Method for inspecting microlithographic photomasks, computer program product, system for inspecting a microlithographic photomask, method for repairing a microlithographic photomask, and method for microlithography
Patent Information
- Authority / Receiving Office
- DE · DE
- Patent Type
- Patents
- Current Assignee / Owner
- CARL ZEISS SMT GMBH
- Filing Date
- 2024-10-30
- Publication Date
- 2026-07-23
AI Technical Summary
Existing defect detection methods for microlithographic photomasks have high sensitivity but low specificity, leading to a high false positive rate, which burdens human inspectors with numerous non-existent defect candidates, increasing time and cost.
A method involving capturing overlapping image lines of a photomask structure, applying defect detection methods to each, and comparing the results to verify actual defects, using calibration information and reliability assessment to reduce false positives.
Enhances defect detection reliability by verifying candidate defects through redundant image information, reducing human inspection workload and improving accuracy.
Smart Images
Figure 00000000_0000_ABST
Abstract
Description
[0001] The invention relates to a method for inspecting a microlithographic photomask. The invention also relates to a computer program product, a system for inspecting a microlithographic photomask, a method for repairing a microlithographic photomask, and a method for microlithography.
[0002] Photomasks are used in microlithographic projection exposure systems to produce components such as integrated circuits with extremely small structures. The photomask, illuminated with very short-wavelength, extreme ultraviolet (EUV) radiation, is projected onto a lithography object to transfer the mask structure.
[0003] For high-quality images produced on lithographic objects, the photomask must be dimensionally accurate and free from structural defects. It is known to inspect photomasks before use in a microlithographic projection exposure system or during downtime. For this purpose, a so-called aerial image of a section of the photomask is generated using a mask inspection device. In this image, the photomask is projected not onto a lithographic object, but onto the image sensor of an EUV camera. Based on the image on the sensor, a defect detection method can be used to assess whether the photomask is free of defects. Defect detection methods are known in the prior art.
[0004] A defect might consist, for example, of a structural component of the photomask or the distance between two structural components being insufficiently wide. A structural component might also have a defect not intended in the photomask's design or an unwanted "bridge" to an adjacent structural component. If a mask containing a defect is exposed, the defect would appear as an irregular shadow or bright area on the photosensitive layer and would be present in the final manufactured component after subsequent processing steps. Various repair methods are known in the art for eliminating these defects or at least reducing them to a negligible level. These methods typically involve removing, adding, and / or structurally modifying material from the photomask in a repair step.This can be achieved, for example, by using an electron beam or an ion beam on the photomask. Longer, straight defects can also be removed by mechanical means (scratching). The use of a laser is also possible, but this is usually not precise enough.
[0005] An undetected defect on a photomask can, if left unrepaired, render an entire batch of manufactured components unusable. It is therefore desirable that as few defects as possible go undetected during the inspection of microlithographic photomasks, so that all existing defects can be repaired.
[0006] Defect detection methods can identify defect information that includes one or more candidate defects. A candidate defect is a detected potential defect in the photomask structure. Defect detection methods used during mask inspection typically have high sensitivity and low specificity. Sensitivity is the probability that a detected candidate defect actually is a defect. Specificity is the probability that if no candidate defect is detected, then no defect actually is present. Therefore, defect detection methods are typically designed to detect all actual defects, accepting the possibility of detecting some candidate defects where no actual defect is present. In other words, the defect detection methods used typically have a low false negative rate and a high false positive rate.As explained above, a low false negative rate is desirable to ensure that as few actual defects as possible go undetected. However, a high false positive rate is disadvantageous because a human typically performs the final assessment of the detected defect candidates before the photomask undergoes repair. This could result in the human being being presented with numerous defect candidates where no actual defect exists on the photomask, which is time-consuming and costly.
[0007] The invention is based on the objective of presenting a method and a computer program product for the inspection of microlithographic photomasks, in which the aforementioned disadvantages are reduced. This objective is achieved by the features of the independent claims. Advantageous embodiments are specified in the dependent claims.
[0008] In a method according to the invention for inspecting a microlithographic photomask, the image of a photomask illuminated by the illumination lens is projected by means of the projection lens onto an image sensor of a camera arranged in the image plane of the projection lens using a mask inspection device comprising an illumination lens and a projection lens. The photomask is placed on a positioning system designed to move the photomask. The photomask is moved with the positioning system so that the image sensor captures a first image line corresponding to a first area on the photomask.The photomask is then moved by the positioning system so that the image sensor captures a second image line corresponding to a second area on the photomask. There is an overlap between the first and second areas, so that both the first and second image lines contain image information about a structure of the photomask located within the overlap. A defect detection method is applied to the image information in the first image line to determine a first defect from the overlap. A similar error detection method is applied to the image information in the second image line to determine a second defect from the overlap. The first and second defect information are compared to determine the defect-free status of the structure of the photomask located within the overlap.
[0009] The invention recognizes that image information about the structure of the photomask arranged within the overlap is captured twice in two overlapping image lines, so that information about a defect in the structure of the photomask arranged within the overlap is also present twice with the captured image lines. The invention therefore proposes applying a defect detection method to the image information about the structure of the photomask arranged within the overlap in both image lines in order to determine two independent pieces of defect information about the area of overlap on the photomask, which are then compared with each other.
[0010] In a mask inspection device, the image field corresponding to the area of the image sensor on the surface of the photomask is small relative to the photomask's area. A scanning movement of the positioning system moves the photomask relative to the image sensor during an exposure. This makes it possible to capture an image line extending across the length of the photomask in a single, continuous exposure. To create a complete image of the photomask, several successively captured image lines are typically aligned and stitched together based on the photomask's overlapping structure.
[0011] The positioning system can be configured to shift the photomask in an X and / or Y direction. The first image line can be captured by the positioning system shifting the photomask in the X direction. The second image line can also be captured by the positioning system shifting the photomask in the X direction. Before capturing the second image line, the positioning system can shift the photomask in the Y direction. Multiple image lines can also be captured by shifting the photomask with the positioning system in opposite directions. The X direction can therefore be positive for the first image line and negative for the second image line, or vice versa. The Y-direction shift determines the overlap of the image lines, or rather, the overlap of the areas on the photomask that correspond to the captured image lines.
[0012] The overlap between the first and second areas on the photomask can, in principle, be of any size. The essential requirement for the invention is simply that an overlap exists between the first and second areas. In other words, the overlap size is greater than 0 (0%) and less than or equal to 1 (100%). The overlap size refers to the ratio between the overlap and an image line. In other words, the overlap size indicates the extent to which an image line is encompassed by the overlap. The overlap size can be less than 0.1 (10%), preferably less than 0.05 (5%), and more preferably less than 0.02 (2%). Such small overlap sizes have the advantage that the entire photomask can be captured more quickly. For an overlap size less than 1, the photomask must be shifted in the Y-direction before capturing the second image line.The greater the overlap, the more image lines must be captured to capture the entire photomask, thus requiring more time for the entire process. However, the greater the overlap, the more areas of the photomask are also captured at least twice in a single image line, resulting in double overlap for more areas. For complete double overlap of the photomask, the overlap size can range from 1 / 2 (50%) to 2 / 3 (approximately 66.6%). An overlap size of 1 / 2 has the advantage that each area of the photomask is captured exactly twice in a single image line. It is preferable if the overlap size can be adjusted by the operator, allowing for a balance between speed and accuracy.
[0013] The photomask shifting steps can be repeated multiple times to capture a plurality of first- and second-line image pairs. It is then possible to specify a rate at which the steps of applying the error detection procedure and comparing the error information are performed only for every xth pair of image lines. In other words, at the specified rate, the error-free status is determined only for a subset of the plurality of first- and second-line image pairs. This allows the frequency of error-free status determination to be predetermined. For example, error-free status can be determined only for every second or third pair of captured image lines, halving or thirding the determination effort and thus offering a trade-off between speed and reliability.
[0014] An area preset can be provided, allowing the operator to define a specific area on the photomask that is to be captured at least twice by means of an overlap. This has the advantage that only the specified area, and not the entire photomask, is captured twice, thus offering a balance between time expenditure and accuracy. In one embodiment, the overlap area is fixed based on the structures to be registered.
[0015] The information on the absence of defects determined by the method according to the invention can be compared with the expected absence of defects in the structure of the photomask arranged within the overlap. Based on such a comparison, calibration information for the mask inspection device can be determined. It is also possible to determine calibration information for a defect detection method based on this comparison. The expected absence of defects (so-called "ground truth") can result from the availability of a reference photomask that has been pre-prepared with structural defects (so-called "defect programmed mask," DPM). In other words, the method according to the invention can be carried out with the reference photomask and the result compared with the ground truth. The reference photomask can also be a photomask with pre-confirmed production quality (so-called "pre-tested mask")."Production-grade mask," which can be assumed to be truly defect-free and therefore free of structural flaws. The calibration information can be used to adjust or readjust the mask inspection device and / or a defect detection method to eliminate or at least minimize any deviations from the ground truth. In other words, the calibration information can preferably reduce the discrepancy between the determined information about defect-free status and the expected defect-free status. This has the advantage of simplified calibration based on fewer image lines being captured.
[0016] A defect report can include a defect candidate if a defect detection procedure has detected a potential defect. Defect candidates can be presented to a human observer for final evaluation to verify them as actual defects or reject them as non-actual defects. Verified defect candidates, which should represent actual defects in the photomask structure, can be repaired in a subsequent repair step.
[0017] Due to various random processes, the image information in the captured image lines can be distorted in different ways. For example, the image information can be distorted by superimposed noise. In other words, the image information in the first image line can be superimposed by a first realization of a random process, and in the second image line by a second realization of the same random process. Because of this superimposition of random processes, an error detection method can therefore produce different results for the image information in the first image line and the image information in the second image line. The ratio between the values of the image information and a random process is often called the "signal-to-noise ratio" (SNR). For example, a low SNR can indicate that the image information is very noisy.
[0018] It is also possible that the image information in the first image line and the image information in the second image line differ due to optical effects, which may be based, for example, on optical scattering. These different optical effects can also lead to a defect detection method producing different results for the image information in the first image line and the image information in the second image line.
[0019] A different result means that, for example, a particular error detection method identifies error information with a candidate error in the image information of the first image line, while the same error detection method does not identify any error information with a candidate error in the image information of the second image line.
[0020] The first and second fault detection methods can be the same. However, it is also possible that the first and second fault detection methods are different. The advantages and disadvantages of various specific fault detection methods can then be combined or weighed against each other.
[0021] By comparing the error information determined by an error detection method according to the invention, the reliability of the applied error detection method is increased. Any error in the photomask structure arranged within the overlap can thus be detected correctly with increased reliability. If, for example, the first error information contains a candidate error—that is, if the error detection method has detected an error in the image information about the photomask structure arranged within the overlap in the first image line—the candidate error can be verified using the second error information. For example, it is possible that a candidate error is only verified as an actual error if it is contained in both the first and the second error information.By comparing the first and second error information, a candidate error in the first error information can, in other words, be verified by a candidate error in the second error information. The information about the error-free status of the photomask structure arranged within the overlap can therefore indicate that an error exists if a specific candidate error is present in both the first and second error information ("logical AND").
[0022] Whether a candidate defect in the first error information corresponds to a candidate defect in the second error information can be based on an identity information. This identity information can state that two candidate defects must have the same center or be located within a specific radius on the photomask to be considered identical. The identity information can also state that the shape of two candidate defects must be identical or within a specific shape tolerance. It is preferred if the identity information can be specified by an operator.
[0023] By comparing the specific error information according to the invention, the susceptibility of applied error detection methods to a superimposed random process or optical effects is also reduced. For example, if the image information in the captured image line is superimposed by a random process such as strong noise, it can happen that an error detection method fails to detect an actually existing error despite its high sensitivity. In such a scenario, it is preferred that an error is verified as such even if only one error piece of information contains a specific candidate error.In other words, the information about the error-free nature of the photomask structure arranged within the overlap can indicate that an error exists if a specific error candidate is present in at least one of the first and second error information ("logical OR").
[0024] If none of the error information includes a candidate for an error, the information about the error-free nature of the structure of the photomask arranged within the overlap can state that the structure of the photomask arranged within the overlap is error-free.
[0025] Information regarding the integrity of the photomask structure within the overlap can preferably be output. For example, this information can be communicated or transmitted to a human user or a computer. This information can be displayed on a screen—particularly as text or an image—or output via a loudspeaker. It is also possible to communicate the information—particularly to a computer—in a way that is unreadable to humans. Output in the form of a database entry or a text file (a so-called "log file") is also possible.
[0026] Information regarding the error-free nature of the photomask structure arranged within the overlap can include reliability information. Reliability information indicates how reliable or certain the result is considered to be. For example, a result can be classified as highly reliable if it has been verified against both the first and second error information. In other words, a detected error can be classified as highly reliable if a specific error candidate is present in both the first and second error information. A detected error can be classified as less reliable if a specific error candidate is present in only one of the first and second error information. The reliability information can include a number representing the number of error candidates that verify the corresponding error.However, it is also possible for the reliability information to include the number of non-conforming error reports. The reliability information can also be expressed as a percentage or a decimal. Preferably, the reliability information can be output instead of, or together with, the error-free status information. This has the advantage of providing an assessment of the reliability of the mask inspection. A final human evaluation can then, for example, focus on the detected errors for which a low reliability rating was given. In other words, the human evaluation can then concentrate on detected errors for which the inventive method has determined a less reliable result.
[0027] If the reliability information indicates that a result is considered unreliable, further steps can be taken to improve its reliability. For example, the procedure can be repeated using other defect detection methods that might yield a more reliable result. Settings of the mask inspection device can also be adjusted in response to the reliability information. For instance, the acquisition of the image lines can be performed more slowly, which typically results in a higher and therefore better signal-to-noise ratio (SNR). Similarly, the illumination of the photomask can be increased, which also typically leads to a higher SNR.
[0028] An error detection method can involve the application of a trained algorithm. The algorithm can be a machine learning model. Training, validating, and applying machine learning models are well-established techniques. In particular, models encompassing artificial neural networks, which form the basis of so-called "deep learning," are well-known. The algorithm can be trained using training data. The training data can consist of multiple pairs, each comprising an image line and an error message (so-called "supervised learning"). It is also possible to use only excerpts of image lines in the training data.The image lines in the training data can also include various random processes, so that image information in the image lines is superimposed by one or more random processes, making the trained algorithm less susceptible to certain random processes. The input of the trained algorithm can consist of a single image line or a portion of an image line. The output of the trained algorithm can consist of error information. It is also possible for the trained algorithm to perform the step of comparing the first error information with the second. The output of the trained algorithm can then consist of information about the error-free nature of the photomask structure arranged within the overlap, while the input consists of multiple image lines.
[0029] Based on the associated reliability information, results can be added to the training data. For example, results classified as highly reliable can be added to the training data. This has the advantage that the algorithm can be further trained using results classified as reliable. Future applications of the corresponding fault detection method can thus be further improved without requiring a human to manually select additional training data.
[0030] The photomask can also be shifted by the positioning system so that the image sensor captures a third image line, corresponding to a third area on the photomask. An overlap can then occur between the first and third areas, as well as between the second and third areas. In other words, there is a simultaneous overlap of the first, second, and third areas, i.e., a triple overlap. The first, second, and third image lines then contain image information about a structure of the photomask located within the overlap. In other words, the image information is then present three times. A third error detection method can then also be applied to the third image line to determine a third piece of error information.The first, second, and third error information can then be compared to determine the reliability of the photomask structure arranged within the overlap. The above explanations regarding the examples with two image lines or two error information can, of course, be applied to three image lines or three error information. For example, the reliability information can reach its highest value if all three error information contain a candidate error, meaning the logical AND is confirmed by all three image information. The training data pairs can also comprise three image lines, and so on.
[0031] If the comparison according to the invention is based on three error pieces of information, it is also possible to refine the decision-making process for verifying a candidate error even further. For example, a candidate error can be verified as an actual error if it is present in all three error pieces of information, i.e., if all three error pieces of information satisfy the logical AND condition. It is also possible for a candidate error to be verified as an actual error if it is encompassed by a plurality of the error pieces of information—in the case of a three-fold overlap, by at least two error pieces of information.
[0032] If the overlap size is between 2 / 3 and 3 / 4 (75%), then a complete triple overlap of the photomask can be achieved. In other words, each area of the photomask can then be captured at least three times in one image row. Even with an overlap of more than 1 / 2, partial triple overlap can occur.
[0033] A person skilled in the art is aware that quadruple overlaps and even higher-order overlaps are possible with a correspondingly large overlap size. The embodiments based on double or triple overlaps can therefore also be adapted to quadruple overlaps or even higher-order overlaps. In one embodiment, the photomask is repeatedly and completely captured, thus capturing an even greater number of image lines that are available for the comparison of error information according to the invention.
[0034] If the overlap size is 1 (100%), then a specific area of the photomask is completely and repeatedly captured in image lines. The positioning system then does not shift the photomask in the Y-direction. The captured image lines then differ only due to superimposed random processes and possibly optical effects, for example, due to minimal shifts of components of the mask inspection device or the photomask.
[0035] In one embodiment, several image lines are superimposed and combined (e.g., added or averaged) to create a combined image line. If superimposed random processes across multiple image lines are uncorrelated, meaning that individual realizations of a random process are not correlated, this offers the advantage of an improved signal-to-noise ratio (SNR) for the combined image line. The method according to the invention can then preferably be carried out using combined image lines. This, in turn, has the advantage that error detection methods that are more susceptible to superimposed random processes can be used. Furthermore, the identity information can be adapted to combined image lines. In particular, a combined image line can be based on image lines that were acquired with an overlap of 1.
[0036] The invention can in principle also be applied to photomasks designed to be illuminated with "deep ultraviolet light" (DUV radiation).
[0037] The invention also relates to a computer program product or a set of computer program products comprising program parts which, when loaded into a computer or into interconnected computers connected to a device according to the invention, are designed to carry out the method according to the invention.
[0038] The invention further relates to a method for repairing a microlithographic photomask, in which, in a repair step, material from the photomask is removed, added, and / or structurally modified. Prior to the repair step, information about the defect-free condition of the photomask is determined using a mask inspection method according to the invention. The repair step is then carried out in an area of the photomask with insufficient defect-free condition.
[0039] The invention further relates to a microlithography method in which, using a projection exposure system comprising an illumination lens and a projection lens, the image of a photomask illuminated by the illumination lens is projected by the projection lens onto a substrate coated with a photosensitive layer and arranged in the image plane of the projection lens. Before exposure of the substrate, information about the absence of defects in the photomask is determined by means of a mask inspection method according to the invention. Exposure of the substrate is only carried out if sufficient defect-free properties are found.
[0040] An area may not be sufficiently error-free if one or more errors have been verified.
[0041] The invention further relates to a system comprising a mask inspection device and a computer. The computer is configured to carry out a mask inspection method according to the invention.
[0042] The disclosure includes further developments of the method with features that are described in the context of the system according to the invention.
[0043] The invention is described below by way of example with reference to the accompanying drawings and advantageous embodiments. The drawings show: Fig. 1: a schematic representation of a mask inspection device; Fig. 2: a schematic representation of a capture of two image lines on a photomask according to the invention; Fig. 3: a schematic representation of a double overlap according to the invention; Fig. 4: a schematic representation of a triple overlap according to the invention.
[0044] With a Fig. Microlithographic photomasks 17 can be examined using the mask inspection device shown in Figure 1.
[0045] Microlithographic photomasks 17 are generally intended for use in a microlithographic projection exposure system (not shown). In the microlithographic projection exposure system, the photomask 17 is illuminated with extreme ultraviolet (EUV) radiation with a wavelength of, for example, 13.5 nm to image a structure formed on the photomask 17 onto the surface of a lithographic object in the form of a wafer. The wafer is coated with a photoresist that reacts to the EUV radiation. The mask inspection device is used to check whether the photomask meets the specifications and is free of contaminants.
[0046] In so-called DUV lithography, corresponding projection exposure systems and mask inspection devices are illuminated with DUV radiation with a wavelength of, for example, 193 nm to 248 nm.
[0047] The mask inspection device is designed according to Fig. 1. The photomask 17 is arranged such that an EUV beam path 15 emanating from an EUV radiation source 14 is directed onto the photomask 17 via an illumination lens 16. The illumination lens 16 shapes the EUV radiation into a beam that illuminates an inspection area on the surface of the photomask 17 with uniform brightness. The illuminated area can, for example, have dimensions of 0.5 mm x 0.8 mm. A field stop is arranged in the illumination lens 16, which limits the illuminated area to the inspection area on the surface of the photomask 17. A positioning system 26 allows the photomask to be moved in the XY plane to bring different inspection areas on the surface of the photomask 17 into the area of the EUV beam path 15.
[0048] The edge lengths of the photomask 17 can, for example, be between 100 mm and 200 mm. The photomask can have an aspect ratio between 1:1 and 1:3, preferably between 1:1 and 1:2, and particularly preferably between 1:1 or 1:2. The photomask can be substantially rectangular. The photomask can preferably be 5 to 7 inches (12.7 cm to 17.8 cm) long and wide, and particularly preferably 6 inches (15.2 cm) long and wide. Alternatively, the photomask can be 5 to 7 inches (12.7 cm to 17.8 cm) long and 10 to 14 inches (25.4 cm to 35.6 cm) wide, and preferably 6 inches (15.2 cm) long and 12 inches (30.5 cm) wide.
[0049] The EUV beam path 15, reflected from the photomask 17, continues via a projection lens 22 to an EUV camera 23, which is equipped with an image sensor 24. The projection lens 22 maps the inspection field on the surface of the photomask 17 onto the image sensor 24 of the EUV camera 23. The EUV radiation source 14, the illumination lens 16, the photomask 17, the projection lens 22, and the EUV camera 23 are arranged in a vacuum housing 40, in which a negative pressure is maintained during operation of the mask inspection device.
[0050] The EUV radiation source 14 is a plasma radiation source in which EUV radiation with a wavelength of 13.5 nm is emitted from a plasma. Tin is a medium suitable for generating a plasma for emitting such EUV radiation. To generate the plasma, a droplet of the medium can be exposed to a laser beam.
[0051] The illumination lens 16 and the projection lens 22 can include mirrors onto which the EUV radiation is reflected. These mirrors can be designed as EUV mirrors, exhibiting particularly high reflectivity for EUV radiation. The optical surface of the EUV mirrors can be formed by a highly reflective coating. This can be a multilayer coating, in particular a multilayer coating with alternating layers of molybdenum and silicon. With such a coating, approximately 70% of the incident EUV radiation can be reflected.
[0052] The projection lens 22 has a magnification factor of more than 100. In order to fully capture the image generated by the field of view on the surface of the photomask 17, the area of the image sensor 24 is larger than the area of the field of view, corresponding to the magnification factor. The image sensor 24 can, for example, have dimensions on the order of 100 mm to 200 mm. The image sensor 24 comprises a multitude of parallel pixel rows that span a pixel array. The image sensor 24 is oriented such that the longitudinal direction of the pixel rows 36 corresponds to the X-direction. When the photomask 17 moves in the X-direction, the image of the photomask 17 on the image sensor 24 moves parallel to the longitudinal direction of the pixel rows, thus enabling the capture of an image row.
[0053] A first image line 101 is thus captured by the positioning system 26 shifting the photomask 17 in the positive X direction. The first image line 101 contains image information about the area on the photomask 17 illuminated during the shift. A second image line 102 is also captured by the positioning system 26 shifting the photomask in the X direction, in this case, however, in the negative X direction. The second image line 102 also contains image information about the area on the photomask 17 illuminated during the shift. Before the second image line 102 is captured, the positioning system 26 shifts the photomask 17 in the positive Y direction. As in Fig. As shown schematically in Figure 2, the photomask 17 is shifted in the Y direction so that there is an overlap between the areas on the photomask 17 that correspond to the first image line 101 and the second image line 102, which is indicated by the dashed lines.
[0054] Fig. Figure 3 shows the first image line 101 and the second image line 102 correspondingly on the photomask 17. There is a double overlap 200 between the two image lines 101 and 102, as this results from the overlap of two image lines. Due to the double overlap 200, the first and second image lines 101 and 102 contain redundant image information about the area on the photomask 17 that corresponds to the overlap. In other words, the same image information is contained once in the first image line 101 and once in the second image line 102.
[0055] In the example from Fig. The overlap size is approximately 0.3. This overlap size was set by the operator of the mask inspection device before image lines 101 and 102 were captured. More precisely, in this example, the operator specified the area of double overlap 200 as a range preset, so that only this specific area, and not the entire photomask, is captured twice.
[0056] An error detection method is applied to the image information in the first image line 101 to determine an initial error. This initial error allows conclusions to be drawn about whether the area on the photomask 17 corresponding to the overlap is free of errors. The same error detection method is applied to the image information in the second image line to determine a second error. This second error also allows conclusions to be drawn about whether the area on the photomask 17 corresponding to the overlap is free of errors.
[0057] The first and second error information are compared to determine the defect-free status of the photomask structure within the overlap. The two error pieces of information are thus combined to obtain further information about defect-free status. In one example, the first error piece contains a candidate defect. This represents a defect detected on photomask 17 by the defect detection method. If the second error piece also contains the candidate defect—that is, if the defect detection method detects the same defect on photomask 17 in both image lines—then the candidate defect is verified as the actual defect. The actual defect is then repaired in a subsequent repair step.If the second error information does not contain the candidate error, then in this example the candidate error is discarded as a non-actual error.
[0058] The image information in the captured image lines is disturbed by superimposed noise. The image information in the first image line 101 therefore differs from the image information in the second image line 102 in that it is disturbed by a different realization of the superimposed noise. Because the first image line 101 is captured by shifting the photomask 17 in the positive X direction and the second image line 102 is captured by shifting the photomask 17 in the negative X direction, the image information in the first image line 101 also differs from the image information in the second image line 102 due to optical effects resulting from minimal displacement of the individual components of the mask inspection device.
[0059] In one example, superimposed noise causes the error detection method to identify a candidate error in the first image line 101, while failing to detect the candidate error in the second image line 102. However, since an error detection method with a low false negative rate was used, the candidate error detected in the first image line 101 can be considered a false positive and is therefore discarded as a non-actual error. This approach makes the error detection method more robust against noise superimposed on the image information. The same applies to any optical effects. The information indicating that the photomask 17 is error-free in the overlapping area is displayed as text on a screen to the operator.This has the advantage that the operator does not have to deal with assessing the erroneously detected error in the first image line 101. In another example, the operator is shown reliability information via a yellow indicator, signaling that an error has been detected in at least one of the two image lines. Alternatively, a green color can be displayed, for example, if no error has been detected in either image line, indicating higher reliability. In this example, the operator then only concerns themselves with the errors displayed in yellow, as these require particular attention and human evaluation.
[0060] Whether the defect candidate in the first error information matches a defect candidate in the second error information is determined using an identity information. In one example, the identity information states that two defect candidates must be located within a specific radius on the photomask to be considered identical. The identity information, or rather the radius, was set by a person operating the mask inspection device before the procedure was carried out.
[0061] In one embodiment, a first error detection method is used for the first image line 101, which is optimized for detecting an image line in the positive X direction. A second error detection method is used for detecting the second image line 102, which is optimized for detecting an image line in the negative X direction. This reduces the influence of certain optical effects.
[0062] In a further embodiment, a third image line 103 is captured by the positioning system 26 shifting the photomask 17 in the positive X direction. The third image line 103 also contains image information about the area on the photomask 17 illuminated during the shift. Before the third image line 103 is captured, the positioning system 26 shifts the photomask 17 in the positive Y direction. The photomask 17 is shifted in the Y direction such that there is an overlap between the areas on the photomask 17 that correspond to the first image line 101, the second image line 102, and the third image line 103, respectively.
[0063] Fig. Figure 4 shows the first image line 101 (dashed line), the second image line 102 (solid line), and the third image line 103 (dash-dotted line) corresponding on photomask 17. There is a triple overlap 300 between the three image lines 101, 102, and 103, as this results from the overlap of three image lines. The overlap size in this example is approximately 0.75. Due to the triple overlap 300, the first, second, and third image lines 101, 102, and 103 contain redundant image information over the area on photomask 17 that corresponds to the overlap. In other words, the same image information is contained once in the first image line 101, once in the second image line 102, and once in the third image line 103.
[0064] As from Fig.As can be seen in Figure 4, in this example, in addition to the triple overlap, there are also 300 areas where only two image lines overlap in a double overlap of 200.
[0065] In one example, the error detection method is also applied to the third image line 103 to determine a third error piece of information. The first, second, and third error pieces of information are compared to determine the accuracy of the photomask structure located within the overlap. The three error pieces of information are thus combined to obtain further information about the accuracy of the structure. In one example, the first error piece of information includes a candidate error. If the second and third error pieces of information also contain the candidate error, meaning the error detection method detects the same error on photomask 17 in all three image lines, then the candidate error is verified as the actual error. The actual error is then repaired in a subsequent repair step.If the second and / or third error information does not contain the candidate error, then in this example, the candidate error is rejected as a non-actual error. In another example, a candidate error is verified or rejected according to the majority principle. Therefore, if at least two of the error information pieces contain the candidate error, then it is verified as an actual error. If only one error information piece contains the candidate error, then it is rejected as a non-actual error.
[0066] In another example, the first image line 101 and the second image line 102 are superimposed and combined by averaging them to create a combined image line. Because the image information in the first and second image lines 101 and 102 is superimposed with uncontrolled noise, the combined image line has an improved signal-to-noise ratio (SNR) compared to the first image line 101 and the second image line 102. The error detection procedure is applied to the combined image line to determine combined error information. In this example, the combined error information is compared with another combined error information, also determined by applying the error detection procedure to an image line obtained by combining two image lines.
[0067] If the information regarding the defect-free status of photomask 17 indicates that the photomask 17 has at least one defect, a repair step is performed to correct the defect on the photomask 17. If the information regarding the defect-free status of photomask 17 indicates that the photomask 17 has no defects, the defect-free photomask 17 is used in a microlithography process to illuminate a substrate.
Claims
[1] Method for inspecting a microlithographic photomask (17) in which, using a mask inspection device comprising an illumination lens (16) and a projection lens (22), the image of a photomask (17) illuminated by means of the illumination lens (16) is projected by means of the projection lens (22) onto an image sensor (24) of a camera (23) arranged in the image plane of the projection lens (22), comprising the steps: a) Placing the photomask (17) on a positioning system (26) designed to move the photomask (17); b) Moving the photomask (17) with the positioning system (26) so that the image sensor (24) captures a first image line (101) corresponding to a first area on the photomask (17); c) Moving the photomask (17) with the positioning system (26) so that the image sensor (24) captures a second image line (102) corresponding to a second area on the photomask (17), wherein there is an overlap (200) between the first area and the second area, such that the first and second image lines (101, 102) include image information about a structure of the photomask (17) arranged within the overlap (200); d) Applying an error detection method to the image information in the first image line (101) to determine an initial error information; e) Applying a fault detection method to the image information in the second image line (102) to determine a second fault information; f) Comparing the first error information with the second error information to determine information about the error-free status of the structure of the photomask (17) arranged within the overlap (200). [2] Method according to claim 1, wherein the first image line (101) is captured by the positioning system (26) moving the photomask (17) in a positive X direction, and wherein the second image line (102) is captured by the positioning system (26) moving the photomask (17) in a positive or negative X direction. [3] Method according to claim 2, wherein, prior to acquiring the second image targets, the positioning system (26) shifts the photomask (17) in a Y direction such that the overlap size is less than 1. [4] Method according to claim 3, wherein the overlap size is less than 0.1, preferably less than 0.05, further preferably less than 0.02, further preferably less than 0.
01. [5] Method according to one of the preceding claims, wherein the overlap (200) corresponds to an area on the photomask (17) which is specified by reference to an area specification. [6] Method according to any of the preceding claims, wherein the specific information about the absence of defects is compared with an expected absence of defects. [7] Method according to claim 6, wherein, by comparing the determined information on the absence of defects with the expected absence of defects, calibration information for the mask inspection device and / or a defect detection method is determined in order to reduce a deviation between the determined information on the absence of defects and the expected absence of defects. [8] Method according to any of the preceding claims, wherein the overlap (200) is specified by means of structures to be registered. [9] Method according to any of the preceding claims, wherein steps b) and c) are repeated to capture a plurality of pairs of first and second image lines (101, 102), and wherein steps d) to f) are performed at a predetermined rate for a subset of the plurality of pairs of first and second image lines (101, 102). [10] A method according to one of the preceding claims, wherein a defect candidate in the first or second defect information is verified by a defect candidate in the other respective defect information, such that the information on the defect-free status of the structure of the photomask (17) arranged within the overlap (200) then indicates that an actual defect exists if a specific defect candidate is present in the first and the second defect information. [11] Method according to any one of claims 1 to 6, wherein the information on the defect-free status of the structure of the photomask (17) arranged within the overlap (200) indicates that an actual defect exists if a specific defect candidate is present in the first or the second defect information. [12] Method according to one of claims 10 or 11, wherein the information on the defect-free nature of the structure of the photomask (17) arranged within the overlap (200) comprises reliability information. [13] Method according to claim 12, wherein the reliability information indicates how many fault information a particular fault candidate is present in. [14] Method according to one of claims 12 or 13, wherein a setting of the mask inspection device is changed in response to the reliability information. [15] Method according to any of the preceding claims, wherein, based on a given identity information, it is determined whether a candidate error in the first error information corresponds to a candidate error in the second error information. [16] Method according to claim 15, wherein the identity information states that two defect candidates are identical if they have the same center or are arranged within a certain radius on the photomask (17) or have the same shape or a shape difference is within a certain shape tolerance. [17] Method according to one of the preceding claims, wherein the information on the error-free nature of the structure of the photomask (17) arranged within the overlap (200) and / or the reliability information is output, preferably acoustically, visually, as a database entry or as a text file. [18] Method according to any of the preceding claims, wherein at least one error detection method comprises the application of a trained machine learning model, wherein the input of the model comprises an image line (101, 102, 103) or a section of an image line and wherein the output of the model comprises error information. [19] Method according to one of the preceding claims, wherein the photomask (17) is moved with the positioning system (26) so that the image sensor (24) detects a third image line (103) corresponding to a third area on the photomask (17), wherein there is an overlap (300) between the first area and the third area as well as between the second area and the third area, such that the first, second and third image lines (101, 102, 103) comprise the image information about the structure of the photomask (17) arranged within the overlap (300), and wherein the third error information is also compared with the first error information and the second error information to determine the information about the defect-free status of the structure of the photomask (17) arranged within the overlap (300). [20] Method according to claim 19, wherein at least one image line is combined from two superimposed image lines. [21] Method according to any of the preceding claims, wherein the photomask (17) is an EUV or DUV mask. [22] A computer program product or set of computer program products comprises program parts which, when loaded into a computer or into interconnected computers, are designed to carry out a method according to any of the preceding claims. [23] System for inspecting a microlithographic photomask (17) comprising a mask inspection device and a computer configured to perform a method according to any one of claims 1 to 21. [24] Method for repairing a microlithographic photomask (17) in which, in a repair step, material of the photomask (17) is removed, added and / or structurally modified, characterized by , that prior to the repair step, information about the defect-free status of the photomask (17) is determined by means of a method according to one of claims 1 to 21. and the repair step is subsequently performed in an area of the photomask (17) with insufficient defect freedom. [25] Method according to claim 24, wherein the information on the defect-free status of the photomask (17) is determined by a method according to claim 9 and the repair step is carried out only if sufficient reliability is found. [26] A microlithography method comprising a projection exposure system comprising an illumination lens (16) and a projection lens (22) in which the image of a photomask (17) illuminated by means of the illumination lens (16) is projected by means of the projection lens (22) onto a substrate coated with a photosensitive layer and arranged in the image plane of the projection lens (22), characterized by , that prior to exposure of the substrate, information on the defect-free status of the photomask (17) is determined by means of a method according to one of claims 1 to 21, and exposure of the substrate is only carried out if sufficient defect-free status and / or reliability is determined.