118 of the image 104 are determined.

[0105] For example, gradient values ​​G of the two-dimensional intensity distribution 118 are determined and assigned to each pixel 116 of the image 104 and / or each pixel 116 of the region of interest 120 of the image 104. A gradient and thus an associated gradient value G i(e.g., a mathematical value of the gradient) for a specific i-th pixel 116 is determined, for example, based on the evaluation of a predetermined number of pixels 116 surrounding this i-th pixel 116. The predetermined number comprises, for example, the pixels 116 arranged around the central i-th pixel 116 in a 3 × 3 pixel, 5 × 5 pixel, 7 × 7 pixel, 9 × 9 pixel, and / or 11 × 11 pixel square. The individual pixels 116 can be included in the calculation of the corresponding gradient with different predetermined weightings.

[0106] Determining the gradient values ​​G i for the image 104 and / or the region of interest 120 of the image 104 comprises, for example, determining a gradient image for the image 104 or the region of interest 120.

[0107] Optionally, the determined gradient values ​​G can be squared for further evaluation, so that squared gradient values ​​G are assigned to the pixels 116i be obtained.

[0108] In Fig. 6, a one-dimensional intensity distribution 124 is shown for illustrative purposes, which results from the two-dimensional intensity distribution 118 from Fig. 4 at the y-position corresponding to the line VI-VI in Fig. 4 corresponds to. In Fig. 6, the intensity I of the image 104 is shown as a function of the x-position of the image 104 in a region of the defect D. In Fig. 6 the increase of intensity I in the area of ​​defect D is clearly visible.

[0109] In Fig. 7 is a one-dimensional distribution 126 of the determined gradient values ​​G of the Fig. 6. Although the determined two-dimensional gradient image (not shown) is evaluated in the method, for illustrative purposes the method is described here using the one-dimensional case ( Fig. 6 and Fig. 7). Examples are Fig. 7 squared gradient values ​​G are shown. Although the procedure is described below for squared gradient values ​​G, non-squared gradient values ​​G can also be evaluated in other examples.

[0110] In a second step S102 of the method, a candidate K of a defect D of the blank 100 depicted in the image 104 is determined based on gradient values ​​G that are greater than a gradient threshold value SW1, SW2.

[0111] The gradient threshold SW1, SW2 can be determined according to two different embodiments (first and second embodiments), as described below, so that a first gradient threshold SW1 and / or a second gradient threshold SW2 are determined accordingly. The two embodiments are preferably applied one after the other, with the first embodiment being used to determine a rough position (130a, Fig. 24) of a candidate K of a defect D is determined and, using the second embodiment, a fine position (130b, Fig. 24) of the candidate K of the defect D. This is also in connection with the procedure from Fig. 21. Alternatively, however, only one of the two embodiments, ie, one of the first gradient threshold SW1 and the second gradient threshold SW2, may be applied.

[0112] In the first embodiment S102a of step S102, the gradient threshold value SW1 (first gradient threshold value SW1) is 70% or more, 80% or more and / or 90% or more of a maximum value G max the determined gradient values ​​G, as in Fig. 7 shown.

[0113] To do this, the maximum value G max of the determined gradient values ​​G. The maximum value G max is used, for example, for the (e.g. squared) gradient values ​​G i of the entire image 104 (ie all pixels 116 of the entire image 104) and / or for the (e.g. squared) gradient values ​​G i of the area of ​​interest 120 of the image 104 (ie all pixels 116 of the area of ​​interest 120).

[0114] An example is Fig. 7 as the first gradient threshold SW1 a value of 90% of the maximum value G max of the determined gradient values ​​G. It is then checked whether there are gradient values ​​G in the image 104 or in the region of interest 120 that are greater than the first gradient threshold SW1. Gradient values ​​G that are greater than the first gradient threshold SW1 (in Fig. 7 with the reference number 128) are interpreted as an indication of a candidate K for a defect D of the blank 100 recorded in the image 104.

[0115] In addition, in step S102, a pixel-dependent location 130 and / or a pixel-dependent property 131 of the candidate K of the defect D depicted in the image 104 can be determined. Fig. 8 shows a section of Fig. 4. As in Fig. 8, the pixel-dependent location 130 of the candidate K comprises, for example, a centroid Z, a contour C, and / or a region B of the candidate K. Furthermore, the pixel-dependent property 131 of the candidate K comprises, for example, an area F and / or a geometric shape Q of the candidate K. The pixel-dependent location 130 and / or the pixel-dependent property 131 of the candidate K of the defect D can, for example, comprise a plurality of pixels 116 and / or subpixel positions of the image 104 corresponding to the centroid Z, the contour C, the region B, the area F, and / or the geometric shape Q of the candidate K.

[0116] For example, based on the first gradient threshold SW1, several pixels 116 and / or subpixel positions of the image 104 are determined for which the determined gradient values ​​G are greater than the first gradient threshold SW1. In this way, for example, an area B of the candidate K of the defect D is determined in the image 104. A geometric center of gravity Z of the determined area B can be determined as the position of the candidate K of the defect D. In Fig. 7 is an x-position x1 of the geometric center of gravity Z determined based on the first gradient threshold SW1 ( Fig. 8). Although in Fig. 7, a corresponding y-position of the geometric center of gravity Z of the candidate K in the image 104 is also determined based on the first gradient threshold SW1.

[0117] Based on the first gradient threshold SW1, for example, a rough position of the defect candidate K in the image 104 can be determined.

[0118] In the case of very low-contrast defects D, it may happen that in step S102 several candidates K, K1, K2, K3 are determined for a corresponding number of defects D of the blank 100 depicted in the image 104 and / or in the region of interest 120. This is shown in Fig. 9. In this case, a respective pixel-dependent area F, F1, F2, F3 of the corresponding candidates K, K1, K2, K3 can optionally be determined. Subsequently, the one of the several determined candidates K, K1, K2, K3 that has the largest area F among the several candidates K, K1, K2, K3 can be determined as a real defect D of the blank 100 depicted in the image 104.

[0119] In the second embodiment S102b of step S102 of the method, a differently determined gradient threshold SW2 (second gradient threshold SW2) is applied to determine a candidate K for a defect D of the blank 100.

[0120] In particular, Fig. 7 that in the first embodiment S102a of step S102, ie, when applying the first gradient threshold value SW1, possibly only one peak 132 of the gradient distribution 126 is detected. As a result, it may happen that a determined location 130 of the defect candidate K (e.g., the marked x-position x1) is inaccurate (see also reference numeral 130a in Fig. 24).

[0121] Therefore, in the second embodiment S102b of step S102, first - as in the first embodiment S102a - a gradient maximum value G maxof the determined gradient values ​​G. In addition, in the second embodiment S102b, a gradient mean value G mean of the determined gradient values ​​G, as in Fig. 10. The gradient maximum value G max and the gradient mean G mean are determined, for example, for the entire image 104 and / or for the area of ​​interest 120 of the image 104.

[0122] Furthermore, the second gradient threshold SW2 is determined as a value that is greater than the gradient mean G mean and smaller than the gradient maximum value G max For example, the second gradient threshold SW2 is an average of the gradient mean G mean and the gradient maximum value G max , as in Fig. 10 illustrated.

[0123] As in Fig. 10, the defect D and its location 130 can be better detected. For example, x- and y-positions of a contour C ( Fig. 8) of the defect candidate K better captured. In Fig. 10, in which only the one-dimensional case is shown for illustrative purposes, the x-positions x2 and x3 can be determined for a contour C of the defect candidate K. Based on the thus determined two-dimensional contour C of the defect candidate K, a center of gravity Z (e.g. geometric center of gravity Z, see x-position x4 in Fig. 6 and Fig. 10) of the defect candidate K can be determined.

[0124] Based on the second gradient threshold SW2, for example, a fine position 130b ( Fig. 24) of the defect candidate K in the image 104.

[0125] The Fig. 6, Fig. 7 and Fig. 10 illustrate a candidate K for a so-called bright defect Dh , whose intensity values ​​I brighter than a background intensity I H,mean ( Fig. 6) of image 104. In addition to such bright defects D h , there are also so-called dark defects D d ( Fig. 23), whose intensity values ​​I darker than the background intensity I H,mean Furthermore, so-called mixed defects D' ( Fig. 11) which in an image 104 with both a bright portion that is brighter than the background intensity I H,mean as well as a dark portion that is darker than the background intensity I H,mean is, can be depicted.

[0126] In Fig. Figure 11 illustrates a one-dimensional intensity distribution 134 of a candidate K for such a mixed defect D'. The intensity distribution 134 has a bright portion 136 that is brighter than the background intensity I H,meanThe intensity distribution 134 also has a dark portion 138 which is darker than the background intensity I H,mean is.

[0127] By calculating the gradient values ​​G' ( Fig. 12) and applying the second gradient threshold SW2', which is, for example, an average between a gradient average G' mean and a gradient maximum value G' max , a location 140 of the candidate K for the mixed defect D' can also be easily found. In particular, based on gradient values ​​G' that are greater than the second gradient threshold SW2', x- and y-positions of a contour C (cf. Fig. 8) of the defect candidate K'. In Fig. 12, the x-positions x5, x6 of the contour C are marked as examples. Based on this, for example, a geometric center of gravity (in Fig. 12 at x7) of the defect candidate K' can be determined.

[0128] In the method, for light, dark, and mixed defects D, it may happen that for a single defect D, two defect candidates K'' are found in the image 104, each corresponding to only a portion of the defect D. This can be taken into account by the following image evaluation steps.

[0129] As in the Fig. 13 to 15, the method (steps S101, S102) for example selects a first and a second candidate K A , K B ( Fig. 13) for a defect D shown in Figure 104. As shown in Fig. 13, can then be started with the procedure for the first candidate K A a pixel-dependent first contour C A , a first pixel-dependent area B A , which is from the first contour C A and a first pixel-dependent center of gravity Z A In addition, the procedure for the second candidate K Ba pixel-dependent second contour C B , a second pixel-dependent area B B , which is from the second contour C B and a second pixel-dependent center of gravity Z B be determined.

[0130] Fig. 14 indicates a first candidate K' for another case A and the one for this first candidate K' A location 130 determined by the method in the form of a pixel-dependent first contour C' A , a first pixel-dependent area B' A , which is from the first contour C' A and a first pixel-dependent center of gravity Z' A . In addition, Fig. 14 a second candidate K' B and the one for this second candidate K' B location 130 determined by the method in the form of a pixel-dependent second contour C' B , a second pixel-dependent area B' B , which is formed by the second contour C' Band a second pixel-dependent center of gravity Z' B shown.

[0131] Furthermore, Fig. 15 for another case a first candidate K'' A and the one for this first candidate K'' A location 130 determined by the method in the form of a pixel-dependent first contour C'' A , a first pixel-dependent area B'' A , which is from the first contour C'' A and a first pixel-dependent center of gravity Z'' A . In addition, Fig. 15 a second candidate K'' B and the one for this second candidate K'' B location 130 determined by the method in the form of a pixel-dependent second contour C'' B , a second pixel-dependent area B'' B , which is from the second contour C'' B and a second pixel-dependent center of gravity Z''3.

[0132] It can then be determined whether the first and second area B A , B B cut ( Fig. 13), the first area B' A a subset of the second area B' B forms ( Fig. 14) and / or a minimum distance a between the first area B'' A and the second area B'' B is smaller than a distance threshold a' ( Fig. 15).

[0133] If it is determined that the first and second area B A , B B cut ( Fig. 13), the first area B' A a subset of the second area B' B forms ( Fig. 14) and / or the minimum distance a between the first area B'' A and the second area B'' B smaller than a distance threshold a' ( Fig. 15), then it can be determined that the first and second area B A , B B (B' A , B' B ; B'' A, B'' B ) of the first and second candidate K A , K B (K' A , K' B ; K'' A , K'' B ) are areas of a single defect D shown in image 104.

[0134] In addition, a common focus Z G (Z' G , Z'' G ) of the first and second area B A , B B based on the first and second pixel-dependent center of gravity Z A , Z B of the first and second candidate K A , K B Furthermore, a common contour C'' G of the first and second area B'' A , B'' B based on the first and second pixel-dependent contour C'' A , C'' B of the first and second candidate K'' A , K'' B . can be determined, as exemplified in Fig. 15 shown.

[0135] In an optional third step S103 of the method, a defect probability P D for a determined candidate K of a defect D shown in Figure 104.

[0136] For this purpose, a background intensity I H of the image 104 and / or the region of interest 120 of the image 104. In addition, a first and a second intensity distance I E and I F to background intensity I H determined, where the first intensity distance I E greater than the second intensity distance I F is.

[0137] In Fig. 16 is the one-dimensional intensity distribution 124 of candidate K from Fig. 6 together with the background intensity I H and the first and second intensity distance I E and I F The first and second intensity distance I E and I Fcan also be called intensity thresholds.

[0138] In Fig. 17 is an example of determining the first and second intensity distance I E and I F In this example, the first and second intensity distance I E and I F based on a statistical evaluation of intensity values ​​I i of pixels 116 of a background area 142 of the two-dimensional intensity distribution 118 of the image 104 ( Fig. 4). The background area 142 of the image 104 is, for example, the entire area of ​​the image 104 outside the area of ​​interest 120 of the image 104. In Fig. 17 is on the left side a one-dimensional intensity distribution 144 of intensity values ​​I iof the background area 142 as a function of the lateral x-position of the image 104. In the statistical evaluation of the intensity distribution 144, in particular a frequency distribution 146 of the intensity values ​​I i the intensity distribution 144 of the background area 142 is determined as in Fig. 17 on the right. The frequency distribution 146 gives the frequency N of certain intensity values ​​I i as a function of the intensity I. Furthermore, a standard deviation σ and a mean I H,mean of the frequency distribution 146. The first and second intensity distance I E and I F are then determined based on the standard deviation σ of the frequency distribution 146 (e.g., as a multiple of the standard deviation σ). For example only, the first intensity distance I E four times the standard deviation σ (4σ) and the second intensity distance IF three times the standard deviation σ (3σ), as in Fig. 16 marked.

[0139] The Fig. The intensity distribution 124 shown in Figure 16 is an intensity distribution of a high-contrast defect candidate K, whose intensity values ​​I i differs significantly from the level of intensity values ​​I i the background intensity I H lift up.

[0140] To determine the defect probability P D of the candidate K is initially based on the first intensity distance I E a first pixel-dependent area F E of candidate K. The first area F E of the candidate K is based in particular on the intensity values ​​I i the intensity distribution 124 which is greater than the first intensity distance I E The x-positions x E1 and x E2 mark exemplary boundaries of the first area FE . Only exemplary and schematic is shown in Fig. 16 a circular area F E as an example of a first surface F E sketched, where the area F E may also have a shape other than circular and, in particular, may have an irregular shape.

[0141] Similar to the first intensity distance I E is also based on the second intensity distance I F a second pixel-dependent area F F of candidate K. Fig. 16 illustrates, by way of example and schematically, a second surface F F .

[0142] The defect probability P D for the defect candidate K is then calculated as the quotient of the first area F E and the second area F F formed: PD=FE / FF.

[0143] Since the Fig. 16 corresponds to a high-contrast candidate K, a size of the areas F E , F F only slightly from each other, resulting in a relatively high defect probability P D results.

[0144] In Fig. 18 shows another example of a one-dimensional intensity distribution 148 of another candidate K' for a defect D of the blank 100. The Fig. The intensity distribution 148 shown in Figure 18 is an intensity distribution of a low-contrast defect candidate K', whose intensity values ​​I i only slightly above the level of intensity values ​​I i a background intensity I' H (Background intensity distribution 150) of the image 104. The background intensity distribution 150 is, similar to the example of Fig. 17, statistically evaluated and a standard deviation σ' and a mean I' H,meana frequency distribution of the background intensity I' H The first and second intensity distance I' E and I' F are then, similar to the example of Fig. 16, for example, a fourfold standard deviation σ' (4σ') and a threefold standard deviation σ' (3σ') are applied. The determination of the first and second areas F' E , F' F of the candidate K' is carried out analogously to Fig. 16 described. In Fig. 18 are x-positions x' E1 and x' E2 as exemplary boundaries of the first surface F' E Unlike the example of Fig. 16, in which a high-contrast defect candidate K was considered, the low-contrast defect candidate K' results in Fig. 18 one - relative to the first surface F' E significantly larger second area F' F The reason for this is that the intensity values ​​I ithe background intensity I' H in many places 152 (ie for many x-positions in Fig. 18) greater than the second intensity distance I' F and thus to the second surface F' F contribute, as in Fig. 18. This gives the defect probability P' D as the quotient of the first area F' E and the second surface F' F (P' D = F' E / F' F ) is a relatively small value.

[0145] In the optional step S103, a pre-determined probability threshold value SW P ( Fig. 19) can be used to determine for a candidate K, K' of a defect D depicted in the image 104 whether it corresponds to a real defect or whether it is an artifact. In particular, it is determined that the candidate K corresponds to a real defect if the defect probability P Dgreater than the probability threshold SW P Furthermore, it is determined that the candidate K' is an artifact if the defect probability P' D equal to or less than the probability threshold SW P is.

[0146] The following is based on reference to Fig. 20 describes a method of fabricating a photomask 102 for microlithography according to one embodiment.

[0147] A first step S201 and a second step S202 of the method for producing the photomask 102 correspond to steps S101 and S102 of the method described with reference to the Fig. 1 to 19 (see especially flow chart in Fig. 5) by means of which a candidate K of a defect D of the blank ( Fig. 1). In other words, the blank 100 for a photomask 102, the image 104 of which is formed in the method according to Fig. 1 to 19 (especially Fig. 5) was analyzed, in which with reference to Fig. 20 described method for producing the photomask 102 is used.

[0148] Furthermore, in the method for producing the photomask 102 ( Fig. 20) in step S202 (which corresponds to step S102 in Fig. 5) a pixel-dependent location 130 of the found defect candidate K is determined. The determined pixel-dependent location 130 includes, for example, a center of gravity Z, a contour C and / or an area B of the candidate K ( Fig. 8). The pixel-dependent location 130 of the candidate K of the defect D may, for example, comprise several pixels 116 and / or subpixel positions of the image 104 corresponding to the centroid Z, the contour C, and / or the region B of the candidate K.

[0149] A third optional step S203 of the method for producing the photomask 102 corresponds to step S103 of the method described with reference to the Fig. 1 to 19 (see especially flow chart in Fig. 5) described procedure (determining a defect probability).

[0150] In a fourth step S204 of the method for producing the photomask 102, microstructures 110 ( Fig. 3) are arranged on the blank 100. The microstructures 110 are, for example, absorber structures. For example, when arranging the microstructures 110, locations 130 (e.g., regions B) of the blank 100 in which one or more candidates K for defects D and / or defects D were detected can be avoided. This can improve the accuracy of the produced lithography mask 102.

[0151] The arrangement of the microstructures 110 on the blank 100 in step S204 can, in addition to the determined pixel-dependent location 130 of the candidate K, also be based on a defect probability D determined for the determined defect candidate K P The defect probability D P is used, for example, as in connection with the Fig. 16 to 19 described.

[0152] The following is based on reference to Fig. 21 a method for analyzing an image 104 ( Fig. 4) a blank 100 ( Fig. 1) for a photomask 102 according to another embodiment.

[0153] In a first step S301 of the method according to Fig. 21 one or more parameters 154 ( Fig. 17) for intensity values ​​I i (ie an intensity distribution 144) of a background area 142 ( Fig. 4) a two-dimensional intensity distribution 118 of the image 104. The one or more parameters 154 have, for example, a frequency distribution 146, a mean value I H,mean , a median value, a maximum value I H,max and / or a minimum value I H,min of intensity values ​​I i of the background area 142 of the image 104. The one or more parameters 154 for intensity values ​​I i of the background area 142 of the image 104 are, for example, as in connection with Fig. 17 described.

[0154] A second and third step S302, S303 of the method according to Fig. 21 correspond to steps S101 and S102 of the process with reference to Fig. 1 to 19 (see especially flow chart in Fig. 5) by means of which a candidate K of a defect D of the blank 100 depicted in the image 104 is determined. In particular, in step S303 of the method according to Fig. 21 the first embodiment S102a of step S102 of the method according to Fig. 5 is applied. In other words, in step S303 of the method according to Fig. 21 a gradient threshold SW1 (first gradient threshold SW1) is applied which is 70% or more, 80% or more and / or 90% or more of a maximum value G max of the determined gradient values ​​G. Furthermore, in step S303 of the method according to Fig. 21 a coarse pixel-dependent location 130, 130a of the determined defect candidate K in the form of a (e.g. geometric) center of gravity Z ( Fig. 8, Fig. 24) of the defect candidate K is determined based on the first gradient threshold SW1. In particular, the coarse pixel-dependent location 130, 130a of the defect candidate K is determined based on gradient values ​​G that are greater than the first gradient threshold SW1.

[0155] In step S303 of the method according to Fig. 21, a rough position 130, 130a, Z of the defect candidate K in the image 104 is determined, in particular based on the first gradient threshold value SW1.

[0156] In a fourth step S304 of the method according to Fig. 21, it is determined whether the defect candidate K found in step S303 corresponds to a candidate K for a bright defect D h or a candidate K for a dark defect D d In particular, it is determined that the candidate K corresponds to a candidate K for a bright defect D hcorresponds if one or more intensity values ​​I assigned to the pixel-dependent location 130 of the candidate K i greater than parameter 154 (e.g. greater than the maximum value I H,max ) for the intensity values ​​I i of the background area 142. It is also determined that the candidate K is a candidate K for a dark defect D d corresponds if one or more intensity values ​​I assigned to the pixel-dependent location 130 of the candidate K i smaller than the parameter 154 (e.g. smaller than the minimum value I H,min ) for the intensity values ​​I i of the background area 142.

[0157] A fifth step S305 of the method according to Fig. 21 corresponds to the second embodiment S102b of step S102 of the embodiment with reference to the Fig. 1 to 19 (see especially flow chart in Fig. 5). In particular, in step S305, a second gradient threshold SW2, as described with reference to Fig. 10. To do this, a gradient maximum value G max and a gradient mean G me-an of the determined gradient values ​​G ( Fig. 10). Furthermore, the second gradient threshold SW2 is determined as a value that is greater than the gradient mean G mean and smaller than the gradient maximum value G max For example, the second gradient threshold SW2 is defined as an average of the gradient mean G mean and the gradient maximum value G max determined as in Fig. 10. The second gradient threshold SW2 serves in particular to determine a fine position 130b ( Fig. 24) of candidate K for a defect D.

[0158] In a sixth step S306 of the method according to Fig. 21, an upper reference intensity threshold UL and a lower reference intensity threshold LL ( Fig. 22) of the area of ​​interest 120 or the envelope area 122 ( Fig. 4) was determined.

[0159] In the case of a bright defect D h the upper reference intensity threshold UL ( Fig. 22) and in case of a dark defect D d the lower reference intensity threshold LL' ( Fig. 23) based on the calculated gradient values ​​G as follows.

[0160] Fig. 22 shows in the lower part an enlarged section of Fig. 6 and in the upper part an enlarged section of Fig. 10. In particular, in the lower part of Fig. 22 the one-dimensional intensity distribution 124 (ie the intensity I as a function of the location x) of a candidate K of a bright defect D h In the upper part of Fig. 22 shows the gradient values ​​G calculated for the intensity distribution 124. In particular, in the upper part of Fig. 22 shows the one-dimensional distribution 126 of the gradient values ​​G determined for the intensity distribution 124. Furthermore, in the upper part of Fig. 22, similar to Fig. 10, the gradient mean G mean , the gradient maximum value G max and the second gradient threshold value SW2 calculated from it.

[0161] The areas of the image 104 (e.g. pixel 116) in which the gradient values ​​G are greater than the second gradient threshold SW2 are in Fig. 22 with the reference number 156. These areas 156 of large gradients are also shown in the lower graph of Fig. 22. The reference number 158 indicates those intensity values ​​I i the intensity distribution 124, which lie in the areas of 156 large gradients.

[0162] A mean value I GG of the intensity values ​​I lying in the areas of 156 large gradients i of the intensity distribution 124. This mean value I GG defined in the case of a candidate K for a bright defect D h the upper reference intensity threshold UL.

[0163] The lower reference intensity threshold LL in the case of a candidate K for a bright defect D h , as in Fig. 22 is based on a mean I H,mean the background intensity. In particular, the lower reference intensity threshold LL in the case of a candidate K for a bright defect D h a sum of the mean I H,meanof the background intensity and a multiple of a standard deviation σ of a frequency distribution 146 of background intensities. A multiple of the standard deviation σ means that the standard deviation σ is multiplied by a number n, where n is an integer greater than or equal to 1 (e.g., n is equal to 1, 2, or 3). For details on determining the frequency distribution 146 of background intensities, the standard deviation σ of the frequency distribution 146, and the mean I H,mean the background intensity is based on the description of Fig. 17. In the example of Fig. 22 is the lower reference intensity threshold LL as I H,mean +σ (i.e., n = 1). Particularly in cases of low-contrast candidates K for defects D, n can take a value greater than 1 to determine a suitable value for the lower reference intensity threshold LL.

[0164] In the case of a candidate K for a dark defect D d ( Fig. 23) defines a mean I' GG (similar to the mean I GG in Fig. 22) of in areas of large gradients (similar to the areas of 156 large gradients in Fig. 22) lying intensity values ​​I i an intensity distribution (similar to the intensity distribution 124 in Fig. 22) the lower reference intensity threshold LL'. The upper reference intensity threshold UL' for a candidate K for a dark defect D d is analogous to the lower reference intensity threshold LL for a candidate K for a bright defect D h determined, but using the mean background intensity (similar to the mean I H,mean in Fig. 22) a multiple of a standard deviation σ of a frequency distribution 146 is subtracted from background intensities (and not added as in the case of a bright defect Dh ).

[0165] In a seventh step S307 of the method according to Fig. 21 is used in the case of a candidate K for a bright defect D h ( Fig. 22) an upper intensity threshold SW A based on the upper and lower reference intensity thresholds UL and LL determined in step S306.

[0166] In particular, a difference ΔI between the upper and lower intensity reference threshold UL, LL is determined, as shown in the lower part of Fig. 22. Furthermore, the upper intensity threshold SW A determined in such a way that it is smaller than the upper reference intensity threshold UL by a pre-determined fraction b of the difference ΔI. In Fig. 22, a fraction b of 15% is used merely as an example. In other words, in the example of Fig. 22 the upper intensity threshold SW A as SWA=UL−(b ΔI)=UL−0.15 ΔI calculated.

[0167] However, in other examples, values ​​other than 15% can be used for the fraction b (e.g. any value between 5% and 30%, e.g. also 10% or 20%).

[0168] In the case of a candidate K for a dark defect D d ( Fig. 23) in step S307 a lower intensity threshold SW' A based on the upper and lower reference intensity thresholds UL' and LL' determined in step S306, their difference ΔI' and a pre-determined fraction b': SW'A=LL'+b'ΔI'=LL'+0.15 ΔI.

[0169] That is, the lower intensity threshold SW' A in case of a dark defect D dis determined such that it is greater than the lower reference intensity threshold LL' by the pre-determined fraction b' of the difference ΔI'. Here, too, the fraction b' was set as 15% for the purposes of example, although other values ​​are also possible (see above).

[0170] In an eighth step S308 of the method according to Fig. 21, a pixel-dependent contour C of the candidate K of the defect D depicted in the image 104 is determined based on the determined upper intensity threshold SW A in case of a bright defect D h and based on the determined lower intensity threshold SW' A in case of a dark defect D d determined.

[0171] In particular, in the case of a candidate K for a bright defect D h those pixels 116 of the image 104 whose intensity values ​​I i the upper intensity threshold SW Acorrespond, are determined as pixels of the contour C of the defect candidate K. In order to completely determine the contour C of the defect candidate K, it may also be necessary to adapt a circumferential contour line to the determined contour pixels 116 and thus, for example, also determine subpixel positions of the contour C.

[0172] Furthermore, in the case of a candidate K for a dark defect D d those pixels 116 of the image 104 whose intensity values ​​I i the lower intensity threshold SW' A correspond, are determined as pixels of the contour C of the defect candidate K. Here, too, the adjustment of a circumferential contour line and / or the determination of subpixel positions may be necessary.

[0173] It may happen that in step S308 several candidates K, K1', K2', K3', K4' for defects D, ie several contours C, C1', C2', C3', C4' of defect candidates K, K1', K2', K3', K4', are determined, as in Fig. 24. In this case, the coarse position 130a of the defect candidate K determined in steps S302 and S303, which was determined using the first gradient threshold SW1, can be used. In particular, the one of the plurality of defect candidates K, K1', K2', K3', K4' or the one of the plurality of determined contours C, C1', C2', C3', C4' of defect candidates K, K1', K2', K3', K4' can be selected that is closest to the determined coarse position 130a of the defect candidate K. Reference numeral 130b denotes the fine position of the defect candidate K determined using the second gradient threshold SW2.

[0174] In a ninth step S309 of the method according to Fig. 21 becomes a focus Z ( Fig. 8) of the candidate K of the defect D depicted in the image 104 is determined based on the determined pixel-dependent contour C of the candidate K. The center of gravity Z is determined, for example, as a geometric center of gravity of an area B encompassed by the determined contour C.

[0175] In the procedure according to Fig. 21 can also be a defect probability P D of a detected defect candidate K, as in connection with the Fig. 16 to 19 described.

[0176] In addition, the procedure according to Fig. 21 can also be extended by a manufacturing step of a photomask 102 (similar to that in connection with Fig. 20, step S204). In other words, the method according to Fig. 21 can be extended to a method for producing a photomask 102 from the examined blank 100.

[0177] In Fig. 25 shows a control device 200 for analyzing an image 104 of a blank 100 of a photomask 102 for microlithography. The image 104 ( Fig. 4) has a two-dimensional arrangement 114 of pixels 116 and a two-dimensional intensity distribution 118 depending on the pixels 116. Furthermore, the control device 200 is particularly suitable for carrying out the methods described above according to Fig. 5 and Fig. 21 set up.

[0178] As in Fig. 25, the control device 200 has a first determination device 202 for determining gradient values ​​G of the two-dimensional intensity distribution 118 of the image 104. In addition, the control device 200 has a second determination device 204 for determining a candidate K of a defect D of the blank 100 depicted in the image 104 based on determined gradient values ​​G that are greater than a gradient threshold value SW1, SW2. The control device 200 can optionally also have one or more further determination devices 206 in order to carry out one or more steps of the methods described above according to Fig. 5 and Fig. 21. An example is Fig. 25 shows a third determination device 206 which is used to determine a defect probability P D for a candidate K for a defect D of the blank 100 determined by the second determination device 204.

[0179] Although the present invention has been described using exemplary embodiments, it can be modified in many ways. LIST OF REFERENCE SYMBOLS 100 blanks 102 Photomask 104 Image 106 layer 108 Surface 110 Structure 112 Image section 114 Arrangement 116 pixels 118 Intensity distribution 120 Area of ​​Interest 122 envelope area 124 Intensity distribution 126 Gradient distribution 128 gradient value 130 Place 130a, 130b Place 131 Property 132 lace 134 Intensity distribution 136 share 138 share 140 Place 142 Background area 144 Intensity distribution 146 Frequency distribution 148 Intensity distribution 150 Intensity distribution 152 positions 154 Parameter 156 Area 158 intensity value A artifact a distance a' threshold a S Distance threshold B area B A , B' A , B'' A Area B B , B' B , B'' B Area b, b' fraction C Contour C1'-C4' contour C A , C' A , C'' A contour C B , C' B , C'' B contour D, D' Defect D h , D d defect ΔI, ΔI' difference E Level F Area F1-F3 area F E , F' E Area F F , F' F Area G gradient value G max , G' max Maximum value G mean , G' mean mean G i Gradient value I Intensity I E , I' E Intensity distance I F , I' F Intensity distance I GG mean I H , I' H Background intensity I H,max Maximum value I H,mean , I' H,mean mean I H,min Minimum value I i Intensity value K, K' candidate K1-K3 candidate K1'-K4' candidate K A , K' A , K'' A candidate K B , K' B , K'' B candidate LL, LL' threshold P D , P' D probability S101-S103 Process steps S102a, S102b Process steps S201-S204 Process steps S301-S309 Process steps SW1 threshold SW2, SW2' threshold SW A 'SW' A Threshold SW P Threshold σ, σ' standard deviation UL, UL' threshold x direction X1 - x7 Position x E1 , x E2 position x' E1 , x' E2 position x F1 , x F2 position x' F1 , x' F2 position y direction z direction Z Focus Z A , Z' A , Z'' A Focus Z B , Z' B , Z'' B Focus

Claims

[1] A method for analyzing an image (104) of a blank (100) of a photomask (102) for microlithography, the image (104) having a two-dimensional arrangement (114) of pixels (116) and a two-dimensional intensity distribution (118) as a function of the pixels (116), comprising the steps of: a) determining (S101) gradient values ​​(G) of the two-dimensional intensity distribution (118) of the image (104), and b) Determining (S102) a candidate (K) of a defect (D) of the blank (100) depicted in the image (104) based on determined gradient values ​​(G) that are greater than a gradient threshold value (SW1, SW2). [2] The method according to claim 1, wherein in step b) a pixel-dependent location (130) and / or a pixel-dependent property (131) of the candidate (K) of the defect (D) depicted in the image (104) is determined, the pixel-dependent location (130) of the candidate (K) of the defect (D) depicted in the image (104) has a center of gravity (Z), a contour (C) and / or a region (B) of the candidate (K), and / or the pixel-dependent property (131) of the candidate (K) of the defect (D) depicted in the image (104) has an area (F) and / or a geometric shape (Q) of the candidate (K). [3] Method according to claim 1 or 2, wherein the gradient threshold value (SW1) is 70% or more, 80% or more and / or 90% or more of a maximum value (G max ) of the determined gradient values ​​(G). [4] Method according to one of claims 1 to 3, wherein in step b) several candidates (K, K1, K2, K3) for a corresponding number of defects (D) of the blank (100) depicted in the image (104) and their respective pixel-dependent area (F, F1, F2, F3) are determined, and that of the plurality of determined candidates (K) is determined as a real defect (D) of the blank (100) depicted in the image (104) which has the largest area (F) among the plurality of candidates (K, K1, K2, K3). [5] Method according to one of claims 1 to 4, wherein a gradient mean (G mean ) and a gradient maximum value (G max ) of the determined gradient values ​​(G), and the gradient threshold (SW2) is greater than the gradient mean (G mean ) and smaller than the gradient maximum value (G max ), and / or the gradient threshold (SW2) is an average of the gradient mean (G mean ) and the gradient maximum value (G max ) is. [6] Method according to claims 2, 3 and 5 or according to claims 2 to 5, wherein a first gradient threshold value (SW1) is determined according to claim 3, a second gradient threshold value (SW2) is determined according to claim 5, a coarse pixel-dependent location (130a) of the candidate (K) of the defect (D) depicted in the image (104) is determined based on the first gradient threshold (SW1), and a refined pixel-dependent location (130b) of the candidate (K) of the defect (D) depicted in the image (104) is determined based on the determined coarse pixel-dependent location (130a) of the candidate (K) and the second gradient threshold (SW2). [7] Method according to one of claims 2 to 6, comprising: Determining (S301) a characteristic (154) of a background intensity (144, I H ) of the two-dimensional intensity distribution (118) of the image (104), wherein the parameter (154) comprises a background frequency distribution (146), a background mean value (I H,mean), a background median value, a background maximum value (I H,max ) and / or a background minimum value (I H,min ) of intensity values ​​(I i ) of a background area (142) of the image (104), Determining (S304) that the candidate (K) of the defect (D) depicted in the image (104) corresponds to a candidate (K) for a bright defect (D h ) if one or more intensity values ​​(I i ) is greater than the parameter (154) for the intensity values ​​(I i ) of the background area (142), Determining (S304) that the candidate (K) of the defect (D) depicted in the image (104) corresponds to a candidate (K) for a dark defect (D d ) if one or more intensity values ​​(I i ) is smaller than the parameter (154) for the intensity values ​​(I i) of the background area (142), Determining (S307), for an area of ​​interest (120) of the image (104) and / or for an enveloping area (122) of the determined candidate (K) of the defect (D) depicted in the image (104) h , D d ), an upper intensity threshold (SW A ) of the two-dimensional intensity distribution (118) in the case of a candidate (K) of a bright defect (D h ) and a lower intensity threshold (SW' A ) of the two-dimensional intensity distribution (118) in the case of a candidate (K) of a dark defect (D d ), Determining (308) a pixel-dependent contour (C) of the candidate (K) of the defect (D) depicted in the image (104) based on the determined upper intensity threshold value (SW A ) in case of a bright defect (D h ) and based on the determined lower intensity threshold (SW' A ) in case of a dark defect (Dd ), and / or Determining (S309) a center of gravity (Z) of the candidate (K) of the defect (D, D) depicted in the image (104) h , D d ) based on the determined pixel-dependent contour (C) of the candidate (K). [8] Method according to claim 7, wherein an upper and lower reference intensity threshold (UL, UL', LL, LL') of the area of ​​interest (120) and the envelope area (122) are determined (S306), a difference (ΔI, ΔI') between the upper and lower intensity reference threshold (UL, UL', LL, LL') is determined, the upper intensity threshold (SW A ) is determined such that it is smaller than the upper reference intensity threshold (UL) by a pre-determined fraction (b) of the difference (ΔI), and the lower intensity threshold (SW' A) is determined such that it is greater than the lower reference intensity threshold value (LL') by a pre-determined fraction (b') of the difference amount (ΔI'). [9] Method according to claim 8, wherein the upper reference intensity threshold (UL) corresponds to a maximum intensity value of the area of ​​interest (120) or the envelope area (122), and the lower reference intensity threshold (LL) corresponds to a minimum intensity value of the area of ​​interest (120) or the envelope area (122), or the gradient threshold (SW2) is greater than a gradient mean (G mean ) of the determined gradient values ​​(G) and less than a gradient maximum value (G max ) of the determined gradient values ​​(G), and / or the gradient threshold value (SW2) is an average of the gradient average value (G mean ) and the gradient maximum value (G max ), and large gradient areas (156) in the image (104) are determined based on determined gradient values ​​(G) that are greater than the gradient threshold value (SW2), the upper reference intensity threshold (UL) in the case of a candidate (K) for a bright defect (D h ) based on a mean value (I GG ) of intensity values ​​(I i ) of the large gradient region (156) are determined, the lower reference intensity threshold (LL') in the case of a candidate (K) for a dark defect (D d ) based on a mean value (I' GG ) of intensity values ​​(I i ) of the large gradient region (156) are determined, and the lower reference intensity threshold (LL) in the case of a candidate (K) for a bright defect (D h ) and the upper reference intensity threshold (UL') in case of a candidate (K) for a dark defect (D d) based on the parameter (154) of the background intensity (144, I H ) can be determined. [10] Method according to one of claims 1 to 9, wherein with the procedure a first and a second candidate (K A , K B ) for a defect (D) depicted in the image (104), a pixel-dependent first contour (C A ) of the first candidate (K A ), which has a first pixel-dependent area (B A ), a first pixel-dependent center of gravity (Z A ) of the first candidate (K A ), a pixel-dependent second contour (C B ) of the second candidate (K g ), which has a second pixel-dependent area (B B ), and a second pixel-dependent center of gravity (Z B ) of the second candidate (K B ) and the procedure has: Determine whether the first and second area (BA , B B ) cut, the first area (B' A ) a subset of the second area (B' B ) and / or a minimum distance (a) between the first region (B'' A ) and the second area (B'' B ) is smaller than a distance threshold (a S ) is, Determine when the first and second area (B A , B B ) cut, the first area (B' A ) a subset of the second area (B' B ) and / or a minimum distance (a) between the first and second region (B'' A , B'' B ) is smaller than the distance threshold (a S ) is that the first and second area (B A , B B , B' A , B' B , B'' A , B'' B ) of the first and second candidate (K A , K B , K' A , K' B , K'' A , K'' B) are regions of a single defect (D) depicted in the image (104), and Determining a common center of gravity (Z G , Z' G , Z'' G ) of the first and second area (B A , B B , B' A , B' B , B'' A , B'' B ) based on the first and second pixel-dependent center of gravity (Z A , Z B , Z' A , Z' B , Z'' A , Z'' B ) of the first and second candidate (K A , K B , K' A , K' B , K'' A , K'' B ). [11] Method according to one of claims 1 to 10, wherein the image (104) of the blank (100) of the photomask (102) is recorded using an illuminating light with a wavelength in the UV range, DUV range, less than 300 nm, less than 250 nm, less than 200 nm and / or less than 100 nm, and / or the method comprises a step of recording one or more images (104) of the blank (100) of the photomask (102), wherein the blank (100) is illuminated with an illumination light having a wavelength in the UV range, DUV range, less than 300 nm, less than 250 nm, less than 200 nm and / or less than 100 nm, an illumination light reflected back from the blank (100) is detected, and the two-dimensional intensity distribution (118) of the image (104) is based on the reflected back illumination light. [12] Method according to one of claims 1 to 11, comprising: Determining (S103) a defect probability (P D ) for a determined candidate (K) of a defect (D) depicted in the image (104), wherein a first pixel-dependent area (F E ) of the candidate (K) based on a first intensity distance (I E ) to a background mean (I H,mean ) a background intensity (I H) of the image (104) is determined, a second pixel-dependent area (F F ) of the candidate (K) based on a second intensity distance (I F ) to the background mean (I H,mean ) is determined, the first intensity distance (I E ) is greater than the second intensity distance (I E ), and the defect probability (P D ) based on a quotient of the first pixel-dependent area (F E ) and the second pixel-dependent area (F F ) is determined. [13] Method according to claim 12, comprising: Determining a frequency distribution (146) and a standard deviation (σ) of the frequency distribution (146) for intensity values ​​(I i ) of a background area (142) of the image (104), and Determining the first and second intensity distance (I E , I F) based on the standard deviation (σ) of the frequency distribution (146). [14] Method according to claim 12 or 13, comprising: Determining that the identified candidate (K) corresponds to a real defect (D) depicted in the image (104) if the defect probability (P D ) is greater than a pre-determined probability threshold (SW P ), and Determine that the identified candidate (K) is an artifact if the defect probability (P D ) is equal to or less than the pre-determined probability threshold (SW P ) is. [15] Method according to one of claims 2 to 14, wherein the method produces a photomask (102) for microlithography, and the method according to step b) comprises: Arranging (S204) microstructures (110) on the blank (100) based on the determined pixel-dependent location (130) of the candidate (K) of the defect (D) depicted in the image (104). [16] Control device (200) for analyzing an image (104) of a blank (100) of a photomask (102) for microlithography, wherein the image (104) has a two-dimensional arrangement (114) of pixels (116) and a two-dimensional intensity distribution (118) depending on the pixels (116), and wherein the control device comprises: a first determination device (202) for determining gradient values ​​(G) of the two-dimensional intensity distribution (118) of the image (104), and a second determination device (204) for determining a candidate (K) of a defect (D) of the blank (100) depicted in the image (104) based on gradient values ​​(G) that are greater than a gradient threshold value (SW1, SW2).

Images