Method and apparatus for correcting an imaging error of a particle beam during sample processing

Abstract

The present application relates to a method (800, 900) and an apparatus (100) for correcting at least one imaging defect of a particle beam (127) during the processing of a sample (105). The method (800, 900) comprises the steps of: (a) determining (920) the at least one imaging defect of the particle beam (127) after at least one interruption of the processing of the sample (105); and (b) correcting (930) the determined at least one imaging defect of the particle beam (127) if it exceeds a predetermined threshold.

Description

1. Technical area

[0001] The present invention relates to a method and a device for correcting at least one imaging error of a particle beam during sample processing. 2. State of the art

[0002] Due to the ever-increasing integration density in microelectronics, photolithographic masks must image ever smaller structural elements into a wafer's photoresist layer. To meet these requirements, the exposure wavelength is being shifted to ever shorter wavelengths. Currently, argon fluoride (ArF) excimer lasers emitting at a wavelength of 193 nm are primarily used for exposure purposes. Intensive work is being conducted on light sources emitting in the extreme ultraviolet (EUV) wavelength range (10 nm to 15 nm) and corresponding EUV masks. To increase the resolution of wafer exposure processes, several variants of conventional binary photolithographic masks have been developed simultaneously. Examples include phase masks, phase-shifting masks, and masks for multiple exposure.

[0003] Due to the ever-shrinking dimensions of the structural elements, lithographic masks, photomasks, or simply masks, cannot always be manufactured without visible or printable defects on a wafer. Due to the costly production of photomasks, defective masks are repaired whenever possible. This also applies to microscopic samples or components, such as stamps for nanoimprint lithography.

[0004] Two important groups of defects in lithographic or photolithographic masks are dark defects. These are areas where absorber and / or phase-shifting material is present, but which should be free of this material. These defects are repaired by removing the excess material, preferably using a local etching process.

[0005] Second, there are so-called clear defects. These are defects on the photomask that exhibit greater light transmission during optical exposure in a wafer stepper or wafer scanner than an identical defect-free reference position. In mask repair processes, these defects can be corrected by depositing a material with suitable optical properties. Ideally, the optical properties of the material used for the repair should match those of the absorber or phase-shifting material. Then, the layer thickness of the repaired area can be adjusted to the dimensions of the layer of the surrounding absorber or phase-shifting material.

[0006] Typically, both defect types are local defects whose dimensions are typically in the submicrometer range. These defects are often repaired by particle beam-induced local chemical processes. To repair local defects in a sample, at least one process gas or a precursor gas is provided at the processing location on the sample, where a focused particle beam induces a local chemical reaction. For the local removal of excess material from the sample, the process gas comprises at least one etching gas. In the case of the deposition of missing material, the precursor gas comprises at least one deposition gas or a deposition gas.

[0007] In the following, masks, stamps for nanoimprint lithography, wafers and various types of components to be repaired, such as MEMS (Micro-Electro-Mechanical System), NEMS (Nano-Eletro-Mechanical System) or PICs (Photonic Integrated Circuits) are summarized under the term probe.

[0008] In contrast to mechanical processes, such as sputtering, chemical processes are often slow processes. During the processing of a defect, the sample may heat up locally. Furthermore, micromanipulators used to align the sample with the focused particle beam may drift over extended periods. Furthermore, the local irradiation of a sample with a charged particle beam can locally charge the sample electrostatically. In addition to sample processing, these effects can also occur when examining the sample with a focused particle beam. The charged particle beam therefore cannot impinge on the sample at the intended location.

[0009] All of the effects outlined above ultimately lead to a drift between the intended processing position on the sample and the incident particle beam. This degrades the lateral resolution of the particle beam with respect to a defective area of ​​the sample. This leads, on the one hand, to a deterioration in the (imaging) quality of the sample and, on the other hand, to a reduction in the quality of a sample repair.

[0010] This problem can be mitigated by placing one or more drift correction markers on the sample near a defect to be repaired. The drift correction marker(s) can be used as a reference during sample processing to determine the extent of any drift and subsequently correct it based on this data.

[0011] For example, WO 2012 / 146647 A1 describes the deposition of a reference mark on a photomask using a particle beam, a process gas, and an additional gas, which may be an oxidizing gas. WO 2005 / 0179949 A2 describes the deposition of material on a photomask using an electron beam and TEOS (tetraethyl orthosilicate) or an organic or inorganic precursor gas.

[0012] In addition to a shift in beam position due to drift processes, the beam shape or beam geometry of a focused particle beam can also change over an extended period of time. Temperature changes and the associated material changes, or in the case of semiconducting components, their function, can influence the beam shape of a particle beam. Furthermore, the use of a precursor gas can accelerate the change in the beam geometry. For example, a portion of the precursor gas and / or its reaction products can penetrate a microscope column, such as the column of an electron beam microscope, and change the beam geometry of the particle beam. Furthermore, hysteresis effects can occur in magnetic lenses, which also influence the beam geometry.However, varying the particle beam shape during a defect repair process leads to a deterioration in the lateral resolution and ultimately reduces the quality of the defect repair. Furthermore, a change in the particle beam shape can make it difficult to determine its drift and thus adversely affect its correction.

[0013] The present invention is therefore based on the problem of providing a method and a device for correcting an imaging error of a particle beam during sample processing, which enables improved sample processing. 3. Summary of the invention

[0014] According to one embodiment of the present invention, this problem is at least partially solved by the subject matter of independent claims 1 and 17 of the present application. Exemplary embodiments are described in the dependent claims.

[0015] In one embodiment, the method for correcting at least one imaging error of a particle beam during processing of a sample comprises the steps of: (a) determining the at least one imaging error of the particle beam after at least one interruption of the processing of the sample; and (b) correcting the determined at least one imaging error of the particle beam if it exceeds a predetermined threshold value.

[0016] Typically, aberrations are understood to be deviations from an ideal particle-optical image produced by a real optical or particle-optical system. In this application, aberrations are not related to an ideal particle-optical image, but rather to the best possible image of a real particle-optical system. An image acquired with a particle-optical system with the best possible adjustment is referred to below as a reference image. This reference image serves as a reference quantity in this application, even if the reference image contains a small amount of aberration. The reference quantity is therefore not an ideal image of a reference element, but rather its best possible image with a real particle-optical system.

[0017] In this application, sample processing includes, for example, a controlled sample modification. However, sample processing can also include a sample examination or sample analysis, in which a sample is imaged and / or in which the material composition of the sample is analyzed. The sample or the material composition does not necessarily have to be (significantly) modified.

[0018] A processing process of a sample can be interrupted, e.g., from time to time or in a periodically repeated manner. During the interruption of the processing process, it is determined whether the beam shape or beam geometry of the particle beam has changed during the processing time. If this is not the case, or if the change in the beam shape is smaller than a predetermined threshold, the processing process, e.g., processing or repairing a defect in the sample, is continued without changing the geometry of the particle beam. If, on the other hand, the change in the beam shape of the particle beam exceeds a predetermined threshold, the beam shape of the particle beam is corrected so that the imaging error caused by the changed beam geometry is corrected.This prevents any impairment of the (image) quality and / or deterioration of the lateral resolution of the processing process. As a result, consistent sample processing quality can be ensured, i.e., independent of the examination time and / or processing time.

[0019] The interruption of sample processing and the determination of the at least one imaging error can be performed automatically, and the correction of the determined at least one imaging error can be performed manually. However, it is also possible for the interruption of sample processing and / or the determination of the at least one imaging error to be performed through human interaction, and the correction of the determined at least one imaging error to be performed automatically. In a preferred embodiment, all steps of a method according to the invention are carried out completely automatically.

[0020] The magnitude of an aberration can be determined by determining one or more parameters used in image processing. For example, broadening of structural edges can be used as a measure or error measure. When examining edge profiles, the magnitude of the first and / or second derivative can be used. Alternatively and / or additionally, parameters in Fourier space can be used to determine the at least one aberration.

[0021] In other exemplary embodiments, the method for correcting at least one imaging error of a particle beam during sample processing may alternatively or additionally comprise the steps: after optimizing the particle beam and acquiring at least one reference image of one or more structural elements in the vicinity of the position of the sample to be processed, processing the sample; determining at least one imaging error of the particle beam at least partially after the sample processing or processing of the sample; and correcting the determined at least one imaging error of the particle beam if it exceeds a predetermined threshold value.

[0022] Processing the sample may include using the particle beam to process the sample.

[0023] The particle beam can thus assume a dual role. On the one hand, it can be used to image the sample, for example, a defect in the sample and / or a reference element of the sample. On the other hand, the particle beam can be used to induce a local chemical reaction. As already indicated above, performing the second function can accelerate a change in the beam shape of the particle beam, for example, due to the effect of a process or precursor gas and / or due to a temperature change resulting from sample processing. By performing a method according to the invention, a change in the beam geometry of the particle beam can be corrected regardless of its cause.

[0024] Alternatively, however, it is also possible for the particle beam not to actively participate in sample processing, but rather to be configured to visualize the sample processing. In this embodiment of a method according to the invention, the sample processing can be carried out, for example, by a probe of a scanning probe microscope or, more generally, by a micromanipulator. The particle beam opens up the possibility of monitoring sample processing, for example, in real time, and of intervening in an automated processing process if necessary.

[0025] In addition, a first particle beam, such as an electron beam, can be used to observe or image a sample repair process performed by a second particle beam. The second particle beam can induce a local chemical process and / or modify a sample through its sputtering effect.

[0026] The particle beam may comprise a charged particle beam. The particle beam may comprise a focused particle beam. The particle beam may comprise an element from the group: a photon beam, an electron beam, an ion beam, an atomic beam, or a molecular beam.

[0027] Both electron beams and photon beams, especially photon beams from the deep ultraviolet (DUV) and extreme ultraviolet (EUV) wavelength range, can be finely focused, so that the area within which an electron beam and / or a photon beam initiates a local chemical reaction of at least one precursor gas is very small. The spatial resolution when repairing local clear and / or dark defects is thus high.

[0028] An electron beam may have a resolution in a range of 0.4 nm to 10 nm, preferably 0.5 nm to 8 nm, more preferably 0.6 nm to 6 nm, and most preferably 0.7 nm to 4 nm. Furthermore, the electron beam may have an energy in a range of 0.01 keV to 50 keV, preferably 0.02 keV to 20 keV, more preferably 0.04 keV to 5 keV, and most preferably 0.1 keV to 2 keV. Furthermore, the electron beam may have a beam current in a range of 5 pA to 5000 pA, preferably 10 pA to 1000 pA, more preferably 15 pA to 300 pA, and most preferably 20 pA to 100 pA. These numerical values ​​apply both to the acquisition of an image by scanning the electron beam over a structure and to sample processing with the electron beam and at least one precursor gas. Large beam currents, ieElectron currents in the nano-ampere range are particularly required for material analysis, for example for X-ray analysis, of the material of a defective sample area.

[0029] During processing of the sample, the electron beam may have a dwell time in the range of 10 ns to 0.1 s, preferably 100 ns to 10 ms, more preferably 1 µs to 1 ms, and most preferably 10 µs to 0.1 ms. Furthermore, the electron beam may have a repetition time (period with which the electron beam impinges on the same spot on the sample) in the range of 1 µs to 1 s, preferably 10 µs to 0.5 s, more preferably 0.1 ms to 0.2 s, and most preferably 1 ms to 0.1 s.

[0030] Processing the sample may comprise providing at least one precursor gas on the sample.

[0031] Preferably, providing the at least one precursor gas comprises providing the at least one precursor gas locally at the processing location of the sample, for example, at the processing location of a defect in the sample. A processing process of the sample may include a local etching process and / or a local deposition process. Sample processing may repair a defect in the sample.

[0032] Interrupting sample processing may at least comprise interrupting the provision of the at least one precursor gas. Furthermore, interrupting sample processing may comprise changing at least one parameter of the particle beam before performing its second function. However, changing the gas supply may be omitted if the defect repair rate is low, i.e., the time required for sample repair is long compared to the time required to determine and correct an imaging error of the particle beam.

[0033] Determining the at least one imaging error of the particle beam may comprise: recording at least one image of at least one reference element of the sample with the particle beam.

[0034] To acquire at least one image of the at least one reference element, it may be necessary to change one or more parameters of the imaging particle beam relative to the processing particle beam. Acquiring at least one image of the at least one reference element of the sample may involve scanning the particle beam over the reference element. Acquiring at least one image may be performed without providing a precursor gas.

[0035] The at least one reference element may comprise at least one element from the group: at least one structural element of the sample, at least one marking for drift correction or at least one defect of the sample.

[0036] If a structural element can be used as a reference element already present on the sample, the process steps of creating and removing the reference element from the sample after sample processing are completed can be eliminated. This also applies if a first defect has a contour, so that it can be used as a reference element when removing a second defect from the sample.

[0037] The sample may comprise a lithographic mask and the at least one structural element may comprise at least one element from the group: at least one absorbing and / or phase-shifting element of a pattern of the lithographic mask, at least one fiducial mark of the lithographic mask or at least one defect of the lithographic mask.

[0038] Typically, pattern elements and / or markings of a mask have contours suitable for use as reference elements. A mask can comprise a transmissive or reflective photomask. A photomask can comprise any mask type.

[0039] A method according to the invention may further comprise the step of depositing a sacrificial layer on the sample and / or a protective layer around a processing location of the sample using a particle beam and at least one precursor gas in the form of a deposition gas.

[0040] Depositing a protective layer on the sample around, or at least partially around, a defect to be treated or corrected protects the area of ​​the sample surrounding the defect from the negative effects of sample processing. If the defect is corrected by deposition of material, the protective layer prevents the inadvertent deposition of material outside the defect. The same applies to a defect that is treated by a local etching process. In this case, a protective layer surrounding the defect prevents etching and thus unintentional changes to the sample outside the defect.

[0041] Deposition of a sacrificial layer near a processing location provides an additional degree of freedom when depositing a marker for drift correction. This allows, for example, the material composition of the marker to be selected independently of the sample material. Furthermore, the sacrificial layer protects the sample from potentially detrimental effects of scanning the particle beam across the sample to acquire an image of the reference element for determining the at least one aberration and / or for determining a drift.

[0042] Creating the sacrificial layer and / or the protective layer may comprise performing a particle beam-induced local chemical reaction.

[0043] A sacrificial layer and / or a protective layer may have a material composition that allows easy removal of this layer(s) after sample processing is completed.

[0044] The height of the sacrificial layer and / or the protective layer can range from 5 nm to 2000 nm, preferably from 10 nm to 1000 nm, more preferably from 20 nm to 500 nm, and most preferably from 50 nm to 300 nm. The lateral dimensions of the sacrificial layer can exceed the lateral dimensions of a marker for drift correction by a factor of 1.2, preferably by a factor of 1.5, more preferably by a factor of 2, and most preferably by a factor of 5. The protective layer can have dimensions in the smaller lateral dimension in the range from 5 nm to 1000 nm, preferably 10 nm to 500 nm, more preferably from 20 to 300 nm, and most preferably from 30 nm to 200 nm.

[0045] The deposition of the sacrificial layer, the protective layer, and / or material onto a defect-missing material may occur at a rate of 0.01 nm / s to 1 nm / s, preferably 0.02 nm / s to 0.5 nm / s, more preferably 0.04 nm / s to 0.3 nm / s, and most preferably 0.05 nm / s to 0.15 nm / s. The etching of the sacrificial layer, the protective layer, and / or excess defect material may occur at a rate of 0.01 nm / s to 1 nm / s, preferably 0.02 nm / s to 0.5 nm / s, more preferably 0.04 nm / s to 0.3 nm / s, and most preferably 0.05 nm / s to 0.15 nm / s.

[0046] The particle beam used to deposit the sacrificial layer and / or the protective layer can be used to deposit the sample. However, it is also possible to use a different particle beam for this purpose. The deposition of the sacrificial layer and the protective layer can be carried out using a deposition gas or a deposition gas combination. However, it is also possible to use a specific deposition gas or a specific deposition gas combination for each of the layers.

[0047] The at least one deposition gas for depositing the sacrificial layer and / or the protective layer may contain carbon. Layers containing a high proportion of carbon can be removed from a sample using a wet-chemical process. Such cleaning processes are typically used to clean the processed sample after sample processing.

[0048] The at least one reference element can have a material contrast to the sample and / or the sacrificial layer. The at least one marking for drift correction can have a material composition that differs from the material composition of the sample and / or the sacrificial layer.

[0049] A drift correction marker can comprise molybdenum (Mo), carbon (C), and oxygen (O) as its main components. If an electron beam is used as the particle beam, markers with this material composition provide not only topographic contrast but also material contrast, typically with respect to both a sample and a sacrificial layer underlying the marker. For example, in the absence of an oxidizing agent during deposition of a drift correction marker, the latter can have a material composition of: Mo 10%-15% C 70%-75% O 15%When using an effective oxidizer, a deposited drift correction marker can essentially comprise MoO3. This illustrates that a drift correction marker typically always contains at least as much carbon or oxygen as molybdenum.

[0050] The term “essentially” here – as elsewhere in this description – means an indication of a measured quantity within the usual error limits, using state-of-the-art measuring technology to measure the quantity.

[0051] Markers for drift correction can take many different forms. Small markers with sharply defined contours, such as cylindrical or conical structures, are preferred. Markers with easily produced topographical shapes, such as cuboids or cubes, are also advantageous. Preferably, several markers for drift correction are arranged around a sample location to be repaired. Two to five markers have also proven to be advantageous from an economic perspective.

[0052] The height of a drift correction mark can range from 1 nm to 1000 nm, preferably from 5 nm to 500 nm, more preferably from 20 nm to 300 nm, and most preferably from 40 nm to 100 nm. The lateral dimensions of a drift correction mark are preferably smaller than the structural elements of a sample. This means that when a photomask is used as a sample, the lateral dimensions of a mark should be smaller than the smallest dimension of the pattern elements of the mask. The lateral dimensions of a drift correction mark can range from 1 nm to 500 nm, preferably from 2 nm to 250 nm, more preferably from 5 nm to 150 nm, and most preferably from 10 nm to 75 nm.

[0053] A reference element and / or a marker for drift correction always exhibits a topographical contrast in the particle beam image with respect to the sample or a sacrificial layer on which the marker is applied. If the reference element or marker for drift correction has a different material composition than the sample or sacrificial layer, the image recorded by the particle beam additionally contains a material contrast. This facilitates the determination of an imaging error in the particle beam.

[0054] The at least one marking for drift correction can have a material contrast to the sample and / or the sacrificial layer.

[0055] The at least one reference element can be arranged within a scanning range of the particle beam. The particle beam-induced local chemical process can generate the at least one marking for drift correction within a scanning range of the particle beam. A maximum scanning range is defined below as an area with dimensions in the micrometer range, approximately 20 µm x 20 µm or 50 µm x 50 µm.

[0056] When arranging or attaching a reference element, two boundary conditions must be observed. If the distance of the reference element from the processing location is too small, the reference element may be affected by the sample processing, which may reduce its function as a reference for correcting aberrations in the particle beam and / or for drift correction. On the other hand, if the distance of the marker from the processing location is too great, i.e. lies outside the scanning range of the particle beam, position-dependent aberrations such as coma and / or field curvature cannot be correctly detected and corrected because the sample has to be moved relative to the particle beam in order to scan the marker. Moving the sample relative to the particle beam has a detrimental effect on the precision with which at least the drift of the particle beam can be determined. In addition, large distances between the marker and the reference element can result.The distance between the reference elements and the defective sample site can negatively impact positioning accuracy. Advantageous distances are in the range of 0.2 µm to 100 µm.

[0057] Creating the at least one drift correction mark may comprise at least one element from the group: depositing the at least one drift correction mark on the sample, depositing the at least one drift correction mark on the sacrificial layer, etching the at least one drift correction mark into the sample, and etching the at least one drift correction mark into the sacrificial layer. A drift correction mark etched into a sacrificial layer and / or the sample may have similar dimensions to a mark deposited on the sample and / or the sacrificial layer.

[0058] If a marking is produced on an absorbing pattern element of a photomask whose lateral dimensions are significantly smaller than the smaller lateral dimension of an absorbing pattern element and this marking can be deposited centrally on the pattern element, the marking can remain on the sample after completion of processing, since it does not substantially interfere with the operation of the photomask. In all other cases, it is advantageous to apply the at least one marking for drift correction to a sacrificial layer that is designed such that it can be easily removed from the sample, e.g., a photomask, using a wet-chemical cleaning process. This allows the at least one marking for drift correction to be removed from the sample together with the sacrificial layer. This can be achieved particularly easily if the marking is produced as a structure etched into the sacrificial layer.

[0059] A method according to the invention can further comprise the step of carrying out a local particle beam-induced local chemical reaction to generate the at least one drift-correcting marking. To generate the at least one drift-correcting marking, the particle beam whose imaging errors are corrected can be used. However, it is also possible to use a particle beam comprising a different type of particle to generate the at least one drift-correcting marking. For example, the imaging errors of an electron beam used for sample processing can be corrected, and the sacrificial layer, the protective layer, and / or the at least one drift-correcting marking can be deposited using a photon beam or an ion beam.

[0060] The sample may comprise a lithographic mask, and the at least one structural element may comprise at least one element from the group consisting of a pattern element of the lithographic mask and a fiducial mark of the lithographic mask. Furthermore, the sample may comprise an imprinting die of nanoimprint lithography, and the at least one reference element may comprise a structural element of the imprinting die.

[0061] Determining the at least one aberration may comprise: analyzing the at least one captured image with respect to at least one reference.

[0062] The at least one reference can comprise at least one element of the group: at least one reference image of the at least one reference element, stored data determined from the at least one reference image of the at least one reference element, at least one reference image recorded by the particle beam, stored data determined from the at least one reference image recorded by the particle beam, a database with at least two images of the at least one reference element recorded with a particle beam having at least one defined imaging error, or a database with stored data determined from the at least two images of the at least one reference element recorded with a particle beam.

[0063] A reference image of a reference element is an image generated by a particle beam from the reference element, wherein the aberration(s) is / are minimal. The at least one reference image of the at least one reference element can include a reference image that appears as if it were recorded with the particle beam. This means that a reference image of the at least one reference element can be recorded by another, such as a second, particle beam and transformed into a reference image that appears as if the particle beam had recorded the reference image.

[0064] A method according to the invention may further comprise systematically changing at least one parameter of the particle beam before recording at least one of the at least two images of the at least one reference element for generating the at least one defined imaging error.

[0065] Images of a reference element can be stored in a database. During acquisition, one or more parameters of the particle beam are systematically varied. This changes the beam shape of the particle beam in a defined manner, so that the particle beam generates images of the reference element with defined aberrations. The database can contain images that include individual aberrations of varying severity. Furthermore, the database can contain images in which the particle beam exhibits a combination of two or more aberrations of varying severity.

[0066] For these images, it is known which parameter change(s) relative to the best possible setting generated the corresponding image. By identifying the image stored in the database that most closely matches the measured image of the reference element, the parameter change(s) that eliminate the aberrations can be determined. This allows the particle beam to generate a reference image of the reference element with the changed parameter settings when reacquiring an image of the reference element.

[0067] Creating a database can be performed prior to sample processing. During the processing process, only an image comparison is required to determine the parameters that correct the particle beam's aberration(s). This type of aberration correction therefore has little impact on the processing process.

[0068] Correcting the determined at least one aberration may comprise employing at least one element from the group: an auto-correction function and a trained machine learning model.

[0069] The auto-correction function may comprise at least one algorithm designed to determine the imaging error(s) of the particle beam by comparing the recorded image of the at least one reference element with a reference image of the at least one reference element.

[0070] Executing the algorithm underlying the autocorrection function is computationally intensive and therefore takes some time. The algorithm must be executed during sample processing. In particular, only after the algorithm has been executed can a decision be made as to whether the particle beam actually needs correction. Therefore, this method of correcting aberrations can increase sample processing time.

[0071] The trained machine learning model can be provided with at least one element from the group of input data: a material composition of the sample, a sample processing method, a particle beam type, at least one parameter of the particle beam, at least one component of the at least one precursor gas used, or at least one environmental parameter of the sample processing. Furthermore, the machine learning model can be trained using data from similar or identical processing or repair processes.

[0072] The environmental parameters may include at least one element from the group: a temperature at the sample processing location, a pressure at the sample processing location, or a humidity at the sample processing location.

[0073] The trained machine learning model can determine a time interval of a sample processing loop iteration. A time interval of the loop iteration can comprise the time interval between the start of sample processing and the first interruption of sample processing, or more generally, the nth time interval can comprise the time interval between the (n-1)th and nth interruptions of sample processing. The trained machine learning model can determine the nth time interval from the n-1 input data of the previous sample processing.

[0074] Training a machine learning model may include: training a machine learning model with a number of aberration-containing images with associated input data and a number of reference images with associated input data.

[0075] A machine learning model generates knowledge from experience. A machine learning model learns from examples provided to the model in the form of training data during a learning or training phase. This allows internal model variables, such as parameters of a parametric mapping, to be assigned appropriate values ​​to describe relationships in the training data. As a result, the machine learning model typically does not simply memorize the training data during the training phase, but rather identifies patterns and / or regularities in the training data.

[0076] The quality of the learned relationships is typically assessed based on validation data in order to evaluate the generalizability of the trained model to new data, i.e., data unknown during training. A trained machine learning model can be applied to a system for correcting aberrations in a particle beam in order to predict the potential effects of changes in the beam shape of the particle beam in an image of the reference element that is unknown to the machine learning model when the corresponding input data is provided. A successfully trained or trained machine learning model, i.e., a trained machine learning model with good generalizability, is therefore able, after completion of the training phase, to evaluate unknown images of the reference element when the corresponding input data is provided.

[0077] A machine learning model may include at least one element from the group: a parametric map, an artificial neural network (ANN), a deep neural network (DNN), a time-delay neural network, a convolutional neural network (CNN), a recurrent neural network (RNN), a long short-term memory (LSTM) network, and a generative model.

[0078] Correcting the determined imaging error may comprise changing at least one parameter of the particle beam, so that the particle beam, with the at least one changed parameter, essentially generates a reference image when recording an image of the at least one reference element.

[0079] Determining at least one aberration of the particle beam can be performed by acquiring and analyzing one or more images, such as the reference mark(s). To minimize disruption to the machining process by determining the aberration, the acquisition time for the one or more images should be as short as possible. Furthermore, the time required to analyze the image(s) to determine the aberration should be kept as short as possible. Autocorrection processes that require a small number of steps are therefore preferred.

[0080] A method according to the invention can further comprise the step of continuing sample processing without correcting the particle beam if its aberration is smaller than a predetermined threshold value, or correcting the at least one aberration of the particle beam and continuing sample processing with the particle beam with at least one changed parameter if its aberration is greater than or equal to a predetermined threshold value. A threshold value can be defined, for example, as a percentage of deterioration relative to an image acquired with the best possible settings of a particle beam. The deterioration of the image can be related to the increase in an aberration, such as the dominant aberration, or to two or more different types of aberration, such as defocusing, astigmatism, coma, etc.For example, a threshold value can be defined as a doubling of the aberration of an image relative to a reference image, ie with the best possible settings of the particle beam parameters.

[0081] The at least one aberration of the particle beam may comprise at least one element from the group: a focus change, an astigmatism change or a coma change.

[0082] An aberration can exhibit deviations from the optimal setting in at least two directions. For example, a focus change can result in overfocusing or underfocusing. It is often difficult to detect which direction of the deviation is (are) responsible for the detected aberration(s). However, this is necessary in order to correct the aberration(s). In this context, the images stored in the database can be helpful in determining the parameter change(s) of the particle beam that correct the detected aberration(s).

[0083] The at least one aberration may include a drift of the particle beam with respect to the sample and an element from the group: a focus change, an astigmatism change or a coma change.

[0084] Acquiring a single image of at least one reference element makes it possible to determine both the extent of the particle beam's aberrations and its drift relative to the sample. Thus, implementing a method according to the invention can significantly improve the quality of sample processing.

[0085] The at least one precursor gas may comprise at least one element from the group: at least one etching gas, at least one deposition gas or at least one additive gas.

[0086] The at least one precursor gas may comprise a first deposition gas for depositing the at least one marker for drift correction, the at least one sacrificial layer, and / or the at least one protective layer. The at least one first deposition gas may comprise at least one element from the group: metal alkyls, transition element alkyls, main group alkyls, metal carbonyls, transition element carbonyls, main group carbonyls, metal alkoxides, transition element alkoxides, main group alkoxides, metal complexes, transition element complexes, main group complexes, or organic compounds.

[0087] The at least one precursor gas may comprise a second deposition gas for generating the at least one marker for drift correction. The at least one second deposition gas may comprise at least one element from the group: tetraethyl orthosilicate (Si(OC2H5)4), metal carbonyls, transition element carbonyls, or main group carbonyls.

[0088] The at least one metal carbonyl may 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(CO8), triruthenium dodecarbonyl (Ru3(CO) 12 ), and iron pentacarbonyl (Fe(CO)5). The at least one main group element alkoxide can comprise at least one element from the group: tetraethyl orthosilicate (Si(OC2H5)4, TEOS), tetramethyl orthosilicate (Si(OCH3)4, TMOS), or titanium tetraisopropoxide (Ti(OCH(CH3)2)4).

[0089] The at least one precursor gas may comprise at least one additive gas from the group: an oxidizing agent, a halide, or a reducing agent.

[0090] The oxidizing agent may comprise at least one element from the group: oxygen (O2), ozone (O3), water vapor (H2O), hydrogen peroxide (H2O2), nitrous oxide (N2O), nitrogen oxide (NO), nitrogen dioxide (NO2), or nitric acid (HNO3). The halide may comprise at least one element from the group: chlorine (Cl2), hydrochloric acid (HCl), xenon difluoride, hydrofluoric acid (HF), iodine (I2), hydrogen iodide (HI), bromine (Br2), hydrogen bromide (HBr), nitrosyl chloride (NOCl), phosphorus trichloride (PCl3), phosphorus pentachloride (PCL5), or phosphorus trifluoride (PF3). The reducing agent may comprise at least one element from the group: hydrogen (H2), ammonia (NH3), or methane (CH4).

[0091] The first deposition gas may comprise molybdenum hexacarbonyl (Mo(CO)6), the at least one additive gas may comprise nitrogen dioxide (NO2), and the second deposition gas may comprise tetraethyl orthosilicate (Si(OC2H5)4) or chromium hexacarbonyl (Cr(CO)6).

[0092] The at least one reference element may comprise at least one marker for drift correction, and a method according to the invention may further comprise the step of removing the at least one marker for drift correction from the sample.

[0093] The removal of the at least one sacrificial layer and the at least one drift correction marker can be performed in a single process step. The removal of the at least one sacrificial layer and / or the at least one drift correction marker can be performed in a wet-chemical cleaning process of the sample.

[0094] A computer program may comprise instructions that cause a computer system to carry out the method steps of the above-described aspects of a method according to the invention when the computer program is executed by the computer system.

[0095] A computer-implemented method for correcting at least one imaging error during sample processing, comprising: (a) receiving at least one image of at least one reference element; (b) comparing the at least one received image of the at least one reference element with at least one or more stored images of the at least one reference element to automatically determine the at least one imaging error; and (c) automatically changing at least one parameter of the particle beam such that the particle beam with the at least one changed parameter has an imaging error that is smaller than a predetermined threshold value.

[0096] A method according to the invention can be implemented in such a way that it can run without human interaction. This makes the method according to the invention objectified, i.e., freed from subjective human assessment. Furthermore, the method can be integrated into a sample processing process in such a way that the sample processing time is only minimally increased.

[0097] In one embodiment, the device for correcting at least one imaging error of a particle beam during processing of a sample comprises: (a) means for determining the at least one imaging error of the particle beam after at least one interruption of the processing of the sample; and (b) means for correcting the determined at least one imaging error of the particle beam if it exceeds a predetermined threshold value.

[0098] The device may further comprise at least one element from the group: means for providing at least one precursor gas on the sample, means for directing and controlling a gas flow rate of the at least one precursor gas, means for generating, focusing, and scanning the particle beam, means for adjusting a focus, means for correcting astigmatism and coma error of the particle beam, means for adjusting an objective lens used to focus the particle beam, means for displacing the sample in the beam direction of the particle beam, means for adjusting a stigmator, means for positioning the particle beam relative to a multi-hole aperture, means for positioning the multi-hole aperture relative to the particle beam, or means for positioning the particle beam inside and / or outside a particle beam column.

[0099] A device according to the invention may further comprise: a stigmator for changing an astigmatism of the particle beam.

[0100] The stigmator may comprise at least one electric and / or magnetic multipole element that generates an electric and / or a magnetic field directed substantially perpendicular to the beam direction of the particle beam.

[0101] The electric and / or magnetic multipole element of the stigmator can generate a quadrupole field.

[0102] The electrical and / or magnetic multipole element may comprise a quadrupole, a hexapole, an octupole and / or a dodecapole element.

[0103] Coma errors can be corrected by positioning the particle beam within the particle beam column. A drift in the particle beam's impact position on the sample can be corrected by positioning the particle beam outside the particle beam column.

[0104] A device according to the invention may further comprise a multi-hole diaphragm for generating a predetermined particle stream.

[0105] The multi-hole aperture can comprise two to ten, preferably three to nine, more preferably four to eight, and most preferably five to seven apertures or aperture holes. The aperture widths of the various holes or apertures of the multi-hole aperture can vary. The particle beam can be shaped by selecting the aperture with the appropriate aperture width through which the particle beam passes.

[0106] The opening width of at least one of the apertures of the multi-hole aperture can be adjustable in a range from 0.1 µm to 1000 µm, preferably 0.2 µm to 500 µm, more preferably 0.5 µm to 200 µm, and most preferably from 1 µm to 100 µm.

[0107] The multi-hole aperture can be arranged in the microscope column in front of the objective lens in the beam direction. The multi-hole aperture can be used to adjust the size of the particle beam current.

[0108] A device according to the invention may further comprise an extractor aperture and / or an anode aperture at the outlet of the particle gun.

[0109] The device can be configured to carry out the method steps according to one of the aspects specified above.

[0110] The device may further comprise a computer system with a computer program as described above.

[0111] Furthermore, the device may include one or more algorithms configured to interrupt sample processing, determine an aberration, and correct the determined aberration. The one or more algorithms may be implemented in hardware, software, firmware, or a combination thereof. 4. Description of the drawings

[0112] In the following detailed description, currently preferred embodiments of the invention are described with reference to the drawings, wherein Fig. 1 The upper part of the image schematically represents a block diagram of some important components of a device which can be used to correct one or more aberrations of a particle beam during sample processing and which can additionally be used to determine and correct a drift of the particle beam, and the lower part of the image represents a schematic section through a cleaning device for a sample; Fig. 2 shows a schematic plan view of a section of a photomask as an example of a sample, which has a defect of excess material on the mask substrate; Fig. 3 the section of the Fig. 2 after an interruption of the defect processing and a renewed scanning of the marking for drift correction to determine an imaging error and / or a drift of the particle beam; Fig. 4 the section of the Fig. 3 shows the drift correction of the focused particle beam before continuing sample processing; Fig. 5 shows a plan view of an exemplary reference element in the form of three sharply imaged point-like structures; Fig. 6 the reference element of the Fig. 5, which was recorded with an aberration of the focused particle beam in the form of defocusing; Fig. 7 the reference element of the Fig. 5, taken with a focused particle beam exhibiting an aberration in the form of astigmatism; Fig. 8 schematically illustrates an exemplary method for correcting aberrations of a focused particle beam and a drift of the focused particle beam during sample processing; and Fig. 9 shows a flowchart of essential steps of the method for correcting at least one imaging error of the particle beam during sample processing. 5. Detailed description of preferred embodiments

[0113] In the following, currently preferred embodiments of a device according to the invention for correcting imaging errors and / or a drift of a particle beam during sample processing are explained in more detail using the example of a modified scanning electron microscope. However, a device according to the invention is not limited to the example described below. As a person skilled in the art will readily recognize, any scanning particle microscope that uses, for example, a focused ion beam and / or a focused photon beam as an energy source can be used instead of the scanning electron microscope discussed. Furthermore, a method according to the invention is not limited to the use of the photomask discussed below as an example as a sample. Rather, it can be used to repair any lithographic mask.Furthermore, the application of a method according to the invention is not limited to photomasks. Rather, it can be used to repair stamps or templates for nanoimprint lithography or generally to correct microstructured elements, such as wafers, MEMS, NEMS, or PICs. Furthermore, it is possible to use a device according to the invention, in addition to or independently of sample processing, to analyze the material composition of a sample. For example, some analysis methods, such as energy-dispersive X-ray spectroscopy (EDX), require the provision of larger, high-resolution electron quantities or electron doses over an extended period of time in order to obtain spatially resolved information about the material composition of a sample.

[0114] The Fig. 1 schematically shows essential components of a device 100 that can be used to correct imaging errors of a particle beam and / or to correct a drift of the particle beam during a processing process of a sample 105. The sample 105 can be any microstructured component or part. For example, the sample 105 can comprise a transmissive or reflective photomask and / or a template for nanoimprint lithography. Furthermore, the device 100 can be used to process, for example, an integrated circuit, a wafer, a micro-electro-mechanical system (MEMS), a nano-electro-mechanical system, and / or a photonic integrated circuit (PIC) that have defects, for example, missing and / or excess material. In the examples explained below, the sample 105 is a lithographic mask.

[0115] The exemplary device 100 of the Fig. 1 is a modified scanning electron microscope (SEM). An electron gun 115 generates an electron beam 127, which is directed by elements 120 and 125 as a focused electron beam 127 onto the lithographic mask arranged on a sample stage 110.

[0116] The sample table 110 has micromanipulators (in the Fig. 1 not shown), with the aid of which a defective area of ​​the photomask 105 can be brought below the point of impact of the electron beam 127 on the photomask 105. In addition, the sample table 110 can be displaced in height, ie in the beam direction of the electron beam 127, so that the focus of the electron beam 127 comes to lie on the surface of the photomask (in the Fig. 1 (also not shown). By adjusting the height of the sample stage 110, the electron beam 127 can be focused. However, this also means that slippage or drift in the height adjustment of the sample stage 110 results in defocusing of the electron beam 127. The height of the sample 105 can be varied in both directions, i.e., upwards or downwards, which can lead to overfocusing or underfocusing. In addition to adjusting the height of the sample stage 110, the focusing of the electron beam 127 can also be reversed or adjusted by changing the focusing effect of the objective lens of the device 100.

[0117] Furthermore, the sample table 110 may include a device for adjusting and controlling the temperature, which allows the mask to be brought to a predetermined temperature and to be kept at this temperature (in the Fig. 1 not specified). In addition, the temperature and other parameters that characterize the environment of the device 100 and / or the interior of the particle beam column, including its electronic components, such as air pressure and / or humidity, can be measured and controlled. A liquid cooling system and / or a refrigeration machine (technically known as a chiller) can be used for this purpose.

[0118] The device 100 typically comprises several apertures 182, 184 or an aperture system 180. Behind the electron gun 115, an extractor aperture and an anode aperture are mounted in the column of the device 100, which in the Fig. 1 by the reference numeral 182. In the Fig. In the exemplary device 100 shown in Figure 1, the aperture system 180 further comprises an aperture 184 having a plurality of openings with different diameters. Typically, the aperture 184 or the multi-hole aperture 184 is permanently installed in the microscope column. Depending on the operating point to be set (such as the sample current, possibly the resolution and / or the depth of field), first deflection systems guide the electron beam 127 through the corresponding opening of the multi-hole aperture 184. One or more second deflection systems, which are arranged below the multi-hole aperture 184, deflect the electron beam 127 leaving the corresponding opening of the multi-hole aperture 184 such that it passes through the center of the objective lens 125. The first and second deflection systems or the first and second deflection systems can, for example, comprise electrical and / or magnetic deflection systems. The deflection systems are shown in the Fig. 1 is suppressed. By passing the electron beam 127 through the center of the objective lens 125, the coma error of the electron beam 127 can be minimized.

[0119] In an alternative embodiment, however, it is also possible to design the multi-hole aperture 184 to be movable perpendicular to the beam direction and to guide the electron beam 127 or the particle beam 127 through the opening of the multi-hole aperture 184 with the corresponding width or diameter by displacing the multi-hole aperture 184 in a plane perpendicular to the beam direction of the electron beam 127. The at least one first and the at least one second deflection system and / or the lateral displacement of the multi-hole aperture can be adjusted by the control unit 145 of the device 100. The size of the particle stream can be adjusted via the selected opening width of the multi-hole aperture 184. By guiding the electron beam 127 through the center of the objective lens 125 regardless of its beam intensity, its coma error can be minimized.

[0120] The distance between the apertures 182 (extractor aperture and / or anode aperture) and the multi-hole aperture 184 of the aperture system 180 in the beam direction of the electron beam 127 can be 1 cm, preferably 2 cm, more preferably 5 cm, and most preferably 10 cm. The number of aperture openings can range from two to 20. Currently, five to seven aperture openings of the multi-hole aperture 184 are preferred. The aperture diameters of the multi-hole aperture 184 preferably have numerical values ​​in the range of 1 µm to 50 µm. It is advantageous to guide the particle beam 127 through the opening of the multi-hole aperture 184 in a precisely positioned manner so that it can pass the optical center of the subsequent objective lens 125. This allows the particle beam 127 impinging on a sample 105 to be essentially coma-free.

[0121] Furthermore, the device 100 has a stigmator which is arranged in the Fig. 1 is not shown. The stigmator comprises an octupole element that generates an electric field in the form of a quadrupole. The stigmator is used to correct aberrations of the electron beam 127 in the form of astigmatism.

[0122] The device 100 of the Fig. 1 uses an electron beam 127 as an energy source 127 to initiate a local chemical reaction of a precursor gas or a mixture of two or more precursor gases. An electron beam 127 can be focused to a small focal spot with a diameter of < 10 nm, preferably < 5 nm. Furthermore, electrons impinging on the surface of the photomask cause little damage to the mask 105, even if their kinetic energy varies over a wide energy range. However, the device 100 and the methods presented here are not limited to the use of an electron beam 127. Rather, any particle beam can be used that is capable of locally inducing a chemical reaction of a precursor gas at the point of impact of the particle beam on the surface of a sample 105, such as the photomask 105, and that can generate image data from a sample surface.Examples of alternative particle beams are an ion beam, an atom beam, a molecular beam, and / or a photon beam. Furthermore, it is possible to use two or more particle beams in parallel. In particular, it is possible to simultaneously use an electron beam 127 and a photon beam or an ion beam as the energy source 127 (in the . Fig. 1 not shown).

[0123] The electron beam 127 can be used to capture an image of the photomask, particularly a defective area of ​​the photomask. A detector 130 for detecting backscattered electrons (BSE) and / or secondary electrons (SE) provides a signal proportional to the surface contour and / or the composition of the photomask.

[0124] By scanning or rasterizing the electron beam 127 over the photomask with the aid of a control unit 145, a computer system 140 of the device 100 can generate an image of the photomask. In particular, with the best possible parameter settings of the electron beam 127, by rasterizing the electron beam 127 over one or more reference elements present on the sample 105, a reference image of the reference element(s) can be generated before starting a processing process of the sample 105, ie the photomask in the Fig. 1. The best possible setting of the parameters of the electron beam 127 is a setting in which the focused electron beam 127 has the smallest imaging errors or the smallest proportion of aberrations that the device 100 allows.

[0125] In addition to capturing a reference image of one or more reference elements present on the sample 105, a database containing images of one or more reference elements with a defined proportion of aberrations can be generated by systematic parameter mismatching before sample processing begins. The aberrations in the images of this database can be reversed by reversing the defined parameter mismatch, thereby converting the images in the database back into the corresponding reference images. Consequently, the generated database can be used to correct aberrations in a particle beam that it "acquires" during sample processing. For this purpose, the computer system 140 quantifies the image properties used to determine the aberration(s) of the particle beam 127.

[0126] The control unit 145 may be part of the computer system 140, as shown in the Fig. 1, or may be implemented as a separate unit (in the Fig. 1). The computer system 140 may include algorithms implemented in hardware, software, firmware, or a combination thereof that enable an image of a reference element and / or a defect of the sample 105 to be extracted from the measurement data of the detector 130. A screen of the computer system 140 (in the Fig. 1) can display the calculated image. In addition, the computer system 140 can store the measurement data of the detector 130 and / or the calculated image. Furthermore, the control unit 145 of the computer system 140 can control the electron gun 115 and the beam path through the beam-imaging and beam-shaping elements 120 and 125 of the device 100. Control signals from the control unit 145 can further control the movement of the sample stage 110 by micromanipulators (in the Fig. 1 not specified).

[0127] Furthermore, the computer system 140 and / or the control unit 145 may include one or more algorithms stored in a non-volatile memory (in the Fig. 1 not shown) of the computer system 140 and / or the control unit 145. The algorithm(s) can determine the aberrations of the particle beam 127 by comparing a recorded image of one or more reference elements with a reference image of the one or more reference elements. The algorithm(s) can also determine the parameter change(s) that minimize the aberrations of the particle beam 127. Thus, the algorithm(s) can be viewed as an auto-correction function that eliminates the imaging errors of the electron beam 127 as far as possible.

[0128] The electron beam 127 incident on the photomask can electrostatically charge the photomask. This can deflect the electron beam 127, and the spatial resolution when capturing an image of a reference element or a defect and when repairing the defect can be reduced. To reduce electrostatic charging of the photomask, an ion gun 135 can be used to irradiate the surface of the photomask with ions of low kinetic energy. For example, argon ions with a kinetic energy of a few hundred eV can be used to neutralize the photomask 105. Alternatively and / or additionally, an electrically conductive grid attached to the column outlet can be used for charge compensation.

[0129] In order to process the sample 105 arranged on the sample stage 110 or the photomask, i.e., to repair its defects, the device 100 has at least three reservoirs for three different processing gases, process gases, or precursor gases. The first reservoir 150 stores a first precursor gas in the form of a deposition gas. For example, a main group element alkoxide, such as TEOS, can be stored in the first reservoir 150, which can be used to repair a defect in the photomask, such as a mask substrate lacking material.

[0130] The second reservoir 155 stores an etching gas, such as xenon difluoride (XeF2). Other exemplary etching gases that may be stored in the second reservoir include: a halogen, such as chlorine (Cl2), fluorine (F2), bromine (Br2), or iodine (I2), oxygen (O2), ozone (O3), and nitrosyl chloride (NOCl).

[0131] The third reservoir 160 stores an oxidizing agent. An oxidizing agent may, for example, include an element from the group consisting of oxygen (O2), ozone (O3), water vapor (H2O), hydrogen peroxide (H2O2), nitrous oxide (N2O), nitrogen monoxide (NO), nitrogen dioxide (NO2), nitric acid (HNO3), and other oxygen-containing compounds.

[0132] The fourth, optional storage container 165 can, for example, store a reducing agent such as hydrogen (H2), ammonia (NH3) and methane (CH4).

[0133] The fifth, also optional, reservoir 170 can store a carbon-containing precursor gas. The precursor gas stored in the fifth reservoir 170 can be, for example, a metal carbonyl, such as chromium hexacarbonyl (Cr(CO)6) or molybdenum hexacarbonyl (Mo(CO)6). A metal carbonyl can be used to deposit a protective layer around a defect to be processed. Furthermore, a metal carbonyl can be used to deposit one or more markers for drift correction, which are used as reference elements during sample processing. In addition, a metal carbonyl, preferably Mo(CO)6, can be used as a deposition gas for depositing a sacrificial layer and / or a protective layer.

[0134] Finally, the sixth, optional reservoir 175 can contain a second etching gas. This can be used to repair defects caused by excess material (dark defects) on the photomask. Furthermore, the etching gas stored in the sixth reservoir 175 can be used to remove the reference elements in the form of markers for drift correction, a sacrificial layer, and / or a protective layer from the photomask.

[0135] Each reservoir 150, 155, 160, 165, 170, 175 has its own control valve 151, 156, 161, 166, 171, 176 to control the amount of the corresponding gas supplied per unit of time, i.e., the gas flow rate at the point where the electron beam 127 impinges on the sample 105. The control valves 151, 156, 161, 166, 171, and 176 are controlled by the control unit 145 of the computer system 140. This allows the partial pressure ratios of the gases supplied at the processing location to be adjusted within a wide range.

[0136] Furthermore, in the exemplary apparatus 100, each reservoir 150, 155, 160, 165, 170, 175 has its own gas supply system 152, 157, 162, 167, 172, 177, which terminates with a nozzle near the point of impact of the electron beam 127 on the photomask 105. In an alternative embodiment (in the Fig. 1 not represented), a gas supply system is used to deliver several or all processing gases in a common stream to the surface of the sample 105.

[0137] In the Fig. 1, the valves 151, 156, 161, 166, 171, 176 are arranged near the corresponding containers 150, 155, 160, 165, 170, 175. In an alternative arrangement, the control valves 151, 156, 161, 166, 171, 176 can be installed near the corresponding nozzles (in the Fig. 1 not shown). Unlike in the Fig. 1 and not presently preferred, it is also possible to provide one or more of the gases stored in the containers 150, 155, 160, 165, 170, 175 in an undirected manner in the lower part of the vacuum chamber 102 of the device 100. In this case, it is necessary for the device 100 to have an aperture (in the Fig. 1 not shown) between the lower reaction chamber 107 and the upper part 102 of the device 100, which provides the electron beam 127, in order to prevent an insufficient vacuum in the upper part of the device 100.

[0138] Each of the reservoirs 150, 155, 160, 165, 170, and 175 can have its own temperature adjustment and control element, which allows both cooling and heating of the respective reservoirs. These enable the storage and provision of the precursor gases at the optimal temperature (in the Fig. 1 not shown). This also makes it possible to use the control valves 151, 156, 161, 166, 171 and 176 as digital elements and to control the gas flow or the gas mass flow via heating of the storage containers 150, 155, 160, 165, 170, 175, i.e. via the vapor pressure of the precursor gases. In addition, each supply system 152, 157, 162, 167, 172 and 177 can comprise its own temperature setting element and temperature control element in order to provide all process gases or precursor gases at their optimal processing temperature at the point of impact of the electron beam 127 on the photomask 105 (in the Fig. 1 also not specified). The control unit 145 of the computer system 140 can control the temperature setting elements and the temperature control elements of both the storage containers 150, 155, 160, 165, 170, 175 and the gas supply systems 152, 157, 162, 167, 172, 177.

[0139] The device 100 of the Fig. 1 comprises a pump system for generating and maintaining a vacuum required in the vacuum chamber 102 (in the Fig. 1 not shown). When the control valves 151, 156, 161, 166, 171, 176 are closed, a residual gas pressure of ≤ 10 -7 mbar. The pump system may comprise separate pump systems for the upper part 103 of the device 100 for providing the electron beam 127 and the lower part 107, which contains the sample stage 110 with the sample a05. Furthermore, the device 100 may comprise a suction device near the processing point of the electron beam 127 in order to define a defined local pressure condition on the surface of the sample 105 or the photomask 105 (in the Fig. 1 (not shown). The use of an additional extraction device can largely prevent one or more volatile reaction products of the one or more precursor gases that are not needed for the local deposition of the deposit from depositing on the photomask 105 and / or in the vacuum chamber 102 or the reaction space 107. Furthermore, the extraction device can prevent particles created in an etching process from being distributed in the vacuum chamber 102 of the device 100. The functions of the pump system(s) and the additional extraction device can also be controlled and / or monitored by the control unit 145 of the computer system 140.

[0140] Furthermore, the device 100 may comprise one or more atomic force microscopes (AFM) that allow defects of the sample to be analyzed in detail (in the Fig. 1 (not shown). Furthermore, it is possible for the one or more AFMs to be configured to perform sample processing, and the focused electron beam 127 can be used to observe or monitor the sample processing in situ. For example, a probe of an AFM can be configured to remove a defect of excess material by mechanical processing. Furthermore, a probe of the one or more AFMs can be configured to remove a particle from the sample 105. In addition, a probe can be configured to repair a clear defect by depositing absorbing material.

[0141] Finally, the lower part of the Fig. 1 schematically shows a cleaning device 190 with a cleaning fluid 195, which is used to clean a sample 105 during and / or after completion of sample processing in the device 100. The cleaning fluid 195 can be adapted to the sample 105 to be cleaned. The cleaning fluid 195 can comprise water and / or aqueous solutions, such as diluted sulfuric acid or diluted hydrogen peroxide under the influence of ultrasound or megasonics. Furthermore, the presence of ultraviolet (UV) or infrared (IR) light as well as the presence of diluted gases, such as hydrogen (H2) and / or oxygen (O2), can enhance the cleaning effect of the cleaning fluid.

[0142] In addition, the cleaning device 190 can be used to remove the reference mark(s) present as drift correction mark(s) from the sample 105. If the drift correction mark(s), a sacrificial layer, and / or a protective layer were deposited using, for example, the precursor gas Mo(CO)6, these marks or layers can be removed from the sample 105 using a cleaning fluid 195 containing specific solutions, such as mineral acids, bases, or organic ligands.

[0143] The Fig. Figure 2 schematically shows a top view of a section of the photomask as an example of a sample 105. The section 200 shows a defect 260 on a substrate 210 of the mask, which defect is to be repaired with a particle beam-induced machining process. The defect 260 can be a defect of missing material, a defect of excess material, or a particle present on the mask substrate 210. In the example of Fig. 2, the defect 260 is a defect of excess material, ie a dark defect 260, on the mask substrate 210. In the Fig. In the example shown in Figure 2, the defect 260 is isolated, ie there is no pattern element of the mask in its vicinity that could take over the function of a reference element or on which a reference element could be generated.

[0144] In order to compensate for a drift of the particle beam or the electron beam 127 during sample processing, the cutout 200 has a reference element 230 in the form of a marking 230 for drift correction. Fig. In the example shown in Figure 2, the marking 230 has a circular and / or cylindrical shape.

[0145] To avoid damage during multiple scanning or rasterization of the marking 230 for drift correction to determine the position of the marking 230, the marking 230 or the DC marking 230 (DC for drift correction) is deposited on a sacrificial layer 220 of the mask substrate 210. Furthermore, by means of the sacrificial layer 220, a defined material contrast of the electron beam 127, as an exemplary particle beam 127, can be set between the sacrificial layer 220 and the reference element 230. Furthermore, the material composition of the sacrificial layer 220 can be selected such that the latter can be removed from the photomask 105 together with the reference element 230 in a wet-chemical cleaning step after completion of the sample processing. In the exemplary embodiment of the Fig. 2, the sacrificial layer 220 has a rectangular shape.

[0146] The double arrow in the Fig. 2 symbolizes the reference distance 240 of the DC marking 230 with respect to the defect 260. The coordinates of the DC marking 230 (relative to a coordinate system linked to the sample holder 110) at the beginning of the sample processing define a reference position 250 of the DC marking 230. The coordinates of the two-dimensional vector that describes the direction and magnitude of a drift between the electron beam 127 and the position of the defect 260 are related to the reference position 250 of the DC marking 230. The Fig. The mask section 200 shown in Figure 2 has a DC marking 230. It is of course also possible to generate two, three, four, or even more reference elements 230 on the mask around the defect 260 and to use them to determine a drift and / or aberrations of the particle beam 127.

[0147] The dashed rectangle 270 indicates the Fig. 2 shows the area that the focused electron beam 127 scans to determine the reference position 250 of the reference element 230 before starting the machining process of the defect 260 of the sample 105. In the example of Fig. 2, the scanning area of ​​the particle beam 127 lies entirely on the sacrificial layer 220.

[0148] The Fig. 3 shows the section 200 of the photomask after performing a part or a first part of the sample processing, i.e., the processing process of the defect 260. The dark defect 260 is processed by performing a particle beam-induced local chemical reaction in the form of an etching process, i.e., removed from the mask substrate 210 by etching. Xenon difluoride (XeF2), for example, can be used as the etching gas. Depending on the material composition of the defect 260, an additive gas, such as an oxidizing agent such as oxygen (O2) or chlorine (Cl2), can be added to the etching gas. A combination of two or more etching gases can also be used to remove the defect 260.

[0149] Sample processing in the form of a local chemical reaction is a long-lasting process that can extend over a period of several minutes and, in the case of large-area defects, can reach into the range of hours. Furthermore, sample processing can lead to a local change in the temperature on the sample 105. Furthermore, the use of a precursor gas or a mixture of different precursor gases can change the focus or, more generally, the beam shape of the focused electron beam 127. These effects of sample processing can, on the one hand, cause a drift of the electron beam 127 relative to the defect 260 to be processed. The correction of a drift is carried out using the Fig. 3 and Fig. 4. On the other hand, the described effects can change the beam shape or beam geometry of the focused particle beam 127. The consequences of a beam change and its elimination are explained below on the basis of the Fig. 5 to 7 discussed.

[0150] Due to the influence of the above effects, sample processing is interrupted, and the focused electron beam 127 is shifted by the reference distance 240 to scan the DC mark 230 again without providing a precursor gas. If the positioning of the reference mark, its scanning with the focused electron beam 127, and the determination of the aberration of the focused electron beam 127 are performed within a short period of time, interrupting the supply of the precursor gas can be omitted. This avoids the time required to wait after interrupting the gas flow until it has stabilized after being switched on.

[0151] As in the Fig. 3 by the vector 350, the DC marking 230 has shifted within the scan area 270. This means that the processing window of the focused electron beam 127 has shifted relative to the defect 260 to be etched. In order to examine the drift 350 of the DC marking 230 with respect to its reference position 250 in detail, in the present example the scan area 270 of the electron beam 127 is essentially shifted by the drift vector 350, and the DC marking 230 is scanned again. Starting from the new position of the DC marking 230, the focused electron beam 127 is shifted again by the reference distance 240, and the processing of the defect 260 continues.

[0152] This process is in the Fig. 4. In the Fig. 3 and Fig. 4, the defect 260 is shown in a gray scale, whereas the defect 260 in the Fig. 2 is shown in black. This symbolizes that the defect 260 in the Fig. 3 and Fig. 4 has already been partially removed by etching, ie the Fig. 3 and Fig. 4 show a defect residue of the original defect 260. The sample processing can be interrupted a second time, generally n times, if necessary, to determine drift and / or aberrations of the focused electron beam at the end of the n-th processing interval or at the end of the n-th loop pass.

[0153] The Fig. 5 presents an image 500 of a reference element 530, which was recorded with the best possible parameter setting of the electron beam 127, so that the image contains the lowest possible proportion of aberrations. The image 500 thus represents a reference image 500 of the reference element 530. The reference element 530 has three cylindrical structures 505, 510, 515 arranged at the ends of an isosceles or equilateral triangle. The center of gravity of the reference element 530 can be selected as the reference position 550 of the reference element 530.

[0154] Just like the reference element 230, the reference element 530 is not a pattern element present on the sample 105 or photomask. Rather, the reference element 530 is a marker for drift correction. The reference element 530 can, for example, be the reference element 230 of the Fig. 2. This means that the reference element 530 can be deposited on a sacrificial layer 210 of the sample by an electron beam-induced deposition process using a deposition gas or a deposition gas mixture. However, it is also possible to etch the reference element 530 or the marking 530 into the sacrificial layer 210 for drift correction. Furthermore, the DC marking 530 can be deposited directly on the mask substrate 210 of the mask 105 or etched into its mask substrate 210.

[0155] As already mentioned in the context of Fig. 2, the reference image 500 of the reference element 530 or the DC marking 530 can be used to determine the reference position 550 of the DC marking 530. The reference position 550 of the DC marking 530 can be determined with great precision based on the reference image 500.

[0156] As already mentioned in the context of Fig. 3, after a portion of the processing of the defect 260, the sample processing is interrupted and the DC mark 230 is scanned again with the focused electron beam 127, with or without interrupting the precursor gas flow, to capture an image of the DC mark 230. The Fig. 6 presents an image 600 of the DC marking 530 of the Fig. 5 after interrupting sample processing. In image 600 of the Fig. 6, the DC marking 530 exhibits an increased proportion of aberrations compared to the reference image 500 due to a defocusing of the electron beam 127. The increased proportion of aberrations, represented by the enlarged out-of-focus structures 605, 610, 615 of the cylindrical structures 505, 510, 515 of the reference element 530, reduces the accuracy with which the position 650 of the drift marking 530 can be determined. This, on the one hand, reduces the precision of determining the drift vector 350. On the other hand, the defocused electron beam 127 increases the lateral dimensions of a local etching reaction for removing the defect 260, thereby reducing the lateral resolution of the sample processing.

[0157] By determining the extent of the misfocusing and correcting it, a reference image 500 of the DC marker 530 can be acquired again after interrupting the sample processing. The extent of the drift, illustrated by the two-dimensional vector 350, can be optimally corrected, as described in the context of Fig. 3 and Fig. 4, in which the changes in the beam shape of the focused electron beam 127 were ignored. Furthermore, the beam geometry can be optimized for performing the local etching reaction to remove the defect 260 to perform the second part of the sample processing. A deterioration of the process resolution of a local chemical reaction, a local electron beam-induced etching reaction in the example of Fig. 2 to 4, during sample processing can be prevented. The determination and correction of aberrations that occur as defocusing of the focused electron beam 127 has been explained in the context of the explanation of the device 100. It also explains how the device 100 can correct defocusing of the electron beam 127.

[0158] The Fig. Figure 7 presents a second type of aberration that can degrade the image quality of the focused electron beam 127 during sample processing. When recording the image 700 of the reference mark 530, the focused particle beam 127 exhibited astigmatism. The astigmatism of the focused electron beam 127 images the elements 505, 510, 515 of the DC mark 530 as blurred, enlarged ellipses 705, 710, 715. Determining the position 750 of the DC mark 730 is subject to greater uncertainty. As already explained in the context of Fig. 6, the drift vector 350 can only be determined imprecisely and the beam shape of the focused particle beam 127 is only partially suitable for carrying out a local etching reaction to remove the defect of excess material 260.

[0159] The stigmator of device 100 generates an electric or magnetic quadrupole field, which can be used to correct the astigmatism of the focused electron beam 127. The focused electron beam 127, which acquires an image of the DC marking 530 with modified parameter settings of the stigmator that correct its astigmatism, generates a reference image 500 of the marking 530 for drift correction. After determining the extent of the astigmatism and correcting it using the stigmator of device 100, the extent of a drift 350 can be optimally corrected, and sample processing can continue with the optimized beam geometry of the focused particle beam 127.

[0160] Coma, not shown here, is another type of aberration that can magnify the imaging error of a focused particle beam 127 during sample processing. This type of aberration typically arises from an oblique passage of the particle beam 127 through the objective lens 125 of the device 100. This type of aberration also reduces the accuracy with which the position of a DC marker 530 can be determined. By guiding the particle beam 127 through the optical center of the objective lens 125 of the device 100, a coma of the focused particle beam 127 can be largely avoided, and the beam shape of the focused electron beam 127 can be optimized for another part of the sample processing.

[0161] The Fig. 8 schematically presents a flowchart 800 of a method according to the invention for correcting both aberrations of a focused particle beam 127 and a drift of the particle beam 127 relative to a defect 260 to be processed during sample processing. The method begins at step 805. First, at decision block 810, a check is made as to whether one or more structural elements suitable for use as reference elements 230, 530 are present on the sample 105 to be processed. If this is not the case, one or more reference elements 230, 530 are generated in the vicinity of the defect 260 to be processed at step 815. Typically, one or more reference elements 230, 530 are generated in the form of one or more markings 230, 530 for drift correction using a particle beam-induced etching process and / or deposition process on the sample 105.

[0162] In one possible embodiment of a method according to the invention, in step 820, by imaging the one or more reference elements 230, 530 with a defined misaligned focused particle beam 127, a database is generated with images of the at least one reference element 230, 530 in which the focused particle beam 127 has a defined imaging error. Since the execution of step 820 is only necessary in a specific embodiment, it is Fig. 8 shown in dashed lines.

[0163] In step 825, a reference image 500 of the at least one reference element 530 is generated by scanning the focused particle beam 127 over the reference element 530. This step is also not necessary for all variants of a method according to the invention and is therefore also indicated as optional by the dashed lines. For example, a reference image 500 of the at least one reference element 530 can be stored in a database.

[0164] At step 830, the actual sample processing begins by the combination of a focused particle beam 127 and at least one etching gas etching the defect 260 of the Fig. 2 to 4 etched.

[0165] After a fixed or variable time interval, the sample processing or defect processing is interrupted at step 835. For this purpose, the flow of the etching gas can be stopped if necessary. If necessary, one or more parameters of the focused particle beam 127 are additionally adjusted for scanning the reference element 230, 530.

[0166] In step 840, the focused particle beam 127 acquires an image 600, 700 of the reference element 230, 530 by scanning over the reference element 530. In the next step 845, the reference image 500 of the reference element 530 is compared with the image 600, 700 of the reference element 530 at the end of the first part of the sample processing. From this comparison, the aberration(s) of the focused particle beam 127 are determined at the end of the first loop of the defect repair process. Further possible embodiments for determining the aberrations have already been described above, such as the use of a database with defined aberrations, the use of an autocorrection function, and / or the use of a trained machine learning model.

[0167] At decision block 850, it is then determined whether the determined aberration falls below or exceeds a predetermined threshold. If the determined aberration exceeds the threshold, the aberration is corrected, and the method returns to block 840, where an image 500 of the reference element 530 is again acquired. Options for correcting various aberrations are discussed in the context of the Fig. 1 explained.

[0168] If the aberration(s) of the focused particle beam 127 is smaller than the predetermined threshold value, the two-dimensional vector 350 describing the drift of the focused particle beam 127 relative to the defect 260 to be processed during the first processing interval is determined in step 860. If the drift is also smaller than the associated threshold value, the method returns to block 830 at decision block 875, and the second part of the sample processing is continued with the original parameter values ​​of the focused particle beam 127. If, however, it is determined at decision block 875 that the magnitude of the drift vector 350 is greater than the threshold value of the drift correction, the drift of the particle beam 127 with respect to the reference position 550 of the reference element 530 is corrected in step 875.

[0169] Furthermore, at decision block 880, a decision is made as to whether the remaining defect 250 or its residual defect after the first loop pass exceeds a permissible threshold. If this is the case, the method returns to block 830 and a second loop of defect processing is executed. If, however, it is determined at decision block 880 that the remaining defect no longer unacceptably interferes with the function of sample 105, the method ends with step 885.

[0170] Finally, the flowchart 900 represents the Fig.9 essential steps of a method for correcting an imaging error of a particle beam 127 during sample processing. The method begins at step 910. In step 920, at least one imaging error of the particle beam 127 is determined after at least one interruption of the sample processing. This can be done, for example, by comparing a reference image 500 of a reference element 520 with an image 600, 700 of the reference element 530, which is acquired after an interruption of the sample processing. Furthermore, the determination of the imaging error(s) can be based on a database with defined imaging errors and / or by using an autocorrection function. The use of a trained machine learning model is also possible. The method ends at step 940. QUOTES CONTAINED IN THE DESCRIPTION

[0000] This list of documents submitted by the applicant was generated automatically and is included solely for the convenience of the reader. This list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions. Cited patent literature

[0000] WO 2012 / 146 647 A1

[0011] WO 2005 / 017 9 949 A2

[0011]

Claims

[1] Method (800, 900) for correcting at least one imaging error of a particle beam (127) during processing of a sample (105), comprising the steps: a. determining (920) the at least one imaging error of the particle beam (127) after at least one interruption of the processing of the sample (105); and b. Correcting (930) the determined at least one imaging error of the particle beam (127) if it exceeds a predetermined threshold value. [2] Method (800, 900) according to the preceding claim, wherein processing the sample (105) comprises using the particle beam (127) to process the sample (105). [3] Method (800, 900) according to one of the preceding claims, wherein processing the sample (105) comprises providing at least one precursor gas on the sample (105). [4] Method (800, 900) according to one of the preceding claims, wherein interrupting the processing of the sample (105) comprises at least interrupting the provision of the at least one precursor gas. [5] Method (800, 900) according to one of the preceding claims, wherein determining the at least one imaging error of the particle beam (127) comprises: recording at least one image (500, 600, 700) of at least one reference element (230, 530) of the sample (105) with the particle beam (127). [6] Method (800, 900) according to the preceding claim, wherein the at least one reference element (230, 530) comprises at least one element from the group: at least one structural element of the sample, at least one marking (230, 530) for drift correction or at least one defect (260) of the sample (105). [7] Method (800, 900) according to one of the preceding claims, further comprising the step of: depositing a sacrificial layer (220) on the sample and / or a protective layer around a processing location of the sample (105) with a particle beam (127) and at least one precursor gas in the form of a deposition gas. [8] The method (800, 900) of claim 6, wherein determining the at least one aberration comprises: analyzing the at least one captured image (600, 700) with respect to at least one reference (500). [9] Method (800, 900) according to the preceding claim, wherein the at least one reference (500) comprises at least one element of the group: at least one reference image (500) of the at least one reference element (230, 530), stored data determined from the at least one reference image (500) of the at least one reference element (230, 530), at least one reference image (500) recorded by the particle beam (127), stored data determined from the reference image (500) recorded by the particle beam (127), a database with at least two images of the at least one reference element (230, 530) recorded with a particle beam (127) having at least one defined imaging error, or a database with stored data determined from the at least two images of the at least one reference element (230, 530) recorded with a particle beam (127) having at least one defined imaging error,were determined., [10] Method (800, 900) according to the preceding claim, further comprising: systematically changing at least one parameter of the particle beam (127) before recording at least one of the at least two images of the at least one reference element (230, 530) to generate the at least one defined imaging error. [11] Method (800, 900) according to one of the preceding claims, wherein correcting the determined at least one aberration comprises employing at least one element from the group: an auto-correction function or a trained machine learning model. [12] Method (800, 900) according to one of the preceding claims, wherein correcting the determined imaging error comprises changing at least one parameter of the particle beam (127) such that the particle beam (127) with the at least one changed parameter substantially generates a reference image (500) when recording an image (500, 600, 700) of the at least one reference element (230, 530). [13] Method (800, 900) according to one of the preceding claims, wherein the at least one aberration comprises at least one element from the group: a focus change, an astigmatism change or a coma change. [14] The method (800, 900) according to any one of claims 3 to 13, wherein the at least one precursor gas comprises at least one member from the group: at least one etching gas, at least one deposition gas, or at least one additive gas. [15] Method (800, 900) according to one of claims 5, 6 or one of claims 7-14 dependent on one of claims 5, 6, wherein the at least one reference element (230, 530) comprises at least one marker (230, 530) for drift correction, and the method further comprises the step of: removing the at least one marker (230, 530) for drift correction from the sample (105). [16] A computer program comprising instructions causing a computer system to carry out the method steps of claims 1 to 15 when the computer program is executed by the computer system. [17] Device (100) for correcting at least one imaging error of a particle beam (127) during processing of a sample (105), comprising: a. means for determining at least one imaging error of the particle beam (127) after at least one interruption of the processing of the sample (105); and b. Means for correcting (110, 180) the determined at least one imaging error of the particle beam (127) if it exceeds a predetermined threshold value. [18] Device (100) according to the preceding claim, further comprising at least one element from the group: means for providing (150, 155, 160, 165, 170, 175) at least one precursor gas on the sample (105), means for guiding (152, 167, 162, 167, 172, 177) and controlling (151, 156, 161, 166, 171, 176) a gas mass flow of the at least one precursor gas, means for generating (115), focusing (120, 125) and scanning the particle beam (127), means for adjusting (110, 120, 125, 145, 180) a focus, an astigmatism and a coma of the particle beam (127), means for Adjusting (145) an objective lens (125) used to focus the particle beam (127), means for displacing the sample (105) in the beam direction of the particle beam (127), means for adjusting (145) a stigmator,or means for positioning (145) the particle beam (127) relative to a multi-hole aperture (184) and / or for positioning (145) the multi-hole aperture (184) relative to the particle beam (127)., [19] The apparatus (100) of claim 17 or 18, further comprising: a stigmator for changing an astigmatism of the particle beam (127). [20] The device (100) of claim 17 or 19, further comprising: a multi-hole aperture (184) for generating a predetermined particle stream. [21] The apparatus (100) of any of claims 17-20, further comprising: a computer system (140) having a computer program according to claim 16.

Images