Measuring device, method for operating a mask metrological measuring device and computer program product

Abstract

Method for operating a mask metrological measuring device, in which an image of a section of a photomask (17) is acquired with a first image sensor (20) and in which an aerial image is generated by subjecting the raw image data acquired with the first image sensor (20) to a clear normalization (45), and in which the aerial image is subjected to a non-linearity adjustment comprising the following steps: a. Compiling (47) the aerial image with a clear image (CT2T); b. Applying (48) a linearity correction (Plin1) to the image data generated in step a. to correct a linearity error of the first image sensor (20); c. Applying (49) a non-linearity adjustment (P-1lin2) to the linearity-corrected image data obtained in step b. to imprint on the image data a linearity signature of a second image sensor not located in the beam path of the measuring device; d. Apply (50) a clear normalization to the linearity-adjusted image data generated in step c.

Description

[0001] The invention relates to a measuring device, a method for operating a mask metrological measuring device and a computer program product.

[0002] Photomasks are used in microlithographic projection exposure systems for the fabrication of integrated circuits with extremely small structures. The photomask (= reticulum), illuminated with very short-wavelength, deep ultraviolet or extreme ultraviolet radiation (DUV or EUV radiation), is projected onto a lithography object to transfer the mask structure.

[0003] For high-quality images produced on lithographic objects, it is crucial that photomasks are used that precisely meet the specifications. Mask metrological measuring devices are used to examine photomasks, allowing for an assessment of their dimensional accuracy.

[0004] In the mask metrological measuring device, a section of the photomask is illuminated with radiation emitted from a radiation source, and the illuminated section of the photomask is imaged onto an image sensor. The raw image data acquired by the image sensor is then normalized to a clear image, i.e., an image free of any mask structure. A clear image can be generated, for example, by positioning the photomask at a clear position where the radiation can pass through the glass substrate of the photomask without being affected by any mask structure. An image that has undergone clear normalization is subsequently referred to as an aerial image.

[0005] Document DE 10 2011 113 940 A1 discloses a device and a method for determining dose changes for adjusting the structural dimensions of a mask.

[0006] After several years of operation, it may become necessary to replace the image sensor of the measuring device. Since the non-linearity behavior of two image sensors is generally not identical, the aerial images the user receives after the sensor replacement will differ from those obtained before the replacement. This is undesirable because many users have established routines for further processing the aerial images. These routines can be disrupted if there are discrepancies between the aerial images.

[0007] The invention is based on the objective of presenting a method for operating a mask metrological measuring device, a measuring device, and a computer program product such that the measurement results after replacing the image sensor correspond to the measurement results before replacing the image sensor. This objective is achieved by the features of the independent claims. Advantageous embodiments are specified in the dependent claims.

[0008] In the inventive method for operating a mask metrological measuring device, a first image sensor captures an image of a section of a photomask. An aerial image is generated by subjecting the raw image data acquired with the first image sensor to clear normalization. The aerial image is then subjected to non-linearity adjustment, which comprises the following steps. In step a., the aerial image is combined with a clear image. In step b., a linearity correction is applied to the image data generated in step a. to correct a linearity error of the first image sensor. In step c., a non-linearity adjustment is applied to the linearity-corrected image data obtained in step b. to imprint a linearity signature of a second image sensor not located in the beam path of the measuring device onto the image data. In step d., a clear normalization is applied to the image data obtained in step c.Applied to generated linearity-adjusted image data.

[0009] The invention recognizes that losses in accuracy and repeatability occur when linearity adjustments determined for the old and new image sensors are applied directly to the raw images. The invention proposes correcting the clear-normalized aerial image. This allows for better agreement between measurement results obtained with the old image sensor before the sensor replacement and measurement results recorded with the new image sensor. For the purposes of the invention, the new image sensor corresponds to the first image sensor. The old image sensor corresponds to the second image sensor, which is not located in the beam path of the measuring device.

[0010] Clear normalization relates measured image data to clear image data. Clear image data is obtained by passing the light path through a structureless glass substrate, specifically through a clear position of the photomask. The clear image data thus provides information about what the image sensor sees when the radiation emitted from the light source is directed onto the image sensor without any structure being examined. Clear normalization prevents measurement results from being distorted by image sensor drift and calibrates the illumination field inhomogeneity. Clear normalization is performed pixel by pixel; that is, for each pixel of the image sensor, the ratio of the image data from the acquired image to the image data of the clear image is calculated.

[0011] The inventive processing of the aerial image with a clear image can also be performed pixel by pixel by multiplying the image data measured for each pixel of the image sensor with a clear measurement value of the same pixel. The multiplication can be performed in normalized units, for example on a scale ranging from 0 to 1. The value 1 can correspond to a saturation of the image sensor. The measuring device can be adjusted such that a clear image has an average intensity between 0.5 and 0.9, preferably between 0.6 and 0.8, on the relative scale.

[0012] The clear image used for billing, hereinafter referred to as clear image C, T2T This term can refer to a clear image captured with the new image sensor. For example, the clear image C T2T This will be the first clear image captured as part of a measurement series. A linearity error of the new image sensor affects clear image C. T2Tin the same way as other image data captured with the new image sensor. In one embodiment, the Clear Image C is also used. T2T A non-linearity fit was performed before the clear image C T2T is used for clear normalization of linearity-adjusted image data.

[0013] The Clear image C T2T The image can be subjected to linearity correction to correct a linearity error of the first image sensor. This initial linearity correction can be the same initial linearity correction applied to the non-linearity-corrected measured image data.

[0014] The Clear image C T2T can be subjected to non-linearity adjustment to achieve the Clear Image C T2Tto imprint a linearity signature of the second image sensor. The second non-linearity adjustment can be the same non-linearity adjustment to which the linearity-corrected image data from the image acquisition is subjected.

[0015] In a prior process step, which may also be part of the method according to the invention, a measurement may have been carried out to determine the linearity error of the first image sensor. For this purpose, a relationship can be established between the amount of radiation incident on the first image sensor and the image data that the first image sensor generates from it. The amount of radiation directed onto the first image sensor can be varied in a larger number of measurements between zero and the saturation of the image sensor, and a relationship can be derived from this as to which amount of radiation corresponds to which output signal of the image sensor.

[0016] To vary the amount of radiation directed onto the image sensor during an image acquisition, either the radiation power or the exposure time can be adjusted. Keeping the radiation power constant and varying the exposure time has the advantage of eliminating linearity errors in an energy monitor. A beam splitter can be positioned in the beam path of the measuring device so that a portion of the incident radiation is directed to the energy monitor. Keeping the radiation power constant and varying the exposure time has the advantage that linearity errors in the energy monitor are irrelevant because the energy monitor readings remain constant.

[0017] If the first image sensor has a linearity error, a linear increase in the amount of radiation received during an exposure process leads to a non-linear increase in the output signal. A formula is developed for linearity correction, which converts the non-linear response of the output signal into a linear one. This formula can take the form of a polynomial. The polynomial can be of second to eighth order, and in particular, it can be of fifth order.

[0018] A calculation formula for the linearity error of the second image sensor can be determined in a similar manner. In many cases, it will be advantageous to determine the linearity correction for the second image sensor before the linearity correction for the first image sensor, because the second (old) image sensor is already installed, while the first (new) image sensor still needs to be installed.

[0019] In the method according to the invention, the calculation rule for the linearity error of the first image sensor is applied directly, so that the non-linear image data is converted into linear image data. The calculation rule for the linearity error of the second image sensor, on the other hand, is applied in inverse form, so that the image data, which are linear after the application of the first linearity correction, are given a linearity signature of the second image sensor.

[0020] The radiation directed onto the image sensor can be generated by a radiation source. This radiation source can be a laser. If the laser emits radiation pulses of constant energy, the amount of radiation directed onto the image sensor during an exposure can be determined by summing the energies of the radiation pulses. The energy monitor displays a similar reading for each radiation pulse. Despite identical radiation pulses, the readings will generally not be completely identical because the measurements are subject to noise. Any non-linearity of the energy monitor does not affect the measurement. Alternatively, a lamp capable of maintaining a constant emitted radiant power can be used as the radiation source. In this way, too, the amount of radiation reaching the image sensor can be adjusted by changing the exposure time.

[0021] The aerial image can be energy-normalized. For this to be the case, both the image data acquired with the first image sensor during image acquisition and the image data of the clear image used in clear normalization can be energy-normalized. Energy-normalized means that the raw data recorded with an image sensor is set in relation to the amount of radiation used for recording. The amount of radiation used for recording can be determined with an energy monitor as described above.

[0022] When recording aerial images, a procedure can be followed by the acquisition of a clear image and the subsequent acquisition of multiple images. Currently, it is common practice to perform the clear normalization of the measurement images based on the last previously recorded clear image. As an inventive solution that can be applied independently of the described non-linearity adjustment, it is proposed to use a first clear image and a second clear image for clear normalization, with the first clear image being acquired before the measurement images and the second clear image being acquired after the measurement images.

[0023] An interpolation over time can be performed between the first and second clear images, comparing the average intensity of the first and second clear images. This interpolation represents the temporal evolution of the image sensor's clear state between the time of the first and second clear images. A clear state derived from this interpolation can be used for clear normalization of a measurement. It is advantageous if the interpolated clear state corresponds to the same time at which the image was acquired.

[0024] The interpolation between the first clear image and the second clear image can be linear. Higher-order interpolation is also possible. According to current findings, the cost-benefit ratio appears to be favorable for linear interpolation.

[0025] A further discrepancy between the first and second image sensors can result from differing point-imaging characteristics. This means that a point-like object does not result in a perfectly point-like image on the image sensor, and that the spreading of the point-like object differs between the first and second image sensors. This discrepancy can be mathematically described by a modulation transfer function (MTF). In the method according to the invention, a first modulation transfer function (MTF1) can be determined for the first image sensor and a second modulation transfer function (MTF2) for the second image sensor. Determining the modulation transfer function can be achieved, for example, by imaging various structures onto an image sensor and deriving a correction for the image sensor from this.

[0026] The correction can be performed by applying the modulation transfer function MTF1 to the image data acquired with the first image sensor. Unlike linearity correction, the correction of the point imaging properties is not performed pixel by pixel, but rather as a correction in the frequency or Fourier space over the entire area of ​​the image sensor. In a further step of the inventive method, the inverse modulation transfer function MTF2 can be applied. -1 applied to the image data. The first modulation transfer function, MTF1, can be applied before or after linearity correction. The second modulation transfer function, MTF2, can be applied... -1 can be done before or after the non-linearity adjustment.

[0027] The inventive method can also be used to correct a linearity error of the energy monitor. For this purpose, it is advantageous to determine the linearity error of the energy monitor in a prior process step by sending signals to the energy monitor across its bandwidth and determining the ratio between the radiation incident on the energy monitor and the measured value obtained from it. Image data acquired with an image sensor can be used as a reference for comparison. To compensate for any linearity error of the image sensor, the exposure time of the image sensor can be adjusted so that the amount of radiation incident on the exposure sensor remains constant during an exposure process.This means that during a measurement process where a large amount of radiation is directed onto the energy monitor, the exposure time of the image sensor is kept short. The lower the amount of radiation on the energy monitor, the longer the exposure time of the image sensor can be. Linearity-corrected measurements from the energy monitor can be used for energy normalization of the aerial image.

[0028] There are measuring devices in which the image data acquired with the old image sensor has already been corrected. These corrections, which may arise, for example, from an earlier adjustment to a different image sensor, are referred to below as the default correction. The method according to the invention can be carried out such that, after processing the aerial image with the clear image C T2Tand the default correction is recalculated before applying the linearity correction. After the non-linearity adjustment is complete, the default correction can be applied again.

[0029] The invention also relates to a mask metrological measuring device comprising an image sensor, a processing module, and a correction module. The image sensor is designed to capture an image of a section of a photomask. The processing module is designed to generate an aerial image by subjecting the raw image data acquired with the image sensor to clear normalization. The correction module is designed to subject the aerial image to non-linearity adjustment. The non-linearity adjustment comprises the following steps: In step a, the aerial image is combined with a clear image. In step b, a linearity correction is applied to the image data generated in step a to correct a linearity error of the first image sensor. In step c, the image data generated in step b is further processed.In step d., a non-linearity adjustment is applied to the linearity-corrected image data obtained in order to imprint a linearity signature of a second image sensor not located in the beam path of the measuring device. In step d., a clear normalization is applied to the linearity-adjusted image data generated in step c.

[0030] The invention also relates to a computer program product or a set of computer program products comprising program parts which, when loaded into a computer or into interconnected computers connected to a measuring device according to the invention, are designed to carry out the method according to the invention.

[0031] The disclosure includes further developments of the mask metrological measuring device with features that are described in connection with the method according to the invention. The disclosure includes further developments of the method that are described in connection with the mask metrological measuring device according to the invention.

[0032] The invention is described below by way of example with reference to the accompanying drawings and advantageous embodiments. The drawings show: Fig. 1: a schematic representation of a measuring device according to the invention; Fig. 2: a top view of a surface with the measuring device Fig. 1 photomask examined; Fig. 3: a section of the beam path of the measuring device from Fig. 1 in a schematic representation; Fig. 4: a schematic representation of the illumination beam path of the measuring device Fig. 1; Fig. 5: the average intensity of several consecutively recorded clear images and a clear interpolation derived from them; Fig. 6: different positions on the image sensor where images of the measurement field can be created; Fig. 7: a block diagram of a process of the method according to the invention; Fig. 8: a representation of the linearity error of the image sensor; Fig. 9: the linearity error from Fig. 8 in another representation; Fig. 10-12: the view according to Fig. 7 in alternative embodiments of the invention.

[0033] A mask metrological measuring device according to the invention serves to examine the structure of a photomask 17. The photomask 17 is intended for use in a microlithographic projection exposure system (not shown). In the microlithographic projection exposure system, the photomask 17 is illuminated with deep ultraviolet radiation (DUV radiation) with a wavelength of, for example, 193 nm in order to image a structure formed on the photomask 17 onto the surface of a lithographic object in the form of a wafer. The wafer is coated with a photoresist that reacts to the DUV radiation. The measuring device is used to examine whether the structure on the photomask 17 corresponds to the dimensional specifications.

[0034] The measuring device is designed according to Fig. 1. The photomask 17 is arranged such that a beam path emanating from a laser radiation source 14 passes through the photomask 17 and is directed to an image sensor 20. The radiation has a wavelength of 193 nm, which corresponds to the DUV radiation used in the microlithographic projection exposure system. An illumination system 16 is arranged between the laser radiation source 14 and the photomask 17, with which the laser beam emitted by the laser radiation source 14 is spread out so that it uniformly illuminates a measurement field 22 within the area of ​​the photomask 17, see [reference]. Fig. 2. Using an imaging system 19, the structure of the photomask 17 is imaged onto an image sensor 20. The section of the beam path between the laser radiation source 14 and the photomask 17 is referred to as the illumination beam path 15.

[0035] The section of the beam path between the photomask 17 and the image sensor 20 is called the imaging beam path 21.

[0036] The photomask 17 is arranged in the measuring device on an XY positioner 18, which is in Fig. Figure 1 is shown schematically. By moving the photomask 17 in the XY plane, the illumination beam path 15 can be directed onto different areas of the photomask 17. A field stop 28 limits the measuring field 22, onto which the illumination beam path 15 strikes via a condenser optic 31.

[0037] In Fig. Figure 4 schematically depicts the illumination beam path 15 between the laser radiation source 14 and the condenser optics 31. A laser beam emitted from the laser radiation source 14 is first guided through a beam attenuator 32. The beam attenuator 32 is adjusted so that the intensity of the laser beam is matched to the sensitivity of the image sensor 20. A prism arrangement 33 deflects the illumination beam path 15 towards a first optical assembly 34, which images the exit aperture of the laser radiation source 14 onto a pupil-forming mirror element 35. The pupil-forming mirror element 35 can be a mirror array comprising a plurality of mirror elements that are movably suspended from a frame structure and whose orientation relative to the frame structure can be individually adjusted.

[0038] The pupil-shaping mirror element or mirror array 35 is arranged in a pupil plane 38 of the illumination system 16, so that reflected radiation is distributed with uniform intensity over the measurement field 22 on the photomask 17. By adjusting the pupil-shaping mirror element or elements, the illumination setting can be varied, i.e., the angular distribution at which the radiation strikes the measurement field 22.

[0039] A second optical assembly 36 directs the illumination radiation to a third optical assembly 37. The third optical assembly 37 comprises a diffractive optical element in the form of a field DOE 43, which generates the field, and a field aperture 28, which defines the measurement field 22. A beam splitter 41 divides the illumination beam path 15, so that a first part of the radiation is directed via the condenser optics 31 to the photomask 17, and a second part of the radiation is directed to an energy monitor 40.

[0040] The distribution of the illumination radiation between the energy monitor 40 and the condenser optics 31 is fixed, so that the measured values ​​of the energy monitor 40 represent a measure of the amount of radiation that reaches the photomask 17 via the condenser optics 31. The image data acquired with the image sensor 20 can be subjected to energy normalization based on the measured values ​​of the energy monitor 40.

[0041] To examine a photomask 17, a large number of images are captured in a temporal sequence using the image sensor 20. The photomask 17 is moved between shots using the XY positioner 18, allowing different areas of the photomask 17 to be examined.

[0042] At regular intervals, for example, 15 minutes, a clear image is acquired in which the beam path passes through a clear position of the photomask 17, which is free of any structure. A clear image serves as a reference for the measurement by recording an image that is not influenced by ordered structures in the beam path. The clear images can be used to perform clear normalization of recorded image data, thus reducing the influence of drift on the image data. An image produced by clear normalization of raw image data is called an aerial image. The measuring device includes a computing module 23 in which raw image data recorded by the image sensor 20 undergoes clear normalization to generate an aerial image.

[0043] Up to now, it has been standard practice to use the last clear image recorded before the image acquisition to be normalized for clear normalization. Clear normalization involves calculating a ratio between the raw data of the image acquisition and the clear image. Clear normalization can be performed in normalized units on a scale ranging from 0 to 1. Here, 0 corresponds to the value provided by the image sensor 20 when the laser light source 14 is inactive, i.e., no radiation from the illumination system 16 is directed towards the image sensor 20. The value 1 corresponds to the value provided by the image sensor 20 in a saturated state. The average intensity of a clear image might, for example, be 0.7. The average intensity of an image acquisition is lower than the intensity of the clear image.

[0044] A drift of the image sensor 20 can cause the average clear intensity to change significantly between two clear images. Therefore, changing the clear image used for clear normalization can result in an abrupt change between two consecutive aerial images.

[0045] According to the invention, it is proposed to perform a clear interpolation between two successive clear images. For this purpose, the average intensity of a first clear image recorded at time T1 and a second clear image recorded at time T2 are determined, and a linear interpolation is performed that extends between times T1 and T2. Fig. 5 is an example of the average intensity I CThe image is represented by clear images acquired at four time points T1, T2, T3, and T4. The intervals between the clear images define the time periods for the linear clear interpolation, which is performed in Fig. 5 is also shown.

[0046] A process occurring on the image sensor 20 Fig. The measuring field 22 does not fill the entire area of ​​the image sensor 20, see Fig. 6. The position of Fig. The position of the measurement field 22 on the photomask 17 also shifts depending on the position of the measurement field 22 on the image sensor 20. Various possible positions of the Fig. The image sensor contains 20 cells. Fig. Figure 6 shows the mean clear intensity I. C , which forms the basis for the clear interpolation, is determined using a section 26 that is tailored to the target image size and arranged centrally on the image sensor 20.

[0047] From the clear interpolation, a clear correction factor F can be determined for each time point T in the relevant time interval. CI The image correction is performed by applying the clear correction factor F, which applies to time T, to the aerial image acquired at time T, pixel by pixel. CI The data is shared. The result is a corrected AI aerial image. CI . AICI=AIFCI

[0048] After prolonged use of a measuring device according to the invention, it may become necessary to replace the image sensor installed in the measuring device with a new one. This should be done in such a way that the user of the measuring device experiences no change, i.e., that the measurement result with the new image sensor looks exactly the same as the measurement result with the old image sensor. This consistency does not occur automatically because, in general, the linearity errors of two image sensors are not the same.

[0049] To determine the linearity error of an image sensor 20, a sequence of exposure processes can be performed, covering the bandwidth of the image sensor 20 from zero incident radiation to saturation of the image sensor 20. A measure of the actual amount of radiation incident on the image sensor 20 is obtained from the measured values ​​E monof the energy monitor 40. In order to avoid that the measurement of the image sensor 20 is distorted by a linearity error of the energy monitor 40, the exposure processes can be carried out in such a way that the energy monitor 40 measures a constant value, so that any linearity error of the energy monitor 40 does not have an effect.

[0050] For this purpose, the laser beam source 14 of the measuring device can be set to emit laser pulses with constant energy. The energy monitor 40 always detects the same amount of radiation when such a laser pulse strikes it. The amount of radiation striking the image sensor 20 can be varied by changing the number of laser pulses during an exposure process. The energy monitor 40 then provides a measured value E. mon , which corresponds to the sum of all radiation pulses.

[0051] The measurement series can begin with zero laser pulses to determine the measured value E.cam The image sensor 20 delivers without incident radiation. The number of laser pulses can be gradually increased until the image sensor 20 reaches saturation. An ideal image sensor 20 would provide measured values ​​E cam deliver those linearly with E mon , i.e., increase with the number of laser pulses, see Fig. 8. A real image sensor 20 deviates from the ideal curve. In Fig. Figure 8 indicates this by a curve that lies below the ideal linear progression in the lower range and above it in the upper range. Fig. 9 is the same linearity error again as a relative quantity with respect to the incident radiation quantity E. mon depicted.

[0052] A correction function can be calculated, for example in the form of a fifth-order polynomial, which can be used to correct the non-linear behavior of the measured values ​​E. camis transformed into a linear progression. When replacing the image sensor, such correction functions can be determined for both the old and the new image sensor 20. This results in a first polynomial P lin1 for the new image sensor 20 and a second polynomial P lin2 for the old image sensor. Using these correction functions, an image taken with the new image sensor 20 can be converted as if it had been taken with the old image sensor.

[0053] In Fig. Figure 7 shows the image sensor 20, which stores the raw data in a buffer 44 of the processing module 23. In step 45, the raw image data undergoes clear normalization to generate an aerial image, which is stored in a memory 46. The aerial image is subject to the linearity error of the image sensor 20 and therefore differs from an aerial image that would have been produced if a different image sensor had been used instead of the image sensor 20.

[0054] For image correction, a correction module 57 is used, which reads the image data of the aerial photograph from memory 46. In a first step 47 of the image correction, the aerial photograph is treated with a clear image C. T2T billed. The Clear image C T2Tis a clear image acquired at the beginning of the measurement series, whose mean intensity is calibrated to a constant value stored in a database. This value corresponds to the mean of a clear image. The linearity error of the first image sensor 20, with which the raw data were acquired, is calculated in step 48 by calculating the image data obtained from the aerial photograph using the first polynomial P. lin1 will be subjected to a linearity correction.

[0055] In a subsequent step 49, a non-linearity adjustment in the form of a second inverse linearity correction can be applied to imprint the linearity signature of the second image sensor onto the linearity-corrected image data. For this purpose, the inverse second polynomial P is used. -1 lin2applied to the linearity-corrected image data to generate linearity-adjusted image data from which the linearity error of the first image sensor 20 was removed and to which the linearity signature of the second image sensor was instead imprinted.

[0056] To generate a customized aerial image from the adjusted image data, a final clear normalization is performed in step 50. To ensure that the clear normalization accurately reflects the transition from the old image sensor to the new image sensor 20, the clear image C is used. T2T It undergoes a corresponding non-linearity adjustment. In a first step, the polynomial P is applied. lin1 the linearity error of the first image sensor 20 from the clear image C T2T The correction is then applied. In a second step, the linearity-corrected clear image C is used. T2T the inverse polynomial P¯ 1 lin2applied to create the Clear Image C T2T to adapt to the linearity signature of the old camera. To generate the linearity-adjusted aerial image AI T2T The image data from the linearity-adjusted image acquisition are replaced by the linearity-adjusted clear image C. T2T divided. Formulated as an equation, this yields the linearity-adjusted aerial image AI. T2T as follows from the aerial image AI1 taken with the first image sensor 20: AIT2T=Plin2−1(Plin1(AI1∗CT2T))Plin2−1(Plin1(CT2T))

[0057] The linearity-adjusted aerial image AI T2T The result of the measurement can be output by the correction module 57 and made available for further use. For the user, the linearity-adjusted aerial image AI appears T2T as if it had been taken with the old image sensor, so that the user can continue to use their usual routines without any changes.

[0058] In the alternative embodiment according to Fig. In an additional step 51, energy normalization is performed in the computing module 23. For this purpose, the raw data acquired with the image sensor 20 are compared to the measured value of the energy monitor 40 at the time of image acquisition. Similarly, the clear image used for clear normalization can also be subjected to energy normalization. For this, the clear image is compared to the measured value of the energy monitor 40 at the time the clear image was acquired. This results in an energy-normalized and a clear-normalized aerial image, which is stored in memory 46. With image acquisition I, the clear image C, and energy normalization E I image acquisition and energy normalization E C The clear image can be represented as an equation as follows. AI=I∗ECC∗EI

[0059] During energy normalization, a linearity correction of the measured values ​​from the energy monitor 40 can also be taken into account. Similar to the image sensor, a correction function in the form of a polynomial can be determined for this purpose, with which the non-linear measured values ​​of the energy monitor 40 are converted into linear data. The necessary measurement data on the properties of the energy monitor 40 can be obtained by varying the energy measured by the energy monitor 40 while the image sensor 20 detects an unchanged amount of radiation. This can be achieved by reducing the exposure time of the image sensor 20 as the energy of the laser pulses increases. At the level of the correction module 57, a linearity correction of the energy normalization is not necessary because the relevant errors average out anyway when the image acquisition and the clear acquisition are compared.

[0060] In another variant, it is additionally taken into account that the new image sensor 20 and the old image sensor may have different point imaging properties. To correct for this, a first modulation transfer function MTF1 can be determined for the new image sensor 20 and a second modulation transfer function MTF2 for the old image sensor.

[0061] According to Fig. This results in two additional steps in the correction module 57. Following the correction of the linearity error of the new image sensor 20 in step 48 and Fourier transformation into the frequency domain, the linearity-corrected image data are divided by the modulation transfer function MTF1 in step 52 to eliminate the point imaging error of the new image sensor 20. Immediately afterwards, in step 53, a multiplication with the modulation transfer function MTF2 can be performed to imprint the point imaging error of the old image sensor into the image data. Then, the inverse Fourier transformation back into the spatial domain is carried out. Alternatively, it would also be possible to correct the point imaging error of the new image sensor 20 immediately after image acquisition, taking the Fourier transformation into account, by directly dividing the raw image acquisition data by the modulation transfer function MTF1.

[0062] In another embodiment, which is described in Fig. As shown in Figure 12, a default correction is performed in a further step 54 in the calculation module 23. This default correction may have been introduced earlier to compensate for an inaccuracy that occurred when using the old image sensor. The default correction is mathematically described by a polynomial P. def .

[0063] In the non-linearity adjustment according to the invention, the result can be distorted by such a default correction. Therefore, the method can be carried out such that, before the linearity correction, the default correction is applied by inversely using the polynomial P. def The correction is reversed in step 55, and the default correction is reapplied following the non-linearity adjustment in step 56. This can apply to both the image acquisition itself and the clear image C. T2TThese conditions apply. In mathematical terms, this results in the linearity-adjusted aerial image AI. T2T then as follows. AIT2T=Pdef(Plin2−1{Plin1[Pdef−1(AI1∗CT2T)]})Pdef(Plin2−1{Plin1[Pdef−1(CT2T)]})

[0064] In this way, the inventive method can also be applied if the image data of the old image sensor has already been subjected to a default correction.

Claims

[1] Method for operating a mask metrological measuring device in which an image of a section of a photomask (17) is acquired with a first image sensor (20) and in which an aerial image is produced by subjecting the raw image data acquired with the first image sensor (20) to a clear normalization (45) and in which the aerial image is subjected to a non-linearity adjustment comprising the following steps: a. Compiling (47) the aerial image with a clear image (CT2T); b. Applying (48) a linearity correction (Plin1) to the image data generated in step a. to correct a linearity error of the first image sensor (20); c. Applying (49) a non-linearity adjustment (P-1lin2) to the linearity-corrected image data obtained in step b. to imprint on the image data a linearity signature of a second image sensor not located in the beam path of the measuring device; d. Apply (50) a clear normalization to the linearity-adjusted image data generated in step c. [2] Method according to claim 1, wherein the aerial image is multiplied pixel-wise with the clear image in step a. [3] Method according to claim 1 or 2, wherein the clear image (CT2T) is a clear image taken with the first image sensor (20). [4] Method according to any one of claims 1 to 3, wherein the clear image (CT2T) is subjected to non-linearity adjustment and wherein the linearity-adjusted clear image (CT2T) is used for clear normalization in step d. [5] Method according to claim 4, wherein the non-linearity adjustment of the clear image (CT2T) comprises a linearity correction to correct a linearity error of the first image sensor (20). [6] Method according to claim 4 or 5, wherein the non-linearity adjustment of the clear image (CT2T) comprises a non-linearity adjustment to imprint a linearity signature of the second image sensor (20) on the clear image (CT2T). [7] Method according to any one of claims 1 to 6, wherein in a prior method step, a measurement was carried out to determine the linearity error of the first image sensor (20). [8] Method according to claim 7, wherein the measurement is carried out while an energy monitor (40) of the measuring device provides constant measured values. [9] Method according to any one of claims 1 to 8, wherein the aerial image is an energy-normalized aerial image. [10] Method according to any one of claims 1 to 9, wherein before step a. a default correction (Pdef) applicable to the second image sensor is recalculated. [11] Method according to any one of claims 1 to 10, wherein the clear normalization of the aerial image is based on a first clear image (T1) and a second clear image (T2), wherein the first clear image (T1) is taken before the image acquisition and the second clear image (T2) is taken after the image acquisition. [12] Method according to claim 11, wherein a linear interpolation over time is performed between the average intensity of the first clear image (T1) and the average intensity of the second clear image (T2). [13] Method according to one of claims 1 to 12, wherein, within the framework of non-linearity adjustment, a point imaging error of the first image sensor (20) is corrected and a point imaging error of the second image sensor is imprinted on the image data. [14] Mask metrological measuring device, comprising a first image sensor (20) designed to capture an image of a section of a photomask (17), a computing module (23) designed to generate an aerial image by subjecting the raw image data obtained with the first image sensor (20) to clear normalization (45), and a correction module (57) designed to subject the aerial image to non-linearity adjustment, comprising the following steps: a. Compiling (47) the aerial image with a clear image (CT2T); b. Applying (48) a linearity correction (Plin1) to the image data generated in step a. to correct a linearity error of the first image sensor (20); c. Applying (49) a non-linearity adjustment (P-1lin2) to the linearity-corrected image data obtained in step b. to imprint on the image data a linearity signature of a second image sensor not located in the beam path of the measuring device; d. Apply (50) a clear normalization to the linearity-adjusted image data generated in step c. [15] Computer program product or set of computer program products, comprising program parts which, when loaded into a computer or into interconnected computers connected to a measuring device according to the invention, are designed to carry out a method according to any one of claims 1 to 13.

Images