Method and apparatus for repairing a defect in a lithographic mask

Abstract

Method (1900) for repairing at least one defect (240, 750, 940, 1140, 1240, 1330, 1630, 1840) of a lithographic mask (200, 400, 500, 600, 700, 800, 900, 1100, 1200, 1300, 1600, 1800), the method comprising the step: Determining (1920) parameters of at least one repair method for the at least one defect, wherein determining parameters comprises: assigning at least one numerical value to a parameter, wherein the numerical value differs from the numerical value specified for that parameter by the at least one defect, wherein determining the parameters is based on at least one element of the group: an exposure setting with which the lithographic mask is exposed in operation, design data of the lithographic mask, refractive index data of a deposit for repairing a defect of missing material, and resolution-enhancing RET techniques of the lithographic mask, and where the deviation of the numerical value of this parameter is selected from a range whose lower limit is greater than zero and whose upper limit is less than a resolution limit of a defect-free area of ​​the lithographic mask.

Description

1. Technical field

[0001] The present invention relates to a method and a device for repairing at least one defect in a lithographic mask. Furthermore, the present invention relates to a method and a device for repairing at least one pattern element of a lithographic mask. 2. State of the art

[0002] As a consequence of the increasing integration density in the semiconductor industry, lithographic masks must increasingly image smaller structures onto wafers. One way to address this trend is the use of lithographic or photolithographic masks whose actin wavelength is shifted to ever shorter wavelengths. Currently, ArF (argon fluoride) excimer lasers, which emit at a wavelength of approximately 193 nm, are frequently used as light sources in photolithography. By using masks for dual or multiple exposures, structures with dimensions that cannot be achieved with a single exposure step can be created in a single photoresist.

[0003] Currently, photolithography systems are under development that utilize electromagnetic radiation in the EUV (extreme ultraviolet) wavelength range (preferably in the 10 nm to 15 nm range). These EUV photolithography systems are based on a completely new beam guidance concept that employs reflective optical elements, as no materials are currently available that are optically transparent in the specified EUV range. The technological challenges in developing EUV systems are enormous, and vast development efforts are required to bring these systems to industrial readiness.

[0004] Lithographic masks, photolithographic masks, exposure masks, photomasks, or simply masks play a crucial role in imaging increasingly smaller structures in photoresist deposited on a wafer. With each increase in integration density, it becomes increasingly important to reduce the minimum feature size that exposure masks can image. To reliably image the decreasingly smaller structures of a mask onto a wafer-mounted photoresist, resolution enhancement techniques (RETs), such as optical proximity correction (OPC), are increasingly employed. The following exemplary documents describe various aspects of RET and OPC techniques: W.-M. Gan et al.: “Placement of sub-resolution assist features based on a generic algorithm”, DOI 10.1109 / ACCESS.2019.2926102, IEEE ACCESS; P. Gupta et al.: „Manufacturing-aware design methodology for assist feature correctness“, Design and Process Integration for Microelectronic Manufacturing III, Proc. of SPIE Vol. 4756, Bellingham, WA, 2005, doi: 10.1117 / 12.604872, US 2006 / 0 046 160 A1, US 8 739 080 B1, US 8 498 469 B2, US 2008 1 0 077 907 A1, US 2009 / 0 258 302 A1 und US 10 318 697 B2.

[0005] German patent application DE 100 30 143 A1 describes a photomask, a manufacturing process for a photomask, and a semiconductor device. A corrected irradiation area, which is to be irradiated with laser light under given output conditions to remove an opaque expansion defect, is set such that it has: (1) an irradiation area having the opaque expansion defect and widths w1 and w2, and (2) a pattern repair area having width w2 and extending in the negative direction in a first direction D1 by the absolute value of a preset repair offset Δw from the junction between the opaque expansion defect and the pattern edge.The correction offset Δw is set such that the degree of dimensional variation of the resist pattern to be transferred falls within a range acceptable for device quality. A portion of the pattern edge corresponding to the width |Δw| is missing after laser irradiation. Therefore, if the design pattern dimension is reduced to approximately 2 µm, for example, as a dimensional value on the photomask, it is possible to mitigate the detrimental effect of resist pattern dimensional variation on device quality, which is caused by a reduction in the transmission factor in the repaired defect area.

[0006] The German patent application DE 10 2017 205 629 A1 relates to a method and a device for repairing at least one defect in a photolithographic mask for the extreme ultraviolet (EUV) wavelength range, wherein the method comprises the steps: (a) determining at least one defect; and (b) determining a repair shape for the at least one defect; (c) wherein the repair shape is diffraction-based to take into account a phase disturbance caused by the at least one defect.

[0007] The patent application DE 10 2017 203 841 A1 relates to a method for determining a repair shape for processing at least one defect of a photolithographic mask, comprising the following steps: (a) determining at least one correction value for the repair shape of the at least one defect, wherein the correction value takes into account a position of at least one pattern element of the photolithographic mask that does not contact the at least one defect; and (b) correcting the repair shape by applying the at least one correction value.

[0008] German patent application DE 10 2004 004 854 A1 describes a method and a device for producing and repairing a photomask. A shaded area with a transmittance in the range of 0% to 2% is formed in the center of a clear defect in a wiring pattern of a halftone mask. Semi-transparent areas with a transmittance of 10% to 25% are formed adjacent to the shaded area in the regions extending from inside the edge of an imaginary pattern without a defect to outside the edge. In this way, the tolerable margin of error for the correction section of the defect in the halftone mask can be increased.

[0009] Patent DE 10 2018 209 562 B3 relates to a device for examining and / or processing a photolithography element with a beam of charged particles, wherein the device comprises: (a) means for recording measurement data while the photolithography element is exposed to the beam of charged particles; and (b) means for predicting a drift of the beam of charged particles relative to the photolithography element using a trained machine learning model and / or a predictive filter, wherein the trained machine learning model and / or the predictive filter use at least the measurement data as input data.

[0010] The manufacturing process of photolithographic masks is becoming increasingly complex, and therefore more time-consuming and ultimately more expensive, due to the combination of increasingly smaller structural elements or pattern elements and the use of RET techniques.

[0011] Due to the minuscule structural sizes of the pattern elements, even the smallest deviations of the pattern elements produced on a wafer from the design specifications manifest themselves during mask manufacturing as visible or printable defects. These must be repaired whenever possible. Because of the small dimensions of the defects visible on a wafer, which can extend down to the low double-digit nanometer range, their detection is very complex. Furthermore, repairing increasingly smaller defects becomes more and more difficult. On the one hand, positioning a repair tool relative to an identified defect is only possible with very complex measurement technology, and on the other hand, adjusting the repair tool to a specific small defect is very time-consuming.

[0012] The present invention therefore addresses the problem of providing methods and devices that improve the repair of small defects in lithographic masks. 3. Summary of the invention

[0013] According to one embodiment of the present invention, this problem is solved by a method according to claim 1 and a device according to claim 15. A further embodiment is described to aid in understanding the invention.

[0014] In one embodiment, the method for repairing at least one defect in a lithographic mask comprises the step of determining parameters of at least one repair shape for the at least one defect, wherein determining parameters includes assigning at least one numerical value to a parameter, the numerical value being different from the numerical value specified for that parameter by the at least one defect. This can be particularly advantageous for small defects, i.e., defects having at least one dimension that is less than ten times the resolution limit of the mask, less than five times the resolution limit, less than three times the resolution limit, or less than the resolution limit itself. It can be especially advantageous if small defects have at least one dimension within a range of 2% to 50% of the resolution limit.

[0015] Typically, a defect to be repaired determines the parameters of the repair tool used to repair the defect. Through extensive analysis, the inventor has discovered that compensating for or repairing particularly small defects that were previously extremely difficult to eliminate can be significantly simplified if at least one of the parameters of the repair tool is assigned a different numerical value than that actually required for repairing the defect. This can be particularly advantageous for small defects, i.e., defects that have at least one dimension within a range of 2% to 50% of the mask's resolution limit.

[0016] In the following, a repair mold whose parameters exclusively have numerical values ​​determined based on measured values ​​will also be referred to as a nominal repair mold. A nominal repair mold is typically formed from the difference between a measured mask section exhibiting a defect and a measured equivalent defect-free mask section. Alternatively and / or additionally, it is also possible to generate a nominal repair mold by subtracting the design data of a measured, defective mask section. A repair mold used in a method according to the invention differs from a nominal repair mold in that, in the former repair mold, at least one parameter has a numerical value that deviates from the numerical value determined from measurement results.

[0017] Put simply, it can be advantageous not to perform a repair (e.g., of an edge defect) solely based on parameters whose numerical values ​​are determined by the specific defect (e.g., in the case of an edge defect, by the difference between the faulty edge position and the target edge position, which is defined by the mask design). This is because, especially when the numerical value of a parameter is close to or even below the resolution limit (e.g., the edge position is only very slightly incorrect), a very precise repair is necessary to correct this (minor) defect. Small errors during the repair process can result in the repair yielding no significant improvement.However, if a targeted deviation is made from a numerical value for the corresponding parameter specified by the respective defect, it is possible to significantly reduce the impact of errors during the repair, thus potentially easing the requirements for the repair process.

[0018] The inventive method is explained below using the example of an edge placement error of a pattern element. A deviation dx of an edge of a pattern element from the specifications of the mask design is transferred by the mask during an exposure process into an edge placement error (EPE). This EPE is determined by the product of the deviation dx, a possible mask error enhancement factor (MEEF) for the defect or deviation, and the magnification or reduction of the projection lens downstream of the mask. A projection lens of a photolithographic exposure system often has a magnification M = ¼ or M = 1 / 5.

[0019] However, if the deviation dx of an edge of a pattern element becomes smaller than the resolution limit of the photolithographic mask, the latter translates the deviation dx into a reduced edge placement error during an exposure process, albeit to a significantly lesser extent. This is due to the averaging of the actinic radiation of a photomask over structures whose dimensions are smaller than the resolving power of the mask. The details of this averaging process depend on the structures under consideration and the specifics of the exposure process used to image these structural elements. The present application takes advantage of this fact to facilitate the repair of defects whose dimensions are below the resolution limit of the lithographic mask.

[0020] For example, when repairing the edge of a pattern element, where the parameters (or the parameters of the corresponding nominal repair function) actually require material to be deposited from the measured edge to the target edge (numerically defined by the defect), material can be deliberately deposited at a distance from the measured edge, potentially over a shorter length. This way, an error in the precise placement of the deposited material (e.g., due to diffraction effects) can have a significantly smaller impact on the quality of the repaired mask. Similarly, this also applies to a defect requiring material etching, where, in the case of an edge defect, material can be etched at a distance from the measured edge.

[0021] A repair form summarizes the instructions executed by a repair tool to eliminate a defect in a photomask. For example, in the case of a defect with excess material, the repair form describes a local etching process that removes the excess material from the mask. A repair form typically has a base area that a particle beam scans in the manner defined by the repair form. This means the repair form specifies the energy of the particles in the particle beam, the spot size at the focus of the particle beam, its dwell time at a point, the spacing between adjacent points of impact of the particle beam on the mask or defect, and the time it takes for the particle beam to return to the starting point. During the execution of the repair form, the area scanned by the particle beam and the numerical values ​​of the parameters specified above can change.In the case of a local etching process, the repair method further specifies the timing of the supply of the etching gas(es), i.e., the repair method controls the gas flow rate of the etching gas(es) during the processing of at least one defect.

[0022] In this application, the term lithographic mask includes a photolithographic mask.

[0023] The at least one parameter can comprise at least one element from the group: at least one lateral dimension of the at least one repair shape, one height dimension of the at least one repair shape, one distance of the at least one repair shape from the at least one defect, one material composition of the at least one defect, one geometric shape of the at least one repair shape, and one environment of the at least one defect on the lithographic mask. The material composition of the at least one defect largely determines its complex refractive index.

[0024] The repair or compensation of small defects can be facilitated or improved through various measures. Firstly, the defect does not need to be repaired exactly as specified by the parameters of a nominal repair mold. For example, the base area of ​​the repair mold can be smaller than the base area of ​​the defect. Secondly, the height of the repair mold can be less than the height of the defect. Thus, the procedure defined above enables defect correction where the material removed from or added to the mold can be less than the volume of the defect. This has a positive effect on defect processing time.

[0025] Secondly, the averaging of actinic radiation over structures smaller than the resolving power, as explained above, allows the repair shape to be placed slightly away from the original defect. This significantly reduces the very high demands on positional accuracy when placing the repair shape, without significantly impairing the quality of the defect repair or compensation.

[0026] Furthermore, the geometric shape of the repair mold can differ from the actual defect shape. This allows for a significant simplification of defect repair, as the shape of the repair mold can be chosen, at least partially, independently of the defect shape and thus with a much simpler geometric form to manufacture.

[0027] At least one parameter can deviate by a predetermined amount from the numerical value specified by the at least one defect for that parameter.

[0028] The deviation of the numerical value of at least one lateral parameter is selected from a range whose lower limit is greater than zero and whose upper limit is smaller than a resolution limit of a defect-free area of ​​the lithographic mask.

[0029] The mask-side resolution limit R is proportional to the actinic wavelength λ and inversely proportional to the mask-side numerical aperture NA. The NA can have numerical values ​​in the range of approximately 0.1 to 0.5. Furthermore, the resolution limit depends on the exposure setting σ of the exposure system that exposes the mask. R=0.5⋅λNA⋅(1+σ), where σ has a numerical value between 0 (for central illumination) and 1 (for maximally oblique illumination). Oblique illumination is also referred to as off-axis illumination in this field.

[0030] Depending on the numerical aperture NA and the exposure setting σ, the resolution limit for photomasks exposed with the deep ultraviolet (DUV) wavelength λ = 193 nm lies between 150 nm ≤ R ≤ 300 nm. For ELTV masks with an actinic wavelength of λ = 13.5 nm, the resolution limit currently lies in the range of 50 nm ≤ R ≤ 100 nm. This means that the range within which a numerical value of one or more parameters of the repair shape can be selected shrinks with decreasing actinic wavelength of the photolithographic mask. In other words, the method defined above opens up a larger range of new degrees of freedom in the DUV wavelength range than in the EUV range.

[0031] The deviation of the numerical value of at least one parameter can comprise a range of 2% to 80%, preferably 2% to 50%, and most preferably 2% to 30% of a resolution limit of the lithographic mask.

[0032] A dimension of at least one dimension of the at least one repair shape can comprise a range of 10% to 90%, preferably 20% to 80%, more preferably 30% to 70%, and most preferably 40% to 60% of a dimension of the corresponding dimension of the at least one defect.

[0033] The dimension of at least one of the repair shapes can include at least one lateral dimension and / or one height of the repair shape.

[0034] The distance of the at least one repair form from the at least one defect can comprise a range of 2% to 80%, preferably 2% to 50%, more preferably 2% to 30%, and most preferably 2% to 10% of a resolution limit of a defect-free area of ​​the lithographic mask.

[0035] The repair shape can exhibit a lateral shift relative to the nominal repair shape, lateral deviations in its dimensions relative to the nominal repair shape, and a deviation in its height relative to the nominal repair shape, such that the repair shape and the nominal repair shape essentially produce the same optical intensity distribution in a photoresist. This means that the lateral dimension(s) and the height of a repair shape can be coupled or correlated. For example, a small height of a repair shape can be compensated for by increasing one or both of its lateral dimensions, and vice versa. Furthermore, a correlation between the height or lateral dimension(s) of a repair shape and a lateral shift of the repair shape relative to a defect is also possible.The details depend on the repair shape, the environment of the mask in which the repair shape is created, and the exposure process.

[0036] The at least one defect can include at least one element from the group: an edge placement error of a pattern element, a broken and / or bridged connection of a pattern element, an outlier of an edge roughness of a pattern element, a particle adhering to the lithographic mask, a low-printing lateral defect, a defect remnant of a performed defect repair, a sidewall angle error of a pattern element, and a centroid error of a pattern element and / or of a distance range between two pattern elements.

[0037] Determining the parameters for at least one repair method may include: capturing at least one aerial image of the at least one defect. An aerial image can be acquired using a mask inspection system. A mask inspection system may include an optical inspection system and / or an inspection system that scans the mask surface. For example, an optical mask inspection system may include a laser interferometer, and a mask surface scanning system may include an atomic force microscope. An optical mask inspection system may be designed to acquire an aerial image and / or an aerial image focus stack of a photomask.

[0038] The acquisition of at least one aerial image can include: acquiring at least one aerial image of at least one defect at an actinic wavelength of the lithographic mask and / or acquiring an aerial image focus stack of the at least one defect. Acquiring the aerial image at the actinic wavelength reveals the details in the aerial image that will be imaged into the photoresist during a subsequent exposure of the photomask. Therefore, acquiring an aerial image at its actinic wavelength is advantageous. It is even more beneficial to determine the imaging behavior of a defective area of ​​a photolithographic mask by tuning it through its focus.

[0039] Determining the parameters for at least one repair method may include: scanning the at least one defect using a scanning particle microscope and / or a scanning probe microscope. Furthermore, determining the parameters for at least one repair method may include: taking at least one aerial photograph of the at least one defect and scanning the at least one defect using a scanning particle microscope and / or a scanning probe microscope.

[0040] A scanning particle microscope can include at least one element from the following group: a scanning electron microscope (SEM), a focused ion beam microscope (FIB), and a scanning electron microscope with polarization analysis (SEMPA).

[0041] A scanning probe microscope can include at least one element from the following group: a scanning tunneling microscope (STM), an atomic force microscope (AFM), a magnetic force microscope (MFM), a scanning near-field optical microscope (SNOM), and a scanning near-field acoustic microscope (SNAM).

[0042] Determining the parameters of at least one repair method is additionally based on at least one element from the group: an exposure setting with which the lithographic mask is exposed during operation, design data of the lithographic mask, refractive index data of a deposit for repairing a defect of missing material, and resolution-enhancing RET techniques of the lithographic mask.

[0043] Determining the parameters of at least one repair method may include: applying at least one algorithm to measurement data of the at least one defect and design data of the lithographic mask.

[0044] The at least one algorithm can be implemented in hardware, software, firmware, or a combination thereof. Furthermore, the at least one algorithm can be stored in non-volatile memory. In particular, the at least one algorithm can be stored in solid-state storage (SSD, Solid State Drive).

[0045] Assigning the numerical value may involve: applying a trained machine learning model to determine the deviation of the numerical value of the at least one parameter from the numerical value specified by the at least one defect for that parameter.

[0046] The machine learning model can include a transformation model that has at least two transformation blocks, where each of the at least two transformation blocks includes at least one generically learnable function that transforms inputs into outputs that are used as inputs for a subsequent transformation block. The machine learning model can include at least one element from the following group: a parametric mapping, an artificial neural network, a deep neural network, a time-delayed neural network, a convolutional neural network, a recurrent neural network, a long short-term memory network, a generative model, a kernel density estimator, a statistical model, a decision tree, a linear model, and a time-invariant model.

[0047] The machine learning model may include: (a) at least one encoder block for determining information-carrying features of an image of the at least one defect and the design data associated with the image of the at least one defect; and (b) at least one decoder block for generating at least one effect of the at least one defect from the determined information-carrying features, wherein the at least one effect of the at least one defect shows what a superimposition of the image of the at least one defect with the corresponding design data looks like.

[0048] Determining the parameters of at least one repair method may involve: applying a trained machine learning model to determine the parameters of at least one repair method.

[0049] A trained machine learning model can be used in at least two embodiments of a method described above. Firstly, a suitably trained machine learning model can be used to assign a numerical value or values ​​to one or more parameters of a repair pattern that differ from the value or values ​​determined by the at least one defect. However, it is also possible – and this is the currently preferred embodiment – ​​for a suitably trained machine learning model to determine all parameters of the at least one repair pattern for repairing the at least one defect based on measurement data, for example, one or more aerial images of the defect, design data of the mask, and, if applicable, structures of one or more RET techniques.

[0050] The method according to the invention can further comprise the step of generating at least one repair element on the lithographic mask using the determined repair shape. The execution of the repair shape creates a repair element on the photomask that is designed to eliminate the at least one defect as far as possible, i.e., to repair or compensate for it.

[0051] At least one of the repair elements cannot be imaged during the exposure of the photolithographic mask. At least one of the repair elements can alter the imaging behavior of at least one defect during the exposure of the lithographic mask.

[0052] A generated repair element typically exhibits deviations from a nominal repair shape that are below the resolution limit of the photolithographic mask. Therefore, when the mask is exposed with the actinic wavelength, it cannot be imaged onto a photoresist and thus onto the underlying wafer. However, the repair element generated on the mask is designed to modify the imaging behavior of the defective area of ​​the photomask. In combination with the repair element, the defective area produces an imaging behavior very similar to that of a defect-free area with the same pattern arrangement. As a result, the repaired mask essentially produces the same edge position as a defect-free mask during an exposure process. The effect of at least one repair element is based, at least in part, on diffraction effects of the actinic exposure radiation at the repair element.

[0053] The creation of at least one repair element can include: performing at least one local etching process and / or performing at least one local deposition process using at least one focused particle beam and at least one precursor gas.

[0054] The at least one focused particle beam can include at least one element from the group: a photon beam, an electron beam, an ion beam, an atom beam, and a molecular beam.

[0055] The at least one precursor gas can include at least one element from the group: an etching gas, a separating gas, and an additive gas.

[0056] The corrosive gas can include at least one element from the following group: a halogen (F2, Cl2, Br2, I2), oxygen (O2), ozone (O3), hydrochloric acid (HCl), hydrogen fluoride (HF), xenon difluoride (XeF2), xenon tetrafluoride (XeF4), xenon hexafluoride (XeF6), xenon chloride (XeCl), argon fluoride (ArF), krypton fluoride (KrF), sulfur difluoride (SF2), sulfur tetrafluoride (SF4), sulfur hexafluoride (SF6), nitrosyl chloride (NOCl), phosphorus trichloride (PCl3), phosphorus pentachloride (PCl5), phosphorus trifluoride (PF3), nitrogen trifluoride (NF3), water vapor (H2O), hydrogen peroxide (H2O2), nitrous oxide (N2O), nitrogen oxide (NO), nitrogen dioxide (NO2), and nitric acid (HNO3).

[0057] The at least one separation gas can comprise at least one element from the group: a metal alkyl, a transition element alkyl, a main group alkyl, a metal carbonyl, a transition element carbonyl, a main group carbonyl, a metal alkoxide, a transition element alkoxide, a main group alkoxide, a metal complex, a transition element complex, a main group complex, and an organic compound.

[0058] The metal alkyl, the transition element alkyl and the main group alkyl can include at least one element from the group: Cyclopentadienyl (Cp) Trimethylplatinum (CpPtMe3), Methylcyclopentadienyl (MeCp) Trimethylplatinum (MeCpPtMe3), Tetramethyltin (SnMe4), Trimethylgallium (GaMe3), Ferrocene (Co2Fe), and Bis-aryl chromium (Ar2Cr).

[0059] The metal carbonyl, the transition element carbonyl, and the main group carbonyl can comprise at least one element from the group: chromium hexacarbonyl (Cr(CO)6), molybdenum hexacarbonyl (Mo(CO)6), tungsten hexacarbonyl (W(CO)6), dicobalt octacarbonyl (Co2(CO)8), triruthenium dodecadecarbonyl (Ru3(CO)6)6 12 ), and iron pentacarbonyl (Fe(CO)5).

[0060] The metal alkoxide, the transition element alkoxide, and the main group alkoxide can comprise at least one element from the following group: tetraethyl orthosilicate (TEOS, Si(OC2H5)4) and tetraisopropoxytitanium (Ti(OC3H7)4). The metal halide, the transition element halide, and the main group halide can comprise at least one element from the following group: tungsten hexafluoride (WF6), tungsten hexachloride (WCl6), titanium hexachloride (TiCl6), boron trichloride (BCl3), and silicon tetrachloride (SiCl4).

[0061] The metal complex, the transition element complex and the main group complex can include at least one element from the group: copper bis-hexafluoroacetylacetonate (Cu(C5F6HO2)2) and dimethyl gold trifluoroacetylacetonate (Me2Au(C5F3H4O2)).

[0062] The organic compound can include at least one element from the following group: carbon monoxide (CO), carbon dioxide (CO2), an aliphatic hydrocarbon, an aromatic hydrocarbon, a component of vacuum pump oils, and a volatile organic compound. An aromatic hydrocarbon can include styrene.

[0063] The at least one additive gas can include at least one element from the group: an oxidizing agent, a halide, and a reducing agent.

[0064] The oxidizing agent can comprise at least one element from the following group: oxygen (O₂), ozone (O₃), water vapor (H₂O), hydrogen peroxide (H₂O₂), nitrous oxide (N₂O), nitrogen oxide (NO), nitrogen dioxide (NO₂), and nitric acid (HNO₃). The halide can comprise at least one element from the following group: chlorine (Cl₂), hydrochloric acid (HCl), xenon difluoride (XeF₂), hydrogen fluoride (HF), iodine (I₂), hydrogen iodide (HI), bromine (Br₂), hydrogen bromide (HBr), nitrosyl chloride (NOCl), phosphorus trichloride (PCl₃), phosphorus pentachloride (PCl₅), and phosphorus trifluoride (PF₃). The reducing agent can comprise at least one element from the following group: hydrogen (H₂), ammonia (NH₃), and methane (CH₄).

[0065] The at least one generated repair element can overlap at least partially with the at least one defect. A deposited repair element can comprise a material of the lithographic mask. The deposited repair element can comprise: a metal, such as chromium (Cr), a metal compound, such as tantalum nitride (TaN), silicon (Si), silicon dioxide (SiO2), and molybdenum silicon oxynitride (Mo3). x SiO y N z ), where 0 < x ≤ 0.5, 0 ≤ y ≤ 2, and 0 ≤ z ≤ 4 / 3. An etched repair element can etch a material of the photolithographic mask. The etched repair element can include the mask materials listed above.

[0066] The procedure defined above may comprise the steps of: (a) generating at least one repair element by means of at least one repair shape for which the parameters are determined by the at least one defect; and (b) determining parameters of a repair shape for a remaining defect residue, wherein determining parameters for the repair shape for the remaining defect residue comprises: assigning at least one numerical value to a parameter that differs from the numerical value specified for that parameter by the remaining defect residue.

[0067] The method defined above can be used as a second stage of a general defect repair process. In a first stage, a large defect, i.e., a defect that is large compared to the resolution limit of the photomask, can be repaired by performing a repair in the form of a local etching process or a local deposition process, or by creating the corresponding repair element. The repaired mask is then inspected. If, upon inspection of the mask, it is found that a repaired area still does not meet the specification, a repair incorporation is determined for the remaining defect, the parameters of which are determined according to the method according to the invention. It is assumed that the remaining defect is a small defect, i.e.,A defect is defined as having at least one dimension smaller than the resolving power or resolution limit of the photolithographic mask. The remaining defect can then be repaired or compensated for by creating the corresponding repair element based on the repair shape determined in the second step.

[0068] In a second embodiment, which supports the understanding of the present application, the method for repairing at least one defective pattern element of a lithographic mask comprises the steps of: (a) determining at least one repair element of the lithographic mask that does not image the lithographic mask upon exposure, wherein the at least one repair element is configured to modify the imaging behavior of the at least one defective pattern element; and (b) generating the at least one repair element on the lithographic mask using at least one focused particle beam and at least one precursor gas.

[0069] The generated repair element can have a dimension smaller than the resolution limit R of the photomask in at least one dimension. As explained above, averaging the actinic radiation over structures with dimensions below the resolving power of the mask reduces the impact of a repair element placement error. This significantly simplifies the positioning of the repair element(s) relative to the defect to be repaired. Furthermore, by not imaging the repair element(s), their geometric shape(s) can deviate significantly from the shape of the defect without negatively impacting the defect compensation. This greatly simplifies the repair or compensation of small defects.

[0070] However, the repair element(s) created on the mask locally alter the diffraction behavior of the mask at the actinic wavelength. The repair element(s) are designed such that, in combination with the defective pattern element, they essentially realize the imaging behavior of a corresponding defect-free area of ​​the photolithographic mask.

[0071] The at least one repair element can have at least one dimension that covers a range of 10% to 90%, preferably 20% to 80%, more preferably 30% to 70%, and most preferably 40% to 60% of a resolution limit of the lithographic mask.

[0072] The distance between the at least one repair element and the at least one defective pattern element can comprise a range of 2% to 80%, preferably 2% to 50%, more preferably 2% to 30%, and most preferably 2% to 10% of the resolution limit of the lithographic mask.

[0073] At least one dimension of the at least one repair element can comprise a range of 10% to 90%, preferably 20% to 80%, more preferably 30% to 70%, and most preferably 40% to 60% of the resolution limit of the lithographic mask.

[0074] A computer program may contain instructions which, when executed by a computer system, cause the computer system to perform the procedural steps of one of the aspects listed above.

[0075] In one embodiment, the device for repairing at least one defect of a lithographic mask has means for determining parameters of at least one repair form for the at least one defect, wherein the means for determining parameters comprises: means for assigning a numerical value to at least one parameter that differs from the numerical value specified for that parameter by the at least one defect.

[0076] The device may further comprise means for generating at least one repair element on the lithographic mask using the determined repair shape.

[0077] The means for determining the parameters of the at least one repair shape can include at least one co-processor configured to determine the parameters of the at least one repair shape from measurement data of the at least one defect and design data of the lithographic mask. Furthermore, the at least one co-processor can be configured to assign at least one numerical value to the at least one corresponding parameter, which differs from the numerical value specified by the at least one defect. The assignment of the differing numerical value to the at least one corresponding parameter can be based on the resolution limit of the lithographic mask and the resolution limit during the generation of the at least one repair element.

[0078] The resolution limit for generating at least one repair element is significantly influenced by two parameters. The first parameter is the minimum spot diameter to which a particle beam can be focused to generate a repair element. For a photon beam, the achievable spot diameter is determined by the wavelength of the photons. To generate a repair element on a mask for the DUV wavelength range using a photon beam, photons from the EUV wavelength range are necessary. EUV photon sources are currently still very expensive. It is therefore advantageous to use a massive particle beam, such as an electron beam, to generate a repair element, as its resolution limit is given by the de Broglie wavelength. Electron beams can currently be focused to a spot diameter in the range of a few nanometers.The positioning accuracy of an electron beam is significantly higher and extends down to the sub-nanometer range.

[0079] The second parameter that determines the resolution limit when creating a repair element is the interaction region, or scattering region, of the secondary electrons generated by a massive particle beam. The diameter of this interaction region on the mask surface determines the extent of the local chemical reaction initiated by the particle beam and the at least one precursor gas. The size of the interaction region depends on the energy of the particles impacting the photomask. Furthermore, the local material composition of the photomask at the interaction site has a significant influence on the size of the interaction region. Currently, local chemical reactions can be confined to lateral dimensions of approximately 5 nm.

[0080] The means for determining parameters of at least one repair mode can include at least one algorithm implemented as an application-specific integrated circuit (ASIC), a complex programmable logic device (CPLD), and / or a field programmable gate array (FPGA).

[0081] The means for determining parameters of at least one repair method can include at least one trained machine learning model. Furthermore, the means for assigning a numerical value to at least one corresponding parameter, which differs from the numerical value specified by the at least one defect, can also include a trained machine learning model.

[0082] The means for determining parameters of at least one repair method can include at least one element from the following group: a mask inspection system, an interferometer, a confocal microscope, a scanning particle microscope, and a scanning probe microscope. An element from this group can acquire measurement data from the at least one defect.

[0083] The means for generating at least one repair element may include: at least one focused particle beam and at least one precursor gas configured to carry out a local chemical reaction.

[0084] In a second embodiment, which supports the understanding of the present application, the device for repairing at least one defective pattern element of a lithographic mask comprises: (a) means for determining at least one repair element of the lithographic mask that does not image the lithographic mask upon exposure, wherein the repair element is configured to modify the imaging behavior of the at least one defective pattern element; and (b) means for providing a focused particle beam and at least one precursor gas configured to generate the at least one repair element on the lithographic mask.

[0085] The means for determining at least one repair element can include at least one co-processor configured to determine the at least one repair element from measurement data of the at least one defect and design data of the lithographic mask. The determination of the at least one repair element can be based on the resolution limit of the lithographic mask and a minimum spot size of the focused particle beam at its focus.

[0086] The means for determining the at least one repair element of the lithographic mask may include at least one algorithm implemented as an application-specific integrated circuit (ASIC), a complex programmable logic device (CPLD), and / or a field programmable gate array (FPGA).

[0087] The means for determining at least one repair element may include at least one trained machine learning model.

[0088] The means for determining the at least one repair element may include at least one element from the group: a mask inspection system, an interferometer, a confocal microscope, a scanning particle microscope and a scanning probe microscope. 4. Description of the drawings

[0089] In the following detailed description, currently preferred embodiments of the invention are described with reference to the drawings, wherein Fig. 1 in the upper part of the image shows a schematic section of a pattern element of a photolithographic mask, which has an edge at the position specified by the design, and in the lower part of the image schematically shows an effective dose distribution of the optical intensity around the mask section presented in the upper part of the image in a photoresist during exposure of the photomask; Fig. 2 shows a schematic section through a mask with a defective pattern element in the upper part of the image, and illustrates the effect of the defective pattern element on the effective dose distribution applied to a photoresist in the lower part of the image; Fig. 3 in the upper part of the image shows a repair of the defective pattern element of the photolithographic mask of the Fig. 2 shows the state of the art, and in the lower part of the image illustrates the effect of the defect repair on the optical intensity distribution of the repaired mask section of the upper part of the image; Fig. 4 in the upper part of the image shows an ideal repair of the defective pattern element of the photolithographic mask of the Fig. 2 presented on the basis of a method according to the invention, and in the lower part of the image shows the optical intensity distribution generated by the optimally repaired photomask in the photoresist of a wafer; Fig. 5 in the upper part of the image shows a real repair of the defective pattern element of the mask. Fig. 2 based on a method according to the invention, and in the lower part of the image illustrates the effect of the repaired photomask on the optical intensity distribution generated in a photoresist; Fig. Figure 6 schematically shows an alternative embodiment of a repair element for compensating for the defect of the Fig. 2 illustrated; Fig. Figure 715 schematically presents a perfectly placed edge of a pattern element, Figure 735 shows a pattern element that has a defect of excess absorber material, Figure 755 represents the repair of the defect of Figure 735 according to the prior art, Figure 775 illustrates the compensation of the defect of Figure 735 by creating a repair element described in this application, and Figure 795 illustrates a second embodiment of a repair element according to the invention; Fig. Figure 8 in the upper part 805 shows a pattern element of a photolithographic mask whose edge is positioned exactly at the location provided by the design, Figure 835 in the middle part shows the repair of a defect of missing absorber material by generating a repair element according to the invention, and Figure 855 in the lower part shows the repair of a defect of excess absorber material by generating a repair element according to the invention; Fig. Figure 9 presents a schematic top view of a striped structure with a defect in the upper part of the image, and shows the striped structure after the defect has been repaired according to the state of the art in the lower part of the image; Fig. 10 in the upper part of the image, the upper part of the image Fig. 9 reproduced, in the middle sub-image illustrating an exemplary repair of the defect of the upper sub-image according to one of the methods described in this application, and in the lower sub-image illustrating the averaging of the complex amplitude of the actinic radiation on a length scale of the optical resolution; Fig. Figure 11 in the upper part 1105 schematically shows a top view of an absorbing, angled pattern element, figure 1125 shows a pattern element that has a defect of excess absorber material, figure 1145 represents the repair of the defect of figure 1125 according to the prior art, figure 1165 illustrates the compensation of the defect of figure 1125 by generating a repair element described in this application, and figure 1185 illustrates the generation of a second example of a repair element according to the invention; Fig. 12. Repairing a defect in missing absorber material according to the Fig. The scheme presented in 11 is reproduced; Fig. 13 in the upper part 1305 shows a striped structure of a mask arranged on a substrate, wherein one pattern element has a defect of missing absorber material, the middle part 1335 represents a rigorous simulation of the defective mask section, and the lower part 1365 shows the optical intensity distribution of the defective mask section; Fig. 14 the mask section of the Fig. 13 represents an optimal repair of the defect using a method according to the invention; Fig. 15 the mask section of the Fig. 13 after a repair of the defect of Fig. 13 represents a repair element that was not optimally positioned; Fig. 16 in the upper left sub-image 1605 shows a mask section measured with a scanning electron microscope which has a defect, in the upper right sub-image 1635 shows a repair element in the mask section of the upper left sub-image which compensates for the defect, and in the lower left sub-image 1665 presents the repaired mask section of the upper left sub-image 1605; Fig. 17 in the upper part of the image 1705 the normalized CD variation within the defective mask section of the upper left part of the image Fig. 16 represents, and in the lower sub-image 1755 the normalized CD variation of the repaired mask section of the lower left sub-image 1665 of the Fig. 16 represents; Fig. 18 in the upper part 1805 shows a schematic section through a striped structure with a defect, in part 1825 illustrates the mask section of part 1805 around the repaired defect, in part 1835 shows the variation of the CD caused by the defect of the upper part, and in the lower part 1865 presents the CD variation remaining after the defect repair; Fig. 19 provides a flowchart of a first embodiment of the method for repairing at least one defect in a lithographic mask; Fig. 20 shows a flowchart of a second embodiment of the method for repairing at least one defect in a lithographic mask; Fig. 21 presents a schematic section through a device for repairing one or more defects in a lithographic mask; Fig. Figure 22 illustrates a schematic section through an optical mask inspection system and a comparison to a scanner of a photolithographic exposure system; Fig. 23 presents a schematic section through a device for repairing at least one defective pattern element of a lithographic mask; and Fig. 24 shows a schematic section through a device comprising a particle beam source and a gas supply system of the Fig. 23 realized. 5. Detailed description of preferred embodiments

[0090] Preferred embodiments of a method and device according to the invention for repairing one or more defects in a lithographic mask are explained in more detail below. Furthermore, exemplary embodiments of a method and device according to the invention for repairing a defective pattern element of a lithographic mask are described in detail below. The methods according to the invention are described using the example of a binary photomask for the deep ultraviolet (DUV) wavelength range. However, these methods are not limited to improving the repair of defective DUV masks. Furthermore, the methods according to the invention are mainly explained with reference to edge placement errors of pattern elements. However, these methods are not limited to repairing this type of error.Rather, these can be used to facilitate the repair of small defects of all kinds, as well as for various types of photolithographic masks. The masks can include transmissive and reflective photomasks. Furthermore, small defects in binary and / or phase-shifting masks, as well as small defects in masks for multiple exposures, can be repaired. In the following, the term mask or photomask will also encompass a template for nanoimprint lithography.

[0091] Furthermore, the devices according to the invention for repairing one or more defects in lithographic masks are explained using the example of a modified scanning electron microscope. However, the devices according to the invention are not limited to being based on a scanning electron microscope. Rather, devices according to the invention can be based on any scanning particle microscope; that is, a device defined in this application can employ any type of particle, preferably a particle with mass, for examining and / or generating one or more repair elements of a photomask.

[0092] The upper part of image 105 of the Fig. Figure 1 shows a schematic section of a one-dimensional (1D) portion of a photolithographic mask 100. The mask 100 can be a transmitting or a reflecting mask 100. In the example of the Fig. The photomask 100 comprises a binary transmissive mask 100. The photolithographic mask 100 has a substrate 110 with a surface area 115. A pattern element 120 or a structural element 120 with a surface area 125 is arranged on the surface 115 of the substrate 110. The substrate 110 can comprise a quartz substrate and / or a material with a low coefficient of thermal expansion (LTE (Low Thermal Expansion) substrate). In a transmissive photomask 100, its substrate 110 is essentially optically transparent to electromagnetic radiation at the actinic wavelength. The pattern element 120 can be a structural element 120 of a binary photomask 100. In this case, the pattern element 120 can comprise an element of an absorber structure 120 and, for example, have chromium. An absorbing pattern element 120 essentially absorbs all the electromagnetic radiation of the actinic wavelength incident on the pattern element 120.For DUV masks, the thickness of a pattern element 120 is in the range of 60 nm to 200 nm. Absorbing pattern elements of EUV masks currently have a layer thickness in the range of 50 nm to 70 nm (in the . Fig. 1 not shown).

[0093] The term “essentially” here means – as in other parts of the description – a specification of a measured quantity within the usual measurement errors, if measuring instruments in accordance with the state of the art are used in determining the measured quantity.

[0094] Furthermore, it is possible that the pattern element 120 comprises a structure element 120 that both shifts the phase of the actinic radiation relative to the radiation incident on the substrate 110 and absorbs a portion of the actinic wavelength light incident on the pattern element 120. Examples of such masks are molybdenum silicide-based or silicon nitride-based AttPSM (Attenuated Phase Shifting Mask) masks. Such masks typically transmit 6% to 20% of the incident optical intensity in the dark region with a phase shift of 180° compared to a transparent region of the photomask.

[0095] The pattern element 120 can also be a structure element 120 of a purely phase-shifting photomask 100. A purely phase-shifting mask 100 can be produced, for example, by etching a corresponding pattern into the substrate 110 of the mask 100, which in this case essentially comprises quartz (SiO2). This type of mask is called a CPL (chromeless phase-shifting) mask. Another example of a purely phase-shifting mask type is the AltPSM (alternating phase-shifting mask).

[0096] The upper part of image 105 of the Fig. Figure 1 shows an ideal, design-specified edge 130, which has a side wall angle 135 of essentially 90°. Furthermore, the edge 130 is positioned exactly at the location specified by the design.

[0097] The lower part of image 155 of the Fig. Figure 1 schematically presents an effective dose distribution of the optical intensity 160 during exposure of the pattern element 120 of the photomask 100. The pattern element 120 absorbs the radiation incident on the mask 100 from above, so that at a certain distance from the edge 130 below the pattern element 120, there is essentially no optical intensity in the photoresist of a wafer. In the transparent substrate region 110 of the mask 100, which is located at a distance from the edge 130 of the pattern element 120, a photoresist applied to a wafer is exposed with the maximum optical intensity. Typically, the edge 130 of a pattern element 120 is defined as the location where the optical intensity in the photoresist reaches 50% of the maximum optical intensity or the maximum effective dose. In the lower part of Figure 155 of the Fig. 1 This is illustrated by the intersection of the dashed vertical line 180 and the horizontal line 170.

[0098] In the upper part of image 205 of the Fig. In Figure 2, edge 230 of pattern element 220 is not positioned at the location specified by the design. Instead, edge 230 has a distance dx, which is illustrated by the horizontal double arrow 240. This distance dx from the intended position of edge 230, as specified by the design, can result from an incorrect placement of pattern element 220, whose dimensions have the values ​​specified by the design. In this case, pattern element 220 can be corrected by determining two repair shapes, which are then used to create two repair elements. The first repair element defines the deposition of missing absorber material in the area indicated by the double arrow 240. Fig. 2 marked area. Secondly, a second repair method specifies the removal of the excess or incorrectly positioned absorber material of pattern element 220 (in the Fig. 2 not shown).

[0099] The following assumes that the 1D representation of the Fig. The unreproduced edge of pattern element 220 is correctly positioned, and pattern element 220 only needs correction in the region of edge 230 by depositing missing absorber material. Furthermore, it is assumed that defect 240 is small. This means that the dimension dx of the one-dimensional defect 240 is smaller than the resolution limit of the photolithographic mask 200 at its actinic wavelength. The positioning or placement accuracy of an edge 130, 230 of a pattern element 120, 220 is crucial, especially due to overlay issues, for lithography systems that perform multiple exposure steps to fix a pattern element in a photoresist.

[0100] The lower part of image 255 of the Fig. Figure 2 shows the optical intensity distribution 260 or the effective dose distribution 260, which is caused by the faulty placement or positioning of the edge 230 of the pattern element 220 in a photoresist arranged on a wafer during exposure of the photolithographic mask 200 with actinic electromagnetic radiation. The positional error of the edge 230 of the pattern element 220, which in the one-dimensional example of the Fig. 2, as described by dx, translates the mask 200 into an edge placement error (EPE) when exposing a wafer, which is shown in the lower sub-image 255 of the Fig. 2 is schematically illustrated by the double arrow 280. The relationship between the EPE 280 on the wafer and the mask defect 240 is described by the relationship: EPE = mask error · MEEF · M. Here, in the Fig. In the example shown, mask defect 240, or mask error 240, is the incorrect positioning of edge 230 of pattern element 220 by dx. The mask error enhancement factor (MEEF) represents any magnification or amplification of mask defect 240 caused by mask 200. For simplicity, it is assumed that MEEF = 1. The factor M denotes the magnification or reduction with which a projection lens of an exposure system images pattern element 220 of mask 200 onto a wafer. For currently used projection lenses, M = 1 / 4 or M = 1 / 5. Fig. Figure 2, as well as the following figures, do not represent the relationship between a mask defect 240 and an EPE 280 to scale.

[0101] An ideal repair of the mask defect 240 by perfectly executing a perfect repair shape would place a repair element at the defective edge 230 of the pattern element 220, so that the defective pattern element 220, as in the Fig. The pattern element 120 shown in Figure 1 would look like this. However, this would require ideal positioning of a repair tool with respect to edge 230 of the defective pattern element 220. Furthermore, a perfect repair of defect 240 would require that the resolution limit of the repair tool be very small, ideally zero.

[0102] The upper part of image 305 of the Fig. Figure 3 illustrates a real repair of defect 240 of the Fig. 2 according to the prior art. Due to the finite positioning accuracy of the repair tool, it cannot be perfectly aligned with the edge 230 of the defective pattern element 220. The repair element 310 generated by the repair tool therefore does not ideally correspond to the 1D dimension dx of the defect 240. Rather, due to the limited positioning accuracy of the repair tool, the repair tool for the defect 240 generates a small part 320 of the repair element 310 on the surface 125 of the defective pattern element 220. For this reason, the edge 330 of the repair element 310 deviates from the position specified by the design. The remaining defect 340 has a dimension dx'.

[0103] The lower part of image 355 of the Fig. Figure 3 illustrates the change in the optical intensity distribution 360 or the EPE in a photoresist when exposing the repaired mask section 300 of the upper part of the image. Fig. 3. Compared to the exposure of the unrepaired mask 200, the repaired mask 300 comes closer to the pattern geometry specified by the design. Fig. 1 significantly closer. However, a discrepancy or EPE 380 remains, illustrated by the double arrow 380, which causes a CD (Critical Dimension) deviation of the repaired pattern element 220 that lies outside the permissible error budget of the mask 300. The defect 340 remaining after the repair according to the prior art can be repaired or compensated, as explained below. However, it is more advantageous to repair the defect 240 directly, i.e., in a single step, as shown below using the Fig. 4 described, to repair.

[0104] The Fig. Figure 405 in the upper part illustrates the best possible repair of the defective pattern element 220. Fig. 2 according to one of the methods described in this application. The repair form for correcting the defective pattern element 200 of the Fig. 2 is designed such that executing the repair shape generates a repair element 410 that has a distance 420 from the edge 230 of the defective pattern element 220. The ideal distance 420 between the edge 230 of the defective pattern element 220 and the repair element 410 is approximately dx / 2, i.e., half the distance of the edge 230 from the vertical line 170, or the edge defect 240 of the defective pattern element 220. For a height 430 of the repair element 410, which is essentially equal to the height of the pattern element 220, the repair element 410, and thus the underlying repair shape, has a dimension that is approximately 80% of the distance 240 of the edge 230 of the pattern element 220 from the vertical line 170. As already stated above, the mask-side resolution limit of DUV masks is in the range of 150 nm to about 300 nm and of EUV masks in a range of about 50 nm to 100 nm.

[0105] The lower part of image 455 of the Fig. Figure 4 presents the optical intensity distribution 460 of the repaired photomask 400, which has the repair element 410 located at a distance 420 from the edge 230 of the defective pattern element 220 of the photomask 200. From a comparison of the Fig. 1 and Fig. Figure 4 shows that the optical intensity distribution 460, produced by the combined effect of the defective pattern element 220 with the repair element 410, generates a pattern element in a photoresist that precisely realizes the edge 130 specified by the design. The repair element 420 ideally creates a pattern element specified by the design in a photoresist and thus on a wafer, even though neither the repair element 410 borders the edge 230 nor does the extent of the defect 240 correspond to the one-dimensional dimension 435 of the repair element 410.

[0106] Repair element 410 can be a repair tool, which is discussed below in the context of the Fig. 21, Fig. 22, Fig. 23 to Fig. As explained in section 24, the repair element is created on the photomask 400 based on a corresponding repair pattern. For this purpose, the repair tool can perform a local chemical deposition reaction using a particle beam and at least one precursor gas. The material composition of the repair element 410 can correspond to the material composition of the pattern element 220. However, the material composition of the repair element 410 can also differ from that of the pattern element 220, as long as the material of the repair element 410 substantially completely absorbs the actinic wavelength of the photolithographic mask 420.

[0107] The Fig. Figure 4 represents a perfect compensation of the defective pattern element 220 by creating the repair element 410 at a distance 420 from the edge 230 of the defective pattern element 220. The upper sub-figure 505 of the Fig. Figure 5 represents a realistically executable repair of the defective pattern element 220. Due to the finite positioning accuracy of a repair tool, the determined repair shape cannot position the repair element 510 exactly at the location intended for optimal compensation of the defective pattern element. In the Fig. In example 5, the distance 520 of edge 230 of the defective pattern element 220 is smaller than in the Fig. 4. Optimal repair of the defective pattern element 220 shown. As can be seen from the lower part 555 of the Fig. As can be seen in Figure 5, the placement error 540 of the repair element 510 leads to a barely perceptible deviation of the optical intensity distribution 560 compared to the ideal optical intensity distribution 460. Fig. 4 in a photoresist applied to a wafer. This means the EPE of the mask 500, which is in the Fig. The damage illustrated by arrow 580 is negligibly small after its repair by creating the repair element 510.

[0108] The reason for this lies in a reduction of the placement error 540 of the repair element 510 during transfer into a photoresist, by approximately a factor of 1 / R, where R denotes the resolution limit of the photolithographic mask. This is one of the essential advantages of the repair methods for small defects in photolithographic masks described here. The sensitivity of placing a repair element 410, 510 on the EPE 580 is significantly reduced when carrying out the methods according to the invention. This reduction in the placement sensitivity of one or more repair elements 410, 510 is described by the resolution limit R of the photolithographic mask 400, 500. The repair element 410, 510 can be placed at a distance of approximately 1% to 30% of the resolution limit R of the photolithographic mask 400, 500 without affecting the EPE 580, i.e., the repair or the defect.Compensation of pattern element 220 by more than 10% from the best possible optical intensity distribution 460 of the . Fig. 4 differs.

[0109] The Fig. Figure 6 illustrates a second example of a repair element 610, which is used to repair or compensate for the defective pattern element 220 of the Fig. 2 can be used. Unlike the height 530 of the repair element 510 of the Fig. 5 is the height 630 of the repair element 610 of the upper part image 605 of the Fig. 6 less than the height or thickness of the pattern element 220 of the mask 600. To compensate for the smaller height 630 of the repair element 610, the repair element 610 has a larger 1D dimension 635 than the repair element 510 of the Fig. 5. As shown in the lower part of image 655 of the Fig. As can be seen from Figure 6, the modified repair element 610 does not essentially lead to a change in the optical intensity distribution 660 in a photoresist. The EPE 680 is – similar to the one in the Fig. 5 - negligibly small.

[0110] In addition to the reduced placement sensitivity of the repair elements 410, 510, 610, by executing a determined repair shape, a repair element 410, 510, 610 can be generated whose lateral dimension 435, 535, 635 deviates significantly from the 1D dimension dx of the defect 240. Thus, the defect repair methods presented in this application significantly reduce the sensitivity with which a parameterized repair shape transfers one or more lateral dimensions of a defect 240 into a repair element 410, 510, 610. Besides the relaxed placement sensitivity, this is the second significant advantage of the repair methods described in this application for small defects in photolithographic masks. In particular, the repair element 410, 510, 610 can be significantly smaller than a defect 240 to be repaired. This has a favorable effect on the time required for defect repair, i.e.,the creation of the repair element 410, 510, 610.

[0111] In the Fig. 1, Fig. 2, Fig. 3, Fig. 4, Fig. 5 to Fig. Section 6 describes the application of the methods presented in this application for repairing a defect in missing absorber material, i.e., a clear defect. Fig. Section 7 explains how these methods can be used to repair a defect of excess material, i.e., a so-called dark defect. The upper part of Figure 705 of the Fig. Figure 7 presents a pattern element 720 arranged on the surface 715 of a substrate 710 of the mask 700. The edge 730 of the pattern element 720 is placed exactly at the location specified by the design, which is marked by the dashed vertical line 170.

[0112] In the second-to-last sub-image 735, pattern element 740 exhibits a defect 750 of excess material. The defect 750 has a 1D dimension 725 that extends beyond the edge 730 of the defect-free pattern element 720. The defect 750 can be caused by an incorrect placement of a pattern element 720 whose dimensions precisely match the size specified by the design. As already mentioned in the context of the Fig. 2 discussed, but it is also possible that too little absorber material was removed in the area of ​​edge 730 when structuring the mask 700.

[0113] Figure 755 depicts the repair of defect 750 according to the state of the art. If the repair element 760 is ideally generated, defect 750 can be perfectly repaired, as indicated in figure 755. The challenges associated with repairing particularly small defects 750 according to the state of the art are discussed in the following section. Fig. 3 have already been carried out.

[0114] Figure 775 illustrates the repair of the defect 750 of excess material by creating a repair element 770 according to the invention. The repair element 770 is not designed to remove the excess material of the defect 750 by performing a local etching process. Rather, the creation of the repair element 770 eliminates part of the defect 750 and a small part of the defective pattern element 740. After the creation of the repair element 770, the remaining defect residue 780, in combination with the repaired pattern element 740, influences the imaging behavior of the repaired mask 700 in the area of ​​the repaired pattern element, so that it exhibits imaging behavior similar to the pattern element 720.

[0115] The bottom sub-image 795 of the Fig. Figure 7 presents a second repair element 790, which can be generated as an alternative to repair element 770. Generating repair element 790 removes a portion of the material from the defective pattern element 740 in the region of edge 730 of the defect-free pattern element 720. The 1D dimension of repair element 790 is larger than that of repair element 770. This larger 1D dimension is compensated for by the fact that the execution of the repair shape underlying repair element 790 does not etch pattern element 740 down to the surface 715 of substrate 710 of mask 700.

[0116] Executing the repair elements 770, 790 to repair the defect 740 opens up the additional degrees of freedom mentioned above in the context of the Fig. 5 and Fig. 6 were executed.

[0117] The Fig. Figure 8 presents in the upper part 805 an absorbing pattern element 720 with a surface 725, which is arranged on the surface 715 of a substrate 710 that is optically transparent to the actinic wavelength of the photolithographic mask 800. The edge 730 of the pattern element 720 is precisely positioned at the location specified by the design. This location is in the Fig. 8 marked by the dashed vertical line 170.

[0118] The middle sub-image 835 of the Fig. Figure 8 shows a defective pattern element 840. The edge 830 of pattern element 840 is not positioned at the location specified by the mask design. As a result, pattern element 840 forms a clear defect or a defect of missing absorber material. The defect is compensated for in sub-image 835 by executing a corresponding repair shape to create a repair element 850. In this example, the repair element 850 closes the gap. Fig. 8 to the edge 830 of the defective pattern element 840. The repair element 850 comprises deposited absorber material whose 1D dimension extends beyond the boundary of the missing absorber material. The height of the exemplary repair element 850 is approximately half the height of the pattern element 820, 840.

[0119] The lower part of image 855 of the Fig. Figure 8 presents a pattern element 860 that has a defect of excess absorber material. The defect of excess material is repaired by removing an upper portion of the excess absorber material by creating the repair shape 870. The edge 880 of the repaired pattern element 860 does not correspond to the position specified by the design, which is indicated by the dashed line 170 in the Fig. 8 is indicated. Nevertheless, the repair element 870 eliminates the effect of the defect of excess material down to a negligible amount.

[0120] Due to the reduced placement requirements for the repair tool, the creation of repair elements 850, 870 can be performed with less effort compared to the creation of repair elements 310, 760 according to the state of the art.

[0121] The Fig. Figure 9 presents in the upper part 905 a schematic top view of a mask section 900, which has a striped structure (“lines and space pattern”) with two absorbing strips 920 and 930 arranged on an optically transparent substrate 910. In the Fig. In the example shown in Figure 9, the left pattern element 920 has a two-dimensional (2D) defect 940 of excess absorbing material.

[0122] The lower part of image 955 of the Fig. Figure 9 illustrates the repair of defect 940 by performing a local etching process using a particle beam and an etching gas. The example of a repair process shown in the lower part of Figure 955... Fig. As shown in Figure 9, the defect 940 has been almost perfectly eliminated, except for a negligible residual defect 950. As a rule, the laterally extended defect 940 cannot be removed as precisely as shown in the lower part of Figure 955. Fig. 9 reproduced.

[0123] As explained above, the precision with which a repair tool or the repair shape can position defect 940 relative to defect 940 or relative to pattern element 920 is finite. Furthermore, both the photomask 900 and the repair shape are subject to drift, such as thermal drift, during the creation of a repair element. Therefore, the repaired pattern element 920 often exhibits an edge roughness along the processed right edge that is outside a predefined error interval (in the Fig. 9 not shown). Therefore, executing a corresponding repair function to remove the defect 940 extending along pattern element 930 is a very complex and time-consuming process.

[0124] The Fig. Figure 10 illustrates the repair of the 2D defect 940 of the Fig. 9 using one of the methods presented in this application. The upper part 1005 of the Fig. 10 reproduces the upper part 905 of the Fig. 9. The middle sub-image 1035 of the Fig. Figure 10 illustrates the compensation of excess material in defect 940 by generating repair elements 1010. The repair elements 1010 created local etched structures in defect 940, partly in the substrate 910 of mask 900 and partly also in the pattern element 920. The various repair elements 1010 can be produced based on a single repair mold, which is generated at different positions along the defect extending along the right edge of the pattern element. However, it is also possible to determine a single repair mold that generates the various repair elements 1010, as shown in the middle sub-figure 1035. Fig. 10 illustrated, generated.

[0125] The major advantage of the repair of defect 940 shown in sub-image 1035 is the significantly reduced sensitivity of placing or positioning the repair elements 1010 with respect to defect 940. The lower sub-image 1065 of the Fig. Figure 10 schematically illustrates the reduced sensitivity when placing the repair elements 1010. The dashed ellipses 1050 illustrate the area of ​​a photolithographic mask 900 or the length scale over which the mask 900 averages the complex amplitude of the actinic electromagnetic radiation. If structures are present on a photomask whose dimensions in one or two dimensions are smaller than the resolution limit of the photomask, the actinic radiation averages over the details of these structures. This means that the microroughness of pattern elements does not, or hardly, affect the imaging behavior of the photolithographic mask, whereas the complex amplitude of the electromagnetic actinic radiation averaged over a dimension in the region of the resolution limit does influence the imaging behavior of the photomask.

[0126] The dashed ellipses 1050 show the areas of a mask that contribute optical intensity to a point of an image of the mask in a photoresist. The scale of this averaging area is—as already explained above—determined by the resolution limit R of the photolithographic mask. As long as the dimensions of the repair elements 1010 are small compared to the resolution limit R, their size and placement are determined by a factor of up to R. -1 less sensitive compared to both the size and positioning of prior art repair elements. This means that a very large area of ​​different repair elements 1010 leads to the same or a very similar result with regard to the imaging behavior of a photolithographic mask. The methods described in this application exploit this fact to improve the repair of small defects, which have previously been particularly difficult to repair.

[0127] The resolution limits of current DUV and EUV masks are listed above.

[0128] In the Fig. 11 and Fig. Section 12 explains how to repair a defect using excess and missing absorber material, where the repair element may have a different geometric shape than the defect being repaired. Figure 1105 shows a mask section 1100 with an angled pattern element 1120 arranged on a mask substrate 1110, the shape of which meets the design specifications. Figure 1125 of the Fig. 11 The pattern element 1130 has a defect 1140 of excess absorber material located on the inner side of the angular area of ​​the pattern element 1130.

[0129] Part 1145 of the Fig. Figure 11 presents the repair of the defect by executing a repair method associated with the defect, in accordance with the prior art, which creates the repair element 1150 in the area of ​​the defect 1140. The associated problem has already been discussed above in the context of the Fig. 3 and Fig. 9 has been explained in detail.

[0130] Part 1165 of the Fig. Figure 11 represents a first example of generating a repair element 1170 that compensates for the defect 1140. This element is much less sensitive than the repair element 1150 of sub-image 1145, both with regard to the placement of the repair element 1170 relative to the defect 1140 and the lateral dimensions of the repair element 1170 relative to the dimensions of the defect 1140. Finally, sub-image 1185 shows a repair element 1190 whose geometric shape differs significantly from the shape of the defect to be repaired. Generating a circular or near-circular repair element 1190 is considerably easier than generating a rectangular or square repair element 1170. Nevertheless, the repair element 1190 of sub-image 1185 compensates for the defect 1140 to an extent that is in no way inferior to the compensation provided by the repair element 1170. The reasons for this are discussed in the Fig. 10 has been explained.

[0131] Based on the Fig. Section 12 now describes how to repair a defect in missing absorber material. Figure 1205 of the Fig. Figure 12 shows a square pattern element 1220 placed on an optically transparent substrate 1210 of a photolithographic mask 1200. In the pattern element 1230 of the sub-image 1225, a part 1240 of the pattern element 1230 is missing. This means that the latter has a clear defect 1240 or a defect 1240 of missing absorbing material.

[0132] Part 1245 of the Fig. Section 12 describes the repair of the defect 1240 (missing material) by attaching a corresponding repair element 1250 over the defect 1240 by carrying out the repair method determined for the defect. The repair element 1250 is manufactured according to the state of the art.

[0133] Part 1245 of the Fig. Figure 12 shows a first example of a repair element 1270, which was manufactured according to one of the methods described in this application and which avoids the sensitivity of the repair method of partial figure 1245. In addition, the repair element 1290 illustrates a second embodiment of a repair element 1290, which only partially replicates the complicated contour of the defect 1240 and is therefore even simpler to manufacture than the repair element 1270.

[0134] The Fig. Figure 13 presents in the upper sub-image 1305 a top view of a mask section 1300, which has a striped structure with absorbing striped pattern elements 1320 applied to a substrate 1310 of the photomask 1300. The second pattern element 1320 from the left has a defect 1330 of missing material. The width of the defect 1330 was chosen to be 15% of the half-pitch of the striped structure. The middle sub-image 1335 of the Fig. Figure 13 presents a rigorous simulation of the imaging behavior of the mask section 1300 of the upper sub-image 1305. The defect 1330 is clearly visible as defect 1340 in the middle sub-image 1335.

[0135] The lower part of image 1365 of the Fig. Figure 13 represents the optical intensity distribution of the defective mask section 1300. The dashed horizontal line 1360 indicates the maximum of the optical intensity distribution or the effective dose distribution of a defect-free mask section. The intersection points of the intensity distribution with the dashed horizontal line 1370 describe the points at which a photoresist is considered exposed. Thus, these intersection points represent the width of the striped pattern element produced on a wafer by the mask section 1300. The defect 1330 is reflected in a wider central pattern element produced on a wafer. A variation in the width of the pattern element imaged onto a photoresist by the mask section 1300 results in a variation of the CD (Critical Dimension) along the imaged pattern.

[0136] The Fig. Figure 14 shows the defective mask section 1300 in the upper part of the image 1405, which was repaired by creating a repair element 1440. The repair element 1440 was created in the Fig. In the example shown in Figure 14, the repair element 1440 is not placed on the defect 1330, as is usual in the prior art, but rather positioned directly next to the defect 1330 where the absorber material is missing. As can be seen in the middle sub-image 1435, the repaired defect 1330 is no longer visible in the imaging simulation. Furthermore, it is clearly evident from the lower sub-image 1465 that the repair element 1440 perfectly repairs the defect 1330, so that there is no variation in optical intensity along the length of the pattern elements 1320.

[0137] The Fig. Figure 15 illustrates in the upper part of the figure 1505 a repair of the defect 1330 of the Fig. 13, where the repair element 1540 was placed 15% of the half-pitch away from the pattern element 1320. In the image simulation, which is shown in the middle sub-image 1535 of the Fig. As shown in image 15, no deviation from a defect-free mask structure is discernible. In the lower part of image 1565 of the Fig. In 15, a slight reduction in the maximum of the optical intensity distribution is visible, resulting in a smaller width of the central pattern element in a wafer. However, the deviation in width is still within the error budget specified for mask 1300.

[0138] The upper left part of the Fig. Figure 16 shows a section of mask 1600 photographed with a scanning electron microscope, exhibiting a stripe pattern (lines and space pattern) with a half-pitch of 152 nm on the mask 1600. As indicated by the dashed, kinked vertical line 1625, one pattern element 1620 of the mask section 1600 has a defect 1630 of missing absorbing material. The exemplary defect 1630 of the Fig. Pattern element 16 has a perpendicular extent of 10 nm. This extent is significantly smaller than the resolution limit R of the photolithographic mask 1600.

[0139] In the upper right sub-image 1635, the repair element 1640 is schematically shown, designed to compensate for the defect 1630 in combination with the adjacent pattern elements 1620. The lower left sub-image 1665 of the Fig. Figure 16 shows the repaired mask section of the upper left partial image 1605 as a difference image to the upper right partial image 1635. As indicated by the straight vertical line 1645, the defect 1630 is no longer visible in the image of the mask section 1600.

[0140] The upper part of image 1705 of the Fig. Figure 17 shows the variation of the CD for the different pattern elements of the upper left sub-image 1605 of the Fig. 16 along the pattern elements. The dashed line 1710 defines the design-specified CD of a pattern element created with mask 1600 on a wafer; in which in the Fig. In the example shown in Figure 17, this value is: CD = 37.68 nm. The dashed horizontal lines 1720 and 1730 represent the lower and upper limits of the permissible CD variation (ΔCD), respectively. The CD variation range ΔCD in the example is... Fig. 17 ±2.5%. The curves in the upper part of the diagram (1705) show that curve 1750 touches the upper limit of the CD tolerance interval, while CD curve 1760 lies well outside the CD tolerance interval of ±2.5% for most of its course. The maximum normalized CD variation for curve 1760 is: ΔCD / CD = 8.6%.

[0141] The lower part of the image 1755 Fig. 17 presents the CD variation for the various pattern elements 1620 of the repaired mask section of the lower left part of the image 1665 of the Fig. 16. All CD curves lie within the permissible tolerance interval, i.e., all CD curves meet the requirement: ΔCD / CD < ±2.5%. This means that the repair element 1640 completely compensates for the defect 1630.

[0142] The upper part of the image 1805 of the Fig. Figure 18 shows a 1D section through a mask section 1800 with eight pattern elements 1820 in the form of a striped structure, arranged on an optically transparent substrate 1810 of the mask 1800. The eight pattern elements 1820 generate seven optically transparent stripes, numbered 0 to 6 in sub-image 1805. The half-pitch of the striped structure on the mask 1800 is 152 nm, or 38 nm on a wafer. In the upper sub-image 1805, the second pattern element 1820 from the left has a defect 1840 of excess, absorbing, and phase-shifting molybdenum silicide (MoSi) material at its right edge. To analyze the effect of the defect 1840, the striped structure of the defective mask section 1800, as shown in sub-image 1805, is simulated.The simulation parameters are: NA = 1.35, λ = 193 nm, outer σ value: 1.0, inner σ value: 0.88, exposure setting: Disar, polarization: y-direction, pattern: L&S (lines & space). MoSi with 6% absorption at the actinic wavelength. The electromagnetic radiation incident on the 1800 mask is assumed to be coherent.

[0143] The middle part of the image 1835 of the Fig. Figure 18 presents a simulation of the variation of the CD for the seven optically transparent stripes 0 to 6 along the stripe direction, i.e., perpendicular to the plane of the paper. Additionally, the simulation determines the shift of the center of gravity (CoG) of the optical intensity distribution. The dashed horizontal line 1870 represents the target CD specified for mask 1800. The curves in the middle sub-figure 1835 describe the variation of the critical dimension, i.e., ΔCD, along the pattern elements 1820. The table in the middle sub-figure 1835 summarizes the ΔCD and ΔCoG for the optically transparent stripes 0, 1, and 2. The table shows that the variation of the CD for the zeroth stripe is 3.1 nm, which is significantly larger than the error budget of mask 1800 of 2.5%. The effect of defect 1840 on the second optically transparent stripe is a consequence of the exposure of mask 1800 with coherent radiation.The simulation was performed using the rigorous optical imaging program DrLitho from the Fraunhofer Institute for Integrated Device Technology (IISB) Erlangen.

[0144] The enlarged section 1825 shows the pattern elements 1820 around the defective transparent strip 1 after the defect 1840 has been repaired by executing the corresponding repair shape determined for the defect. When executing the repair shape for the local etching of the defect 1840, it is assumed that the local etching process produces a sidewall angle 1850 with a negative angle of -20°. The negative sidewall angle 1850 causes an amplitude defect in an aerial photograph or when exposing a wafer. It is accounted for by a magnification factor MEEF = 1.4. The negative sidewall angle 1850 is compensated by a (positive) shift of the foot point 1860 of the edge 1830, such that the width of the strip 1 is approximately half the height of the nominal width of the optically transparent strip.Errors in the sidewall angle can be essentially completely compensated for an angle range of approximately +20° by a corresponding shift of the foot of edge 1830 of the repaired pattern element 1820. The required lateral shift for this is in the range of approximately ±3.3 nm for λ = 193 nm.

[0145] The lower sub-image 1865 shows the simulated variation of the CD for optically transparent stripes 0 to 6 along the stripe direction. Furthermore, the simulation determines the CoG shift of the optical intensity distribution. The dashed horizontal line 1870 describes the target value of the CD on a wafer. A comparison of the curve families of sub-images 1835 and 1865 immediately reveals that repairing defect 1840 drastically reduces the variation of the CD. The table of the lower sub-image 1865—similar to the table of the middle sub-image 1865—summarizes the ΔCD and ΔCoG for optically transparent stripes 0, 1, and 2. Compared to the table of the middle sub-image 1835, repairing defect 1840 has reduced the variation of the CD by more than an order of magnitude.

[0146] In connection with the Fig. Section 18 discusses the repair of a defect (1840) of excess material adjacent to the edge (1820) of a pattern element without completely covering the transparent strip. It is, of course, also possible to repair a defect of excess material that completely bridges two pattern elements. Furthermore, this can be done in the context of Fig. The procedures described in section 18 can also be used to repair defects or missing material.

[0147] The flowchart from 1900 of the Fig. Reference 19 describes a first embodiment of the method discussed in this application for repairing at least one defect 240, 750, 940, 1140, 1240, 1330, 1630, 1840 of a lithographic mask 200, 400, 500, 600, 700, 800, 900, 1100, 1200, 1300, 1600, 1800. The method begins at step 1910. In the next step 1920, parameters of at least one repair form for the at least one defect 240, 750, 940, 1140, 1240, 1330, 1630, 1840 are determined, wherein determining parameters comprises: assigning a numerical value to at least one corresponding parameter, which is derived from the value caused by the at least one defect 240, The specified numerical value deviates from the values ​​obtained in steps 750, 940, 1140, 1240, 1330, 1630, and 1840. The procedure ends at step 1920.

[0148] Furthermore, the flowchart 2000 presents the Fig. 20 the steps of a second embodiment of the method for repairing at least one defective pattern element 220, 740, 840, 860, 920, 1130, 1230, 1320, 1620, 1820 of a lithographic mask 200, 400, 500, 600, 700, 800, 900, 1000, 1100, 1200, 1300, 1600, 1800. The method begins at step 2010. In the next step 2020, at least one repair element 410, 510, 610, 770, 790, 850, 870, 1010, 1170, 1190, 1270, 1290, 1440, 1540, 1640 of the lithographic mask determines that the lithographic mask does not image when exposed, wherein the repair element changes an imaging behavior of at least one defective pattern element.

[0149] Then, in step 2030, at least one repair element is created on the lithographic mask using a focused particle beam and at least one precursor gas. The process ends in step 2040.

[0150] The Fig. Figure 21 schematically shows a section through a device 2100 designed to determine parameters for a repair mold for at least one defect 240. For this purpose, the device 2100 has an optical mask inspection system 2110. Fig. Figure 22 illustrates the principle of an optical mask inspection system designed to capture an aerial image of a transmissive mask 200. The left sub-figure 2105 of the Fig. Figure 22 schematically depicts some components of a scanner. An exposure system focuses electromagnetic radiation of the actinic wavelength onto a photolithographic mask. A projection optic or projection lens reduces the size of the radiation passing through the photomask (typically 1:4 or 1:5) and projects it onto a wafer, or onto a photoresist with a large numerical aperture (NA) distributed on the wafer.

[0151] In the right-hand part 2155 of the Fig. Figure 22 shows some components of an optical mask inspection system 2110, designed for the actinic wavelength of the scanner in the left sub-image 2105. The exposure system of the scanner and the optical mask inspection system 2110 are essentially identical. This means that image generation is essentially the same for both systems. The optical mask inspection system 2110 thus images a section of the optical intensity distribution of a mask as it appears on a photoresist located on the wafer. Unlike a scanner, however, in an optical mask inspection system 2110, a lens projects a small, highly magnified section of the optical intensity distribution of a photomask onto a CCD (charge-coupled device) camera. This makes it possible to visualize defects that a photomask exhibits at the actinic wavelength in its aerial image and to detect them using a CCD sensor or a CCD camera.

[0152] The optical mask inspection system 2110 of the device 2100 can provide the measurement data of one or more aerial images to a computer system 2130 of the device 2100 via the connection 2120. The computer system 2130 of the device 2100 can determine the parameters of a repair method associated with the defect in the aerial image(s) from the measurement data. For this purpose, the computer system 2130 can include a co-processor 2140, which is specifically designed to efficiently execute an algorithm that determines the parameters of a repair method associated with the defect from the aerial image(s) of the optical mask inspection system. Furthermore, the computer system 2130 of the device 2100 can include a second algorithm designed to assign a value or numerical value to one or more parameters of the repair method that differs from the numerical value specified by the defect.The second algorithm can also be executed by the 2140 co-processor.

[0153] However, it is also possible that the computer system 2130 has a dedicated hardware component 2150 that executes one or both of the algorithms described above. The hardware component 2150 of the computer system can be implemented as an ASIC (Application Specific Integrated Circuit), a complex programmable logic device (CPLD), and / or a field programmable gate array (FPGA).

[0154] Additionally or alternatively, the computer system 2130 can include a dedicated graphics processor 2160 designed to execute a trained machine learning model. A machine learning model can be trained in at least two ways, or the graphics processor can execute two different trained machine learning models designed for the respective problem. On the one hand, a machine learning model can be trained to determine, from the measurement data of the optical mask inspection system 2110, from design data of the lithographic mask 200, settings of the exposure system, and, if applicable, from RET structures generated on the mask 200, the parameters of one or more repair methods that are executed to repair or compensate for the defect(s), i.e., to generate one or more repair elements 410, 610, 610.

[0155] Alternatively or additionally, a machine learning model can assign a different numerical value to one or more parameters of an already parameterized repair shape to determine the repair elements 410, 510, 610 described above according to the invention. However, the currently preferred embodiment is that a machine learning model directly predicts the parameters of a repair shape from the input data specified above to form one of the repair elements 410, 510, 610 described in this application. The process of training a machine learning model is not addressed in this application.

[0156] Furthermore, the computer system 2130 can include non-volatile memory 2170 in which the algorithm(s), machine learning model(s), and / or trained machine learning model(s) are stored. The non-volatile memory 2170 can comprise solid-state storage (SSD).

[0157] Furthermore, the computer system 2130 can include a control unit 2180 designed to control the optical mask inspection system 2110.

[0158] Furthermore, the device 2100 may include a scanning particle microscope, a scanning probe microscope and / or a confocal microscope designed to scan a defect 240 of a photolithographic mask 200 and to generate an image of the measurement data. If the device 2100 has one or more of these measuring instruments, the control unit 2180 may also control these measuring instruments.

[0159] The Fig. Figure 23 schematically presents a cross-section through a device 2300 that can repair at least one defective pattern element 220. For this purpose, the device 2300 has an optical mask inspection system 2310. This type of measuring device has already been discussed above in the context of the discussion of Fig. 22 described.

[0160] The optical mask inspection system 2310 of the device 2300 can provide the measurement data of one or more aerial photographs or a stack of aerial photographs to a computer system 2330 of the device 2300 via the connection 2320. The computer system 2330 of the device 2300 can then provide the computer system 2130 of the device 2100 with the Fig. 21 will be similar. To avoid excessive length, a description of the 2330 computer system will be omitted. Instead, the discussion will focus on the Fig. 21 referred.

[0161] The device 2300 further comprises a particle beam source 2350, which can provide a focused particle beam. The focused particle beam of the particle beam source 2350 can be used, on the one hand, to analyze a defect 240 in a photolithographic mask 200. Based on the measurement data of the focused particle beam and / or the measurement data of the optical mask inspection system 2310, a repair method for the defect can be determined using one or more algorithms or one or more machine learning models. On the other hand, the focused particle beam of the particle beam source 2350, in combination with the gas supply system 2370 of the device 2300, can be used to repair the analyzed defect 240. Both the particle beam source 2350 and the gas supply system 2370 can exchange data with the computer system 2330 via the connections 2340 and 2360.Furthermore, the control unit 2370 of the computer system 2330 can control the optical mask inspection system 2310, the particle beam source 2350 and the gas supply system 2370.

[0162] The Fig. Figure 24 shows a schematic section through a device 2400, which combines a particle beam source 2350 and a gas supply system 2370 of the device 2300. A repair element 410, 510, 610 for repairing the defect 240 can be produced with the device 2400. The exemplary device 2400 of the Fig. Reference 24 comprises a modified scanning particle microscope 2410 in the form of a scanning electron microscope (SEM) 2410. The device 2400 includes a particle beam source 2350 in the form of an electron beam source 2405, which generates an electron beam 2415 as a massed particle beam 2415. An electron beam 2415 can be focused onto a spot that is significantly smaller than the focus diameter of a photon beam. Due to the small de Broglie wavelength of electrons, the electron beam 2415 can be focused onto a spot diameter in the range of a few nanometers. As an analysis or measurement tool, an electron beam 2415 thus exhibits a very high lateral resolution.

[0163] Furthermore, an electron beam 2415 – compared to an ion beam – has the advantage that the electrons striking the sample 2425, for example the photolithographic mask 200, cannot substantially damage the sample 2425 or the photomask 200. However, it is also possible to use an ion beam, an atom beam, or a molecular beam in the device 2400 to process the sample 2425 (in the Fig. 24 not shown).

[0164] The scanning particle microscope 2410 consists of an electron beam source 2405 and a column 2420 in which the beam optics 2413 are arranged approximately in the form of an electron optic of the SEM 2410. In the SEM 2410 of the Fig. 24 The electron beam source 2405 generates an electron beam 2415, which is directed by the imaging elements arranged in the column 2420, which are in the Fig. 24 (not shown), as a focused electron beam 2415 is directed at position 2422 onto the sample 2425, which can encompass the photolithographic mask 200. Thus, the beam optics 2413 form the imaging system 2413 of the electron beam source 2405 of the device 2400.

[0165] The imaging elements of column 2420 of SEM 2410 can further scan or rasterize the electron beam 2415 across the sample 2425. The sample 2425 can be examined using the electron beam 2415 of the device 2400. The electron beam 2415 typically strikes the sample 2425 perpendicularly.

[0166] The backscattered electrons and secondary electrons generated by the electron beam 2415 in an interaction zone or scattering bulb of the sample 2425 are registered by the detector 2417. The detector 2417, which is arranged in the electron column 2420, is referred to as an "in-lens detector". The detector 2417 can be installed in the column 2420 in various embodiments. The detector 2417 converts the secondary electrons generated by the electron beam 2415 at the measuring point 2422 and / or the electrons backscattered by the sample 2425 into an electrical measurement signal and forwards this signal to an evaluation unit 2480 of the device 2400. The evaluation unit 2480 analyzes the measurement signals from the detectors 2417 and 2419 and generates an image of the sample 2425, which is displayed on the display 2495 of the evaluation unit 2480.The detector 2417 may also contain a filter or filter system to discriminate electrons in energy and / or solid angle (in the . Fig. 24 not reproduced).

[0167] The exemplary device 2400 can include a second detector 2419. The second detector 2419 can be configured to detect electromagnetic radiation, particularly in the X-ray range. This allows the detector 2419 to analyze the material composition of the radiation generated by the sample 2425 during its examination. The detectors 2417 and 2419 can be controlled by the control unit 2370 of the computer system 2330. In an alternative embodiment, the device 2400 has its own control unit (in the Fig. 24 not shown).

[0168] Furthermore, the device 2400 can include a third detector (in the Fig. 24 not shown). The third detector may be in the form of an Everhart-Thornley detector and is typically located outside column 2420. It is generally used for detecting secondary electrons.

[0169] The device 2400 can include an ion source that provides ions with low kinetic energy in the region of the sample 2425 (in the Fig. 24 not shown). The ions with low kinetic energy can compensate for a charging of sample 2425.

[0170] The sample 2425 is arranged for examination on a sample stage 2430 or a sample holder 2430. A sample stage 2430 is also known in the field as a "stage". As in the Fig. The sample stage 2430, symbolized by the arrows, can be accessed, for example, by micromanipulators located in the Fig. 24, which are not shown, are moved in three spatial directions relative to column 2415 of SEM 2410.

[0171] In addition to translational movement, the sample stage 2430 can be rotated about at least one axis oriented parallel to the beam direction of the particle beam source 2405. Furthermore, the sample stage 2430 may be rotatable about one or two additional axes, wherein these axis(s) are arranged in the plane of the sample stage 2430. Preferably, the two or three axes of rotation form a rectangular coordinate system.

[0172] The sample 2425 to be examined can be any microstructured component or part that requires analysis and, if necessary, subsequent processing, for example, the repair of a local defect 240 of a pattern element 220 of a photolithographic mask 200.

[0173] The device 2400 of the Fig. 24 may also include one or more scanning probe microscopes, for example in the form of an atomic force microscope (AFM) (in the Fig. 24 not shown), which can be used to analyze and / or process sample 2425.

[0174] The one in Fig. The scanning electron microscope 2410, shown as an example in Figure 24, is operated in a vacuum chamber 2470. To generate and maintain the required vacuum pressure in the vacuum chamber 2470, the SEM 2410 features... Fig. 24 a pump system 2472 on.

[0175] The following section describes the gas supply system 2370, which is implemented in the device 2400. As mentioned above, the sample 2425 is arranged on a sample stage 2430. The imaging elements of the column 2420 of the SEM 2410 can focus the electron beam 2415 and scan or rasterize the sample 2425. The electron beam 2415 of the SEM 2410 can be used to induce a particle beam-induced deposition (EBID) process and / or a particle beam-induced etching (EBIE) process. To perform these processes, the exemplary device 2400 has the following features: Fig. 24 three different storage containers 2440, 2450 and 2460 for storing different precursor gases.

[0176] The first reservoir 2540 stores a precursor gas, for example, a metal carbonyl such as chromium hexacarbonyl (Cr(CO)6) or a main group metal alkoxide such as TEOS. Using the precursor gas stored in the first reservoir 2440, missing material from the photolithographic mask 200 can be deposited onto it in a local chemical deposition reaction. Missing material from a mask 200 can include missing absorber material, such as chromium; missing substrate material 210, such as quartz; missing material from an OMOG mask, such as molybdenum silicide; or missing material from a multilayer structure of a reflective photomask, such as molybdenum and / or silicon.

[0177] The electron beam 2415 of the SEM 2410 acts as an energy source to split the precursor gas stored in the first reservoir 2540 at the location where material is to be deposited on the sample 2425. This means that by the combined provision of an electron beam 2415 and a precursor gas, an EBID process is performed for the local deposition of missing material, for example, missing material from the photomask 20. The modified SEM 2410 of the device 2400, in combination with the precursor gas stored in the first reservoir 2440, can form a device for generating a repair element 410, 510, 610 on a photolithographic mask.

[0178] As explained above, an electron beam 2415 can be focused to a spot diameter in the range of a few nanometers. The interaction region, or scattering bulb, in which an electron beam 2415 generates secondary electrons depends on both the energy of the electron beam 2415 and the material composition upon which the electron beam 2415 interacts. The diameters of interaction regions range into the low single-digit nanometer range. Thus, the diameter of a scattering bulb of an electron beam 2415 limits the achievable resolution when generating a repair element 410, 510, 610 by executing the corresponding repair shape. This resolution limit is currently in the single-digit nanometer range.

[0179] In the Fig. In the apparatus 2400 shown in Figure 24, the second reservoir 2450 stores an etching gas that enables the execution of a local electron beam-induced etching (EBIE) process. Using an electron beam-induced etching process, excess material can be removed from the sample 2425, such as the excess material of the pattern element 860 from the photolithographic mask 800. An etching gas can, for example, comprise xenon difluoride (XeF2), a halogen, or nitrosyl chloride (NOCl). Thus, the particle beam source 2350, in combination with the gas supply system 2370, forms an apparatus 2400 for generating a repair element 410, 510, 610.

[0180] The third storage container 2460 can store an additive or additional gas that can be added, as needed, to the etching gas held in the second storage container 2450 or to the precursor gas stored in the first storage container 2440. Alternatively, the third storage container 2460 can store a second precursor gas or a second etching gas.

[0181] Each of the storage containers 2440, 2450 and 2460 has in the Fig. The device 2400 shown in Figure 24 has its own control valves 2442, 2452, and 2462 to control the amount of the corresponding gas supplied per unit time, i.e., the gas flow rate, at the point 2422 where the electron beam 2415 strikes the sample 2425. The control valves 2442, 2452, and 2462 can be controlled by the control unit 2370 of the computer system 2330. This allows the partial pressure ratios of the gas(es) supplied at the processing location 2422 for carrying out an EBID and / or an EBIE process to be adjusted over a wide range.

[0182] Furthermore, in the exemplary device 2400 the Fig. 24 each storage container 2440, 2450 and 2460 has its own gas supply system 2445, 2455 and 2465, which terminates with a nozzle 2447, 2457 and 2467 near the point of impact 2422 of the electron beam 2415 on the sample 2425.

[0183] The storage containers 2440, 2450, and 2460 can have their own temperature setting and / or control element, allowing both cooling and heating of the respective storage containers 2440, 2450, and 2460. This enables the storage and, in particular, the provision of the precursor gas at the respective optimal temperature (in the Fig. (24 not shown). The control unit 2370 can control the temperature setting elements and the temperature control elements of the storage containers 2440, 2450, and 2460. During the EBID and EBIE processing operations, the temperature setting elements of the storage containers 2440, 2450, and 2460 can also be used to adjust the vapor pressure of the precursor gases stored therein by selecting an appropriate temperature.

[0184] The device 2400 can comprise more than one storage container 2440 for storing two or more precursor gases. Furthermore, the device 2400 can have more than one storage container 2450 for storing two or more etching gases (in the Fig. 24 not shown).

Claims

[1] Method (1900) for repairing at least one defect (240, 750, 940, 1140, 1240, 1330, 1630, 1840) of a lithographic mask (200, 400, 500, 600, 700, 800, 900, 1100, 1200, 1300, 1600, 1800), the method comprising the step: Determining (1920) parameters of at least one repair method for the at least one defect, wherein determining parameters comprises: assigning at least one numerical value to a parameter, wherein the numerical value differs from the numerical value specified for that parameter by the at least one defect, wherein determining the parameters is based on at least one element of the group: an exposure setting with which the lithographic mask is exposed in operation, design data of the lithographic mask, refractive index data of a deposit for repairing a defect of missing material, and resolution-enhancing RET techniques of the lithographic mask, and where the deviation of the numerical value of this parameter is selected from a range whose lower limit is greater than zero and whose upper limit is less than a resolution limit of a defect-free area of ​​the lithographic mask. [2] Method (1900) according to the preceding claim, wherein the at least one parameter comprises at least one element from the group: at least one lateral dimension of the at least one repair shape, one height dimension of the at least one repair shape, one distance of the at least one repair shape from the at least one defect, one material composition of the at least one defect, one geometric shape of the at least one repair shape, and one environment of the at least one defect on the lithographic mask. [3] Method (1900) according to one of the preceding claims, wherein the at least one parameter deviates by a predetermined amount from the numerical value specified for that parameter by the at least one defect. [4] Method (1900) according to one of the preceding claims, wherein the deviation of the numerical value of the at least one corresponding parameter comprises a range from 2% to 80% of a resolution limit of the lithographic mask. [5] Method (1900) according to one of the preceding claims, wherein a dimension of at least one dimension of the at least one repair form comprises a range of 10% to 90% of a dimension of the corresponding dimension of the at least one defect. [6] Method (1900) according to any of the preceding claims, wherein the at least one defect comprises at least one element from the group: an edge placement error (240) of a pattern element (220), an interrupted and / or a bridged connection of a pattern element, an outlier of an edge roughness of a pattern element, a particle adhering to the lithographic mask, a low-printing lateral defect, a defect residue (340) of a completed defect repair, a sidewall angle error (1850) of a pattern element (1820), and a centroid error of a pattern element and / or of a distance region between two pattern elements. [7] Method according to any of the preceding claims, wherein determining the parameters for the at least one repair method comprises: taking at least one aerial photograph of the at least one defect. [8] Method (1900) according to the preceding claim, wherein taking the at least one aerial image comprises: taking the at least one aerial image of the at least one defect at an actinic wavelength of the lithographic mask and / or taking an aerial image focus stack of the at least one defect. [9] Method (1900) according to any of the preceding claims, wherein the assignment of the numerical value comprises: applying a trained machine learning model to determine the deviation of the numerical value of the at least one parameter from the numerical value specified by the at least one defect for that parameter. [10] Method (1900) according to one of the preceding claims, wherein determining the parameters of the at least one repair mode comprises: applying a trained machine learning model to determine the parameters of the at least one repair mode. [11] Method (1900) according to any of the preceding claims, further comprising the step of: producing at least one repair element (410, 510, 610, 770, 790, 850, 870, 1010, 1170, 1190, 1270, 1290, 1440, 1540, 1640) on the lithographic mask using the determined repair shape. [12] Method (1900) according to the preceding claim, wherein the production of the at least one repair element comprises: performing at least one local etching process and / or performing at least one local deposition process using at least one focused particle beam and at least one precursor gas. [13] Method (1900) according to any one of the preceding claims, comprising the steps: a. Generating at least one repair element (310) by means of at least one repair mode for which the parameters are determined by the at least one defect (240); and b. Determining parameters of a repair form for a remaining defect residue (340), wherein determining parameters for the repair form for the remaining defect residue comprises: assigning at least one numerical value to a parameter that differs from the numerical value specified for that parameter by the remaining defect residue. [14] Computer program comprising instructions which, when executed by a computer system, cause the computer system to perform the method steps of claims 1 to 13. [15] Device (2100) for repairing at least one defect (240, 750, 940, 1140, 1240, 1330, 1630, 1840) of a lithographic mask (200, 400, 500, 600, 700, 800, 900, 1100, 1200, 1300, 1600, 1800), comprising: means for determining parameters of at least one repair method for the at least one defect, wherein the means for determining parameters comprises: means for assigning a numerical value to at least one parameter that differs from the numerical value specified for that parameter by the at least one defect, wherein the means for determining the parameters is based on at least one element of the group: an exposure setting with which the lithographic mask is exposed in operation, design data of the lithographic mask, Refractive index data of a deposit for repairing a defect of missing material and resolution-enhancing RET techniques of the lithographic mask,and wherein the deviation of the numerical value of this parameter is selected from a range whose lower limit is greater than zero and whose upper limit is less than a resolution limit of a defect-free area of ​​the lithographic mask.

Images