Method of adjusting a model for a lithographic process and associated apparatus

Abstract

A method of adjusting a lithographic process for a particular patterning device. The method comprises obtaining wavefront data (420) relating to an objective lens (PS) of a lithographic apparatus (LA), the wavefront data being measured after an exposure of a pattern on a substrate (W) using the particular patterning device (MA); determining a pattern-specific wavefront contribution from the wavefront data (445) and a reference wavefront (430), the pattern-specific wavefront contribution relating to the patterning device; and using the pattern-specific wavefront contribution to adjust (460) the lithographic process for the particular patterning device.

Description

[0001] Cross-reference to related applications

[0002] This application claims priority to European Application 20156986.0, filed on February 12, 2020, and European Application 20178091.3, filed on June 3, 2020, the entire contents of which are incorporated herein by reference. Technical Field

[0003] The present invention relates to methods and apparatus for manufacturing devices, for example, by photolithography, and to methods for manufacturing devices using photolithography. Background Technology

[0004] A photolithography apparatus is a machine that applies a desired pattern onto a substrate (typically onto a target portion of the substrate). Photolithography apparatus can be used, for example, in the manufacture of integrated circuits (ICs). In that case, a patterning apparatus (which is alternatively referred to as a mask or photomask) can be used to generate circuit patterns to be formed on individual layers of the IC. This pattern can then be transferred to target portions (e.g., portions including dies, a die, or several dies) on a substrate (e.g., a silicon wafer). Pattern transfer is typically performed by imaging onto a layer of radiation-sensitive material (resist) disposed on the substrate. Typically, a single substrate will contain a network of adjacent target portions that are continuously patterned. These target portions are typically referred to as “fields.”

[0005] In the fabrication of complex devices, numerous photolithographic patterning steps are typically performed, thereby forming functional features in successive layers on a substrate. Therefore, a key aspect of the performance of a photolithography apparatus is its ability (using the same or different apparatuses) to properly and accurately align the applied pattern relative to features placed in previous layers. For this purpose, the substrate is provided with one or more sets of alignment marks. Each mark is a structure whose position can later be measured using a position sensor (typically an optical position sensor). The photolithography apparatus includes one or more alignment sensors through which the position of the marks on the substrate can be accurately measured. Different types of marks and different types of alignment sensors are known depending on the manufacturer and even different products from the same manufacturer.

[0006] Photolithography equipment includes a projection system. The projection system includes optical elements, such as, for example, lenses. Optical aberrations can arise from defects in the optical elements. Optical aberrations can also arise from projection effects, such as, for example, heating of the optical elements that occurs during photolithography exposure. A projection system model is used to determine one or more adjustments that can be made to the optical elements of the projection system. The determined adjustments can have the effect of reducing optical aberrations within the projection system.

[0007] It may be desirable to provide improvements, for example, in the calibration, adjustment, and / or use of projection system modeling methods, which avoid or mitigate one or more of the problems of the prior art, whether identified herein or elsewhere. Summary of the Invention

[0008] In a first aspect, the present invention provides a method for adjusting a photolithography process for a particular patterning apparatus, the method comprising: obtaining wavefront data associated with an objective lens of a photolithography apparatus, the wavefront data being measured after exposure of a pattern on a substrate using the particular patterning apparatus; determining a pattern-specific wavefront contribution based on the wavefront data and a reference wavefront, the pattern-specific wavefront contribution being associated with the particular patterning apparatus; and using the pattern-specific wavefront contribution to adjust the photolithography process for the patterning apparatus.

[0009] A lithography apparatus is also disclosed, which includes operable methods for performing the first aspect.

[0010] The above and other aspects of the invention will be understood in light of the examples described below. Attached Figure Description

[0011] Embodiments of the invention will now be described with reference to the accompanying drawings, which are by way of example only, in which:

[0012] Figure 1 Describing the photolithography equipment;

[0013] Figure 2 schematic diagram Figure 1 Measurement and exposure processes in the equipment;

[0014] Figure 3 The illustrations show (a) a known empirical method for determining lens heating correction and (b) a known computational method for determining lens heating correction.

[0015] Figure 4 yes Figure 3 (b) Detailed flowchart of the method;

[0016] Figure 5 This is a flowchart of a method according to a first embodiment of the present invention; and

[0017] Figure 6 This is a flowchart of a method according to a second embodiment of the present invention. Detailed Implementation

[0018] Before describing the embodiments of the invention in detail, it is instructive to present example environments that can be used to implement the embodiments of the invention.

[0019] Figure 1 A lithography apparatus LA is schematically depicted. The apparatus includes: an irradiation system (irradiator) IL configured to modulate a radiation beam B (e.g., UV or DUV radiation); a patterning apparatus support or support structure (e.g., a mask stage) MT configured to support a patterning apparatus (e.g., a mask) MA and connected to a first positioner PM configured to accurately position the patterning apparatus according to certain parameters; two substrate stages (e.g., wafer stages) WTa and WTb configured to hold a substrate (e.g., a resist-coated wafer) W and connected to a second positioner PW configured to accurately position the substrate according to certain parameters; and a projection system (e.g., a refractive projection lens system) PS configured to project a pattern imparted by the patterning apparatus MA to the radiation beam B onto a target portion C (e.g., comprising one or more dies) of the substrate W. The reference frame RF connects various components and serves as a reference for setting and measuring the positions of the pattern forming apparatus and the substrate, as well as for setting and measuring the positions of features on the pattern forming apparatus and the substrate.

[0020] Irradiation systems may include various types of optical components for guiding, shaping, or controlling radiation, such as refractive, reflective, magnetic, electromagnetic, electrostatic, or other types of optical components, or any combination thereof.

[0021] The patterning apparatus support MT holds the patterning apparatus in a manner dependent on the orientation of the patterning apparatus, the design of the lithography equipment, and other conditions (such as, for example, whether the patterning apparatus is held in a vacuum environment). The patterning apparatus support MT can use mechanical, vacuum, electrostatic, or other clamping techniques to hold the patterning apparatus. The patterning apparatus support MT can be, for example, a frame or stage, which may be fixed or movable as needed. The patterning apparatus support can ensure that the patterning apparatus is, for example, in a desired position relative to a projection system.

[0022] The term "patterning apparatus" as used herein should be interpreted broadly as any apparatus that can be used to impart a pattern to a radiation beam in the cross-section of the beam in order to produce a pattern in a target portion of a substrate. It should be noted that, for example, if the pattern imparted to the radiation beam includes phase-shifting features or so-called auxiliary features, the pattern may not precisely correspond to the desired pattern in the target portion of the substrate. Typically, the pattern imparted to the radiation beam will correspond to a specific functional layer in a device (such as an integrated circuit) produced in the target portion.

[0023] As depicted herein, the apparatus is of the transmissive type (e.g., using a transmissive pattern forming apparatus). Alternatively, the apparatus may be of the reflective type (e.g., using a programmable mirror array of the type mentioned above, or using a reflective mask). Examples of pattern forming apparatuses include masks, programmable mirror arrays, and programmable LCD panels. Any use of the terms “mask” or “mask” herein is to be considered synonymous with the more general term “pattern forming apparatus.” The term “pattern forming apparatus” can also be interpreted as an apparatus that stores pattern information in digital form for use in controlling such a programmable pattern forming apparatus.

[0024] The term “projection system” as used herein should be interpreted broadly to encompass any type of projection system suitable for the exposure radiation used or for other factors such as immersion in liquids or vacuum, including refractive, reflective, reflective-refractive, magnetic, electromagnetic, and electrostatic optical systems, or any combination thereof. Any use of the term “projection lens” herein may be considered synonymous with the more general term “projection system.”

[0025] Photolithography apparatuses can also fall into the category where at least a portion of the substrate can be covered by a liquid (e.g., water) with a relatively high refractive index to fill the space between the projection system and the substrate. Immersion liquids can also be applied to other spaces within the photolithography apparatus, such as the space between the mask and the projection system. Immersion techniques are well-known in the art for increasing the numerical aperture of projection systems.

[0026] In operation, the irradiator IL receives a radiation beam from the radiation source SO. For example, when the source is an excimer laser, the source and the lithography apparatus can be separate entities. In these cases, the source is not considered part of the lithography apparatus, and the radiation beam is delivered from the source SO to the irradiator IL by means of a beam delivery system BD, which includes, for example, suitable directional mirrors and / or beam expanders. In other cases, for example, when the source is a mercury lamp, the source can be an integral part of the lithography apparatus. The source SO and the irradiator IL, together with the beam delivery system BD, can be referred to as the radiation system when necessary.

[0027] The irradiator IL may, for example, include an adjuster AD, an integrator IN, and a concentrator CO for adjusting the angular intensity distribution of the radiation beam. The irradiator can be used to adjust the radiation beam to have a desired uniformity and intensity distribution in its cross-section.

[0028] A radiation beam B is incident on a pattern forming apparatus MA held on a pattern forming apparatus support MT and is patterned by the pattern forming apparatus. Having traversed the pattern forming apparatus (e.g., a mask) MA, the radiation beam B passes through a projection system PS, which focuses the beam onto a target portion C of the substrate W. The substrate stage WTa or WTb can be accurately moved, for example, to position different target portions C within the path of the radiation beam B, by means of a second positioner PW and a position sensor IF (e.g., an interferometer, a linear encoder, a 2-D encoder, or a capacitive sensor). Similarly, a first positioner PM and another position sensor (this other position sensor is not in...) Figure 1 (as explicitly described in the text) can be used, for example, to accurately position a pattern forming apparatus (e.g., a mask) MA relative to the path of the radiation beam B, either after mechanical acquisition from a mask library or during scanning.

[0029] Mask alignment marks M1, M2 and substrate alignment marks P1, P2 can be used to align a patterning apparatus (e.g., a mask) MA and a substrate W. Although the substrate alignment marks, as illustrated, occupy dedicated target portions, they can also be located in the space between target portions (these marks are referred to as scribing alignment marks). Similarly, in cases where more than one die is disposed on the patterning apparatus (e.g., a mask) MA, mask alignment marks can be located between the dies. Smaller alignment marks can also be included within the die within device features, in which case it is desirable to keep the markings as small as possible and without requiring any imaging or process conditions different from adjacent features. An alignment system for detecting alignment marks is further described below.

[0030] The depicted apparatus can be used in a variety of modes. In scanning mode, the pattern forming apparatus support (e.g., mask stage) MT and substrate stage WT are scanned synchronously while the pattern to be applied to the radiation beam is projected onto the target portion C (i.e., single dynamic exposure). The speed and direction of the substrate stage WT relative to the pattern forming apparatus support (e.g., mask stage) MT can be determined by the magnification (reduction) and image inversion characteristics of the projection system PS. In scanning mode, the maximum size of the exposure field limits the width of the target portion (in the non-scanning direction) in a single dynamic exposure, while the length of the scanning motion determines the height of the target portion (in the scanning direction). Other types of lithography apparatus and operating modes are possible, as is well known in the art. For example, stepping mode is known. In so-called “maskless” lithography, the programmable pattern forming apparatus is kept stationary but has a changing pattern, and the substrate stage WT is moved or scanned.

[0031] You can also use combinations and / or variations or completely different usage patterns described above.

[0032] The lithography apparatus LA belongs to the so-called dual-platform type, which has two substrate stages WTa and WTb and two stations—an exposure station EXP and a measurement station MEA—between which the substrate stages can be exchanged. While a substrate on one stage is being exposed at the exposure station, another substrate can be loaded onto the other stage at the measurement station and various preparatory steps can be performed. This configuration enables a significant increase in the apparatus's throughput. The preparatory steps may include using a level sensor LS to map the surface height profile of the substrate and using an alignment sensor AS to measure the position of alignment marks on the substrate. If the position sensor IF is unable to measure the position of the substrate stage when it is at both the measurement and exposure stations, a second position sensor can be provided to enable tracking of the substrate stage's position relative to the reference frame RF at both stations. Other arrangements are known and available instead of the illustrated dual-platform arrangement. For example, other lithography apparatuses providing substrate stages and measurement stages are well known. These substrate stages and measurement stages are mated together during preparatory measurements and then separated during exposure of the substrate stage.

[0033] The lithography apparatus LA also includes a wavefront sensor WS; for example, a wavefront sensor WS located on each substrate stage WTa, WTb. This sensor can measure the quality of the wavefront image, as well as any aberrations, and thus enable (e.g., parallel) measurements of optical aberrations through the projection slit, thereby achieving (e.g.) more accurate alignment, improved mask heating correction, and in-process lens heating correction. The wavefront sensor can include an interferometric wavefront measurement system and can perform static measurements on up to high-order lens aberrations. The interferometric wavefront measurement system can be implemented as an integrated measurement system for system initialization and calibration. Alternatively, the interferometric wavefront measurement system can be used for "on-demand" monitoring and recalibration. Thus, the wavefront sensor can be able to measure lens aberrations on a batch-by-batch or per-wafer basis, depending on its design.

[0034] Figure 2 The diagram illustrates the exposure of a target portion (e.g., a die) to light. Figure 1 The steps on substrate W in the dual-platform apparatus are shown. The left side within the dashed box represents the steps performed at the measurement station MEA, while the right side shows the steps performed at the exposure station EXP. Sometimes, one of the substrate stages WTa and WTb will be located at the exposure station, while the other substrate stage will be located at the measurement station, as described above. For the purposes of this description, it is assumed that substrate W has already been loaded into the exposure station. At step 200, a new substrate W' is loaded into the apparatus via a mechanism not shown in the figure. Both substrates are processed in parallel to increase the throughput of the lithography apparatus.

[0035] First, consider a newly loaded substrate W', which may be a previously untreated substrate prepared using a new photoresist for the first exposure in the apparatus. However, typically, the described photolithography process will only be one step in a series of exposure and processing steps, such that the substrate W' has been passed through this apparatus and / or other photolithography apparatuses several times, and may undergo subsequent processes. Specifically addressing the problem of improving overlap performance, the task is to ensure that the new pattern is accurately applied to the correct position on the substrate that has already undergone one or more cycles of patterning and processing. These processing steps gradually introduce deformation into the substrate, which must be measured and corrected to achieve satisfactory overlap performance.

[0036] Pre- and / or subsequent patterning steps (as mentioned above) can be performed in other lithography equipment, and even in different types of lithography equipment. For example, some layers in the device fabrication process that have very high requirements for parameters such as resolution and overlap can be performed in more advanced lithography tools compared to other layers with less stringent requirements. Thus, some layers can be exposed in immersion-type lithography tools, while others are exposed in "dry" tools. Some layers can be exposed in tools operating at DUV wavelengths, while others are exposed using EUV wavelength radiation.

[0037] At position 202, alignment measurements using substrate marker P1 and an image sensor (not shown) are used to measure and record the alignment of the substrate relative to the substrate stage WTa / WTb. Additionally, an alignment sensor AS is used to measure several alignment marks across the entire substrate W'. In one embodiment, these measurements are used to establish a “wafer grid” that maps very accurately to the distribution of marks across the entire substrate, including any distortions relative to the nominal rectangular grid.

[0038] At step 204, a horizontal sensor LS is also used to measure a graph or map of the wafer height (Z) relative to the XY position. Typically, height maps or height graphs are used only to achieve accurate focusing of the pattern after exposure. Height maps can also be used for other purposes.

[0039] When substrate W' is loaded, configuration data 206 is received, which defines the exposure to be performed and also defines the properties of the wafer and the previously generated pattern and the pattern to be generated on the wafer. Measurements of wafer position, wafer grid, and height mapping obtained at 202, 204 are added to this configuration data, allowing a complete set of configuration and measurement data 208 to be passed to the exposure station EXP. Alignment data measurements include, for example, the X and Y positions of alignment targets formed in a fixed or nominal fixed relationship with the product pattern (the product of the lithography process). This alignment data, obtained just before exposure, is used to generate an alignment model with parameters that fit the model to the data. These parameters and the alignment model are used during the exposure operation to correct the position of the pattern applied in the current lithography step. The model used interpolates positional deviations between the measured positions. A typical alignment model may include four, five, or six parameters that together define the translation, rotation, and scaling of the "ideal" grid at different sizes. Advanced models using more parameters are well known.

[0040] At position 210, wafers W' and W are swapped so that the substrate W' being measured becomes the substrate W entering the exposure station EXP. Figure 1 In the example apparatus, this interchange is performed by swapping supports WTa and WTb within the apparatus, ensuring that substrates W and W' remain accurately clamped and positioned on those supports to maintain relative alignment between the substrate stage and the substrate itself. Therefore, once the stage has been interchanged, in order to control the exposure steps using measurement information 202, 204 for substrate W (formerly W'), it is necessary to determine the relative position between the projection system PS and the substrate stage WTb (formerly WTa). At step 212, mask alignment is performed using mask alignment marks M1, M2. In steps 214, 216, 218, scanning motion and radiation pulses are applied to successive target locations across the entire substrate W to complete the exposure of multiple patterns.

[0041] By using alignment data and height mapping obtained at the measurement station during the exposure step, these patterns are precisely aligned relative to the desired location, and specifically, relative to features previously placed on the same substrate. The exposed substrate, now labeled "W", is unloaded from the equipment at step 220 to undergo etching or other processes according to the exposed pattern.

[0042] Those skilled in the art will recognize that the above description is a simplified overview of several very detailed steps involved in an example of a real manufacturing scenario. For example, there are often separate stages of coarse and fine measurements using the same or different markings, rather than measuring alignment in a single pass. The coarse and / or fine alignment measurement steps can be performed before or after the height measurement, or alternately.

[0043] The concepts disclosed herein relate to the use of wavefront sensors, such as the wavefront sensor WS described above. Such wavefront sensors can be used to measure wavefront errors or aberrations, which are typically described as coefficients in a series of 2D mathematical functions known as Zernike polynomials. These wavefront errors or aberrations can be regrouped to describe some well-known optical aberrations, such as spherical, coma, and astigmatism.

[0044] Typically, wavefront sensors cannot be used to directly measure wavefronts based on product structure. However, for lenses that have been heated by exposure, the wavefront signal will carry a “memory” or “artifact” of the exposed product features (pattern or mask, irradiation, and dose-specific wavefront contributions). This is because when the product features are exposed, the lens (i.e., the optical column) heats in the pupil plane based on diffraction induced by the irradiation pattern combined with the product features on the mask. Immediately after exposure, wavefront sensors can be used to measure wavefront data; such sensors will observe changes in the wavefront relative to the cooled lens, changes caused by and describing the shape of the product feature and the irradiation setup. This shape will gradually diminish as heat diffuses in the lens material. It is proposed to utilize this phenomenon and use this wavefront data for several different applications (e.g., differential wavefronts including the difference between a heated lens wavefront measurement and a reference wavefront (e.g., a cooled lens wavefront measurement). In embodiments, the concepts described herein propose using raw wavefront sensor signals, which are pixelated images captured by a wavefront sensor camera, for example, without Zernike decomposition, to determine pattern-specific wavefront contributions. This raw wavefront data can be a mapped representation of the wavefront error (scalar) across the entire exit pupil plane of the objective lens, rather than vectorizing the wavefront to Zernike coefficients. Alternatively, pattern-specific wavefront contributions can be determined in a conventional manner from wavefront data processed into Zernike polynomials.

[0045] Among the factors influencing overlap and focal length drift, lens heating is a major contributor to image quality degradation, particularly for frequently used, very localized illumination sources (such as dipole illumination). These localized thermal distributions on the lens can lead to undesirable effects on the imaging wavefront. Aberration drift during batch exposures is unavoidable due to lens heating during exposure. The cumulative absorption of light within the lens can ultimately result in imaging defects, which are a direct consequence of wavefront errors induced by lens heating. Therefore, in this first embodiment, such raw wavefront data is proposed to improve lens heating control, and particularly to improve the calibration and / or verification of lens heating models used for predicting lens heating. In other variations of this first embodiment, the raw wavefront data is not used, and another mask description is used instead.

[0046] Therefore, lens heating (LH) control is crucial for maintaining image quality. A key step in LH control is the ability to accurately predict lens heating. On some systems, a current method known as Application-Specific Calibration (ASCAL) is used to predict lens heating effects by measuring wavefront drift (e.g., using a wavefront sensor WS as described above). The aforementioned Zernike coefficients can be measured by ASCAL at different slit locations and with different wafer numbers during batch exposure to help understand and predict LH evolution behavior. The predicted LH behavior can then be corrected using one or more of the many LH correction options available on modern scanners, such as, for example, image adjuster lens adjustment control. For instance, ASCAL generates a lens heating feedforward (LHFF) dataset at the end of its calibration process for device layers on a scanner. This LHFF dataset is applied to the target scanner and layer to correct for LH-induced wavefront drift.

[0047] Figure 3 (a) The main steps of the online ASCAL workflow are illustrated. An exposure scheme 300, which may include, for example, mask data, dose data, and field size data, is used to determine the scanner settings. These scanner settings are used in the steps of performing calibration tests 310 on the scanner and measuring the wavefront using a wavefront sensor. Based on the obtained measurement data (e.g., wavefront or aberration measurements, which may be in the form of Zernike coefficients), model parameters 320 are fitted to a general (e.g., a black box based on physical properties) model. Steps 300 to 320 are repeated for each application (e.g., each layer, scheme, etc.), and the resulting fitted application-specific model parameter values ​​are stored in a database 330.

[0048] ASCAL is an effective method for reducing the effects of lens heating. However, ASCAL requires at least 1 to 2 hours of dedicated tooling time per layer, and for in-line monitoring and calibration (e.g., within the scanner), test batches need to be exposed and measured for each new mask. This is a significant overhead in terms of the time required on the scanner.

[0049] To address this time constraint, alternative strategies have been employed in the case of simulation measurements. This approach is known as computational ASCAL or cASCAL. For example, cASCAL is described in Beak et al., “Lens heating impact analysis and controls for critical device layers by computational method” (Proceedings of the International Society for Optical Engineering (SPIE), March 2014), which is incorporated herein by reference. In contrast to ASCAL, which generates an LHFF dataset based on measured data, cASCAL generates the LHFF dataset solely through computational simulation. This results in significant scanner time savings and therefore no loss of scanner availability.

[0050] Figure 3 (b) The main steps of the cASCAL process are illustrated. Using application data 340, which may include mask layout data, optical settings, and illumination conditions (e.g., illumination pupil data), a corresponding diffraction pattern 350 is calculated computationally. Based on this simulated diffraction pattern and a lens heating model for the lens used, a lens heating response is simulated. The simulated lens heating response is used to determine application-specific model parameter values ​​360, which are stored in a database 370. This process is repeated for each application as previously described.

[0051] Figure 4This is a flowchart describing the LH workflow for cASCAL. A mask description 400 (which may be in the form of a graphical database system or GDS archive) is combined with illumination data 410 to obtain a diffraction pattern 420. A general lens model 430 (e.g., for a specific lens family or lens type) undergoes a first adjustment step 440 to adjust the model parameters for the specific lens to obtain a lens-specific model 445. This step can be a one-time step performed by the scanner manufacturer before scanner installation. Then, the diffraction pattern 420 is used in a second adjustment step 450, which includes simulating and predicting lens heating on the lens-specific model 445. The result of this second adjustment step is an application-specific model 455. All these steps are computational and do not require physical wafer exposure and readout. The method can be performed on a mask / layer basis to obtain an application-specific model library 455 for each layer. In a later stage, each application-specific model 455 is used when the corresponding application-specific model is exposed to calculate / predict lens heating (LH) and its correction during production 460.

[0052] The problem with the cASCAL framework described above is that it is not as accurate as the measured ASCAL framework. Another problem is that mask data (GDS archives) cannot always be used to calculate diffraction patterns.

[0053] Therefore, it is proposed to address one or both of these problems by defining a machine learning (e.g., data-driven) model to predict lens heating during production (i.e., in real time). The model can be a neural network, such as a deep artificial neural network (DANN) (but any other suitable machine learning model can be used). This neural network model can complement the currently used physical property-based lens-specific models in cASCAL, such that their predictions are improved by the neural network (e.g., by adjusting the model parameters of the neural network).

[0054] In this embodiment, the neural network does not require a training set for each new mask / layer, nor does it require specific initial training for each new mask / layer; instead, the neural network will improve and learn during production as the scanner operates (real-time training). This can be achieved by using real-time data from a wavefront sensor (which is typically used for mask alignment before exposing each wafer) and, for example, keeping it as close as possible to the target value (minimizing deviation from the target).

[0055] Alternatively, the neural network can be trained during the initial training phase, for example, on actual production data (e.g., historical data from previous batches). Training a neural network means changing the weights present in the network. This means that the output of the neural network can change during training, regardless of whether the input is the same. When the scanner operates, a model whose performance is predictable can preferably be used. Therefore, the neural network can be trained on production data, but preferably, during training, this "output" of the neural network (adjustments to the lens model) is not used in production only while the neural network is being trained. Thus, once training is complete and the performance is satisfactory, the output of such a network can be used.

[0056] Figure 5 This is a flowchart of the improved method according to an embodiment. Steps / items 500 to 560 correspond to... Figure 4 The corresponding steps / items of the method are 400 to 460. The difference lies in that: wavefront sensor data (e.g., aberration data) 565 is collected during batch production 560 and used to train a neural network (or a second model) 570 to adjust the parameters of the primary lens model 550 (or lens heating model) or the first model 545; for example, to obtain an application-specific model 555 that allows the wavefront sensor data to improve aberration performance such as that measured by the wavefront sensor. The neural network 570 thus adjusts the lens model 545 for a specific layer / mask (defined by mask data 500) and the application during production using said mask. In other words, this neural network 570 can be considered as an additional feedforward control created immediately following the current feedforward control and adapted to the layer currently in production.

[0057] Figure 6 This is a flowchart of an improved method according to another embodiment, wherein the output of wavefront sensor 665, such as raw pixelated sensor data (image data) instead of Zernike decomposition data, is used together with illumination pupil 610 to calculate diffraction pattern (or related properties, such as transmittance profiles or lens heating profiles) 620. As already mentioned, mask data 400, 500 are not always available. In this way, raw wavefront data 665 is used to resolve mask-specific LH components, and the first model 655 and / or the second model 670 (alone or in combination) do not depend on any mask (design) data, nor on any time-consuming wafer exposure and readout. The wavefront data 665 used to calculate the diffraction pattern may include a difference wavefront between the exposure step and the exposure step.

[0058] This allows for smarter and dynamically updated control over the lens model parameters. For example, the neural network 670 can use the original sensor wavefront 665 to train an unadjusted lens model 645 for a specific layer, so that the adjusted lens model 655 can better predict lens drift characteristics based on the wavefront information 665.

[0059] As already explained, wavefront measurements performed during batch exposure include valuable information about the heating profile of the lens elements (near the pupil plane) and thus indicate the mask diffraction pattern. This is because the details of the pattern layout on the mask are reflected by its diffraction pattern in the exit pupil of the scanner's projection lens. Therefore, the mask's characteristic layout (and transmittance) largely determines how the projection lens is heated (e.g., the spatial intensity distribution at its optical surface) and thus how the wavefront error induced by its heating evolves. One or more wavefront sensors can be used to measure the temporal evolution of the wavefront error (pixelated wavefront), and due to the above, the temporal evolution of the wavefront error indicates the mask-specific lens heating component. Therefore, typically, in the absence of mask design data, raw (e.g., pixelated) wavefront data can be used to determine / adjust LHFF model parameters. Typically, the wavefront is measured during mask alignment prior to exposure of each wafer.

[0060] Steps / components 610 to 670 correspond to Figure 5 The corresponding steps / elements 510 to 570 in the flowchart are described. However, it should be noted that there is no equivalent to mask data 500, as mask data (e.g., GDS files or the like) is not used in this embodiment. Therefore, the step of calculating the diffraction pattern 620 uses raw wavefront data 665 (e.g., difference wavefront) from the wavefront sensor and illumination data 610 to calculate LHFF model parameters 650. The calculated diffraction pattern 620 and / or wavefront data 665 (e.g., raw images and / or processed aberration data), which can be obtained before exposing each wafer, are used to train a second model 670 to adjust the first model 645 for a specific application (layer or mask) to obtain an application-specific model 655.

[0061] As mentioned, Figure 5 or Figure 6 This process can be altered so that the neural network is trained on historical data and not further trained during actual production; instead, the (fixed) trained neural network is used in production to adjust the lens model on production data.

[0062] All of the above methods (e.g., Figure 5 and Figure 6The process described above may include an initial training step for training a third model 580, 680 (e.g., machine learning, neural network, or DANN), which trains an initial general lens model 530, 630 for a specific lens (e.g., modeling a type of lens, where such a lens model may be supplied by a lens vendor) (e.g., as part of a scanner setup phase that may be performed before the scanner is installed). Thus, the third model 580, 680 may perform steps 540, 640 of the flowchart described above. Given a specific lens type, the general lens model 530, 630 is reinforced using, for example, test data 575, 675 from a lens heating test performed on the specific lens during setup. The neural network 580, 680 trained on such test data 575, 675 may be generated specifically for each lens type; for example, using several (e.g., similar) mask inputs. Similarly, this third model 580, 680 may reinforce the first model 530, 630 (and the second model 570, 670). The lens models 530 and 630 based on physical properties can be reinforced by neural networks 580 and 680 specific to each lens type and by another neural network 570 and 670 operating in real time, which are adapted to the layer currently being printed.

[0063] While the above embodiments disclose concepts related to lens heating correction, these concepts are not limited thereto and should not be interpreted in this way. For example, the same concepts can be applied to predicting and correcting mask heating and wafer heating. In such cases, the first model is, where appropriate, a mask heating model or a wafer heating model. In such embodiments, the output of the workflow (from feedforward corrections of a first model, such as those reinforced by at least a second model) can be used to actuate the wafer and / or mask platform, respectively. Therefore, this can improve LH prediction at many levels: e.g., inter-wafer, inter-field, and / or intra-field.

[0064] In another embodiment, the raw wavefront data can be used in a lens setup step, which transforms the lens model from its initial calibration state (typically performed on a cold lens) into a starting point for optimization for a specific mask / layer.

[0065] The goal of lens modeling is to optimize lens aberrations within a lithography system. In some cases, the optimization objective may be zero aberration; otherwise, it may be another distinguishing feature, often expressed in terms of Zernike coefficients. Lens quality and lens setup optimization determine how well the optimization objective is achieved for a given system. The optimization objective can be constrained by an evaluation function consisting of a set of terms with a primary least-squares term, often expressed as a weighted sum of Zernike coefficients using values ​​of weights associated with the coefficients of odd and / or even (lower and / or higher-order) Zernike polynomials (e.g., equal to (1, 1.2, 0.5)). This can be applied to both lens setup (e.g., using a calibration lens model) and production (e.g., using a driver lens model). Such a choice can be based on simple and qualitative considerations. Lens calibration and production performance can be optimized to the same objective for all machines. While Zernike coefficients are independent polynomials, changes in the current state of technology may explicitly force optimization methods to minimize some Zernike coefficients at the expense of others. This is because lenses (and lens elements) can affect more than one Zernike coefficient in different ways.

[0066] This method is described in US2019227441A1, which is incorporated herein by reference. This method may include: receiving an initial evaluation function comprising an initial set of model parameters and weights; and executing an optimization algorithm starting from the initial set of model parameters and weights to determine a second evaluation function comprising a second set of model parameters and weights. The optimization algorithm scores different sets of model parameters and weights based on projection system characteristics of the projection system adjusted according to the output of the lens model using the evaluation function with the set of model parameters and weights. When executed using the second evaluation function, the projection system can be adjusted using optical element adjustments from the output of the lens model. The optimization algorithm may generate other evaluation functions between the first and second evaluation functions. The method can be used to perform a thorough search across the entire complex solution space of the lens model to find suitable model parameters and weights that improve the performance of the lens model. The set of model parameters may correspond to at least some of the lens model characteristics. The set of model parameters may correspond to lens model constraints, such as iterative cutoffs involved in the algorithms or mathematical techniques used in the lens model, such as, for example, single-valued decomposition.

[0067] The optimization described above is determined using a finite number of Zernike polynomials (typically 64), which do not allow for a complete representation of the lens's distinguishing features. Therefore, this paper proposes using raw wavefront sensor data to set and adjust the lens. This enables the capture and corresponding correction of all lens distinguishing feature contents. This approach can be implemented in software (e.g., running on a scanner or offline processing device) that can periodically provide raw wavefront sensor data and continuously evaluate the wavefront.

[0068] The method may include a first adjustment phase for determining layer-specific corrections to a calibrated lens. Based on the heating profile, product mask characteristics (e.g., diffraction patterns derived therefrom) can be inferred using methods already disclosed herein. The heating profile can describe wavefront evolution as the lens heats up; for example, the wavefront difference between a cold lens and a hot lens, or between a relatively cold lens and a hotter lens measured when the mask is exposed. This diffraction pattern can be used to determine product-specific initial lens corrections (e.g., correction sub-fits) for a lens model; for example, the diffraction pattern or the determined mask characteristics can then be combined with the measured wavefront, and optimization can be performed to determine such product-specific initial lens corrections. Product-specific initial lens corrections can be applied to the calibrated lens model. Thus, the application of such product-specific initial lens corrections will change the lens from an initial calibrated state to an initial (starting) state optimized for printing a given layer (e.g., for a specific mask). This can then be used to optimize sub-fits towards a target wavefront (e.g., zero aberration or otherwise) using the mask.

[0069] This embodiment can use an evaluation function to define an image-based rather than polynomial-based optimization objective. The optimization algorithm can be an evolutionary or genetic algorithm with a fitness function configured to score different sets of model parameters and weights. In other embodiments, an ANN or DANN can be used to determine sub-fitting schemes. Optimization can determine lens correction to adjust the measured wavefront to a target wavefront. The optimization and evaluation functions can be similar to those described above (e.g., as described in US2019227441A1), except that the optimization and evaluation functions are performed on the raw wavefront (image) data instead of on the Zernike decomposition.

[0070] This first stage can be performed offline (e.g., using historical data) to determine a static, product-specific initial lens correction or sub-matching scheme for the mask. This operation can be repeated for all masks for the product, allowing the generation of a sub-matching scheme library, each corresponding to a different mask. In this way, during production, the corresponding sub-matching scheme for each mask can be applied to the lens model before exposure using the mask.

[0071] In a variation, instead of determining mask characteristics based on wavefront data, the mask file or (at a more basic level) the mask transmission factor (e.g., the percentage of light transmitted by the mask) and the proposed illumination settings are combined with the measured wavefront data, and image-based optimization is performed to determine product-specific initial lens correction.

[0072] The second stage, during exposure of the same layer in production, may include monitoring wavefront data to check if the lens has drifted from the optimized state achieved using the sub-matching scheme. If drifting is found, in-situ calibration may correct the lens settings (e.g., via a lens manipulator). This may include another (image-based) optimization based on a mask diffraction pattern determined according to the original wavefront data and lens characteristics (i.e., actuator constraints) to optimize the wavefront.

[0073] In this way, sophisticated optimization techniques can be used to achieve rapid convergence of the lens manipulator settings to the target wavefront.

[0074] The above concepts can be extended to additionally utilize per-field alignment and / or level sensor data in the optimization process to determine the optimal lens settings for each field. This approach could include performing per-field lens optimization based on the raw wavefront data and per-field data to achieve real-time optimal focal length and overlap (e.g., real-time correction of focal length and overlap).

[0075] A similar process can be applied to lens setup, the only difference being the setup point: instead of using a product mask (as described above) to determine product-specific initial lens calibration, a test mask is used to perform the above process.

[0076] This embodiment can be combined with Figure 5 or Figure 6 This embodiment can be implemented to provide a starting point for improvements to production steps 560, 660, and optionally monitoring steps. Alternatively, this embodiment can be implemented alone.

[0077] The output of the adjusted lens model may include residual optical aberrations present in the projection system. In this case, optimization algorithms can be used to give better scores to model parameters and weights that result in lower residual optical aberrations. Projection system characteristics need not be optical aberrations and may instead or additionally include overlap error, root mean square wavefront error, focal length error, optical element adjustment constraints, and other characteristics that those skilled in the art will understand from the teachings herein. A focusing example will now be described.

[0078] In another embodiment, a method similar to the lens setup optimization described above will be used to improve focal length calibration and measurement for lens models.

[0079] Currently, focal length testing is used to determine focal length calibration (e.g., to calibrate Zernik coefficients 4 and 5, which are associated with the primary (feature-independent) defocus Zernik coefficients). Specialized masks can be used to expose multiple verification fields on a test substrate at different focal length shifts, each verification field comprising multiple focus marks (e.g., approximately 250 marks). These focus marks on the mask produce corresponding marks on the substrate, which can be read (e.g., using an alignment sensor) to infer the focal length setting during the exposure. Because this testing involves the exposure of the wafer, it is time-consuming. The remaining focal length (even number) Zernik coefficients are calibrated using a wavefront sensor: its signal is typically decomposed into the first 64 Zernik coefficients as already described. During production, mask alignment and focal length drift are measured and corrected using the Zernik decomposition of wavefront sensor data. As already described, this polynomial representation, because it is truncated to 64 or 100 Zernik coefficients, does not allow for a complete representation of the lens's distinguishable features.

[0080] Therefore, a method is proposed to supplement or replace current methods for focal length measurements as described above (e.g., in setup and / or during production) based on a series of original wavefront measurements performed at different focus levels. These measurements enable focal length estimation in a more accurate and efficient manner.

[0081] In the setup phase, it is proposed to use the measured wavefront (or, as described, a difference wavefront) of a heated or thermally applied lens to determine the optimal focal length for a specific mask or product feature to be imaged. The method involves repeatedly exposing the product feature to heat the lens and subsequently measuring the (raw) wavefront, each repetition performed at a different focus level (e.g., covering a focal length range including the expected optimal focal length value). Each measurement will result in a different wavefront, which can be used to determine the optimal focal length via optimization; for example, the optimal focal length for a preferred or specific wavelength.

[0082] The optimization can be similar to the optimization methods described in the previous embodiments (e.g., ANN or evolutionary algorithms), which optimize the raw wavelength data instead of the Zernike representation (or use the raw wavelength data instead of the Zernike representation as input). The main difference is that the optimization in the previous embodiments was described in terms of optimizing wavefront aberrations (and therefore lens manipulator correction), and this optimization specifically optimizes the focal length, and therefore optimizes the optimal relative platform position between the mask and the substrate platform in the z-direction (i.e., perpendicular to the substrate plane). This optimization can be performed alone or in addition to the aforementioned optimization for lens sub-matching schemes (and optionally with...). Figure 5 and Figure 6 (A combination of methods). Alternatively, if the wavefront data used for optimization involves different focus levels, joint optimization can be performed for lens settings (product-specific initial lens correction) and optimal focal length.

[0083] In addition to performing tests during setup to eliminate the need for performing this focal length test, the above method can also be used in production to determine the optimal focal length in real time and monitor optimal focal length drift in order to improve mask alignment. The method may include repeatedly performing the same optimization, including new wavefront data associated with each new wafer, to determine whether the optimal focal length has drifted due to setup, and if so, to determine a new focal length correction. As a specific example, to check whether the optimal focal length has drifted, three (or more) wavefront sensor measurements or captures can be performed (i.e., via a thermal lens) after each wafer exposure: one wavefront sensor measurement or capture at the current optimal focal length, one wavefront sensor measurement or capture at the current optimal focal length with a smaller displacement in a first direction, and one wavefront sensor measurement or capture at the current optimal focal length with a smaller displacement in a second direction. These measurements can be compared to determine whether the optimal focal length has drifted; if so, the optimal focal length or optimal focal length can be adjusted accordingly.

[0084] In another embodiment, a lens matching method using raw wavefront data is proposed. As already stated, the goal of lens setup is to optimize lens aberrations within the lithography system: for example, to optimize to the point where lens aberrations can be zero. Lens quality and lens setup optimization determine the degree to which the goal is achieved for a given system. Each lens has a different lens distinguishing feature (LFP), and therefore performance has a population distribution. The population distribution is directly related to the matching machine overlap (MMO). For later lens models, there is a change in the average LFP of the population. To keep the average LFP constant relative to previous machines, a non-zero target equal to the previous population average is introduced. Therefore, the current state of the art is to use a single target lens distinguishing feature for all lenses of a given type.

[0085] The selected target can be based on the best average performance of the selected lens as measured at the lens manufacturer. This target may differ from the average group measured at the lithography manufacturer and also from the group at a specific wafer fab. All machines used by all users and all wafer fabs are set to the same target, so statistical data and group analysis do not differentiate between groups for each user.

[0086] Users have only a finite set of machines with group averages that typically differ from the global group average. Therefore, it makes sense to target a given user's own group average rather than the global group average. Current methods targeting the global group average reduce the likelihood of machine-to-machine intra-field distinguishing feature matching.

[0087] Furthermore, and most importantly, the current scanner matching test uses a test mask. If the test mask is then used for measurement while using the product mask, the matching performance of the two scanners may differ considerably. Typically, the machine's matching performance is not ideal relative to a production environment for the following reasons:

[0088] • Perform matching on a cold lens;

[0089] • If matching is performed on a hot lens, the mask used is not the production mask;

[0090] • Matching (cold or hot) is performed on one layer at a time, and the matching does not take into account the lens behavior on different layers (i.e., masks).

[0091] Therefore, it is proposed to use raw wavefront data to predict the goodness of scanner matching for a specific product layer (mask). This can be achieved by: exposing the same product mask on different machines (e.g., one adjustment batch / 25 adjustment wafers per machine); recording the wavefront before exposing each wafer; and using this mask to correct the dynamic lens behavior of these machines so that the matching is optimized (e.g., maximizing yield when the mask is exposed on two or more machines).

[0092] It should be understood that this is different from using [the technology / method / etc.]. Figure 5 and Figure 6The first embodiment describes a single-lens heating calibration performed on the raw wavefront data. This embodiment aims to minimize the (e.g., overlap) effect of lens heating on a single scanner at that point. This embodiment aims to match the dynamic behavior of two (or more) different scanners on the same mask; this is not achieved by separately calibrating the lens heating behavior of the first scanner and then calibrating the second scanner to match the dynamic behavior of the first scanner as closely as possible. Instead, the method includes co-optimization to match the dynamic behavior of different scanners to be as close as possible (e.g., to maximize yield or other performance parameters).

[0093] This approach may include using a matching model or algorithm, such as an evolutionary or genetic algorithm, which operates to adjust a lens model and / or adjust a model that adjusts the lens model based on wavefront data obtained during exposure using a specific (e.g., production) mask.

[0094] Therefore, the matching model can operate directly on the lens model during setup or matching operations, thereby adjusting the lens model in a manner similar to the "second model" disclosed above, but using a co-optimization approach. This co-optimization approach can provide co-optimization adjustments for each lens model (e.g., co-optimization adjustments of model weights / coefficients), which optimize performance indicators, such as the yield of the product produced on both (or both) of the matched scanners. It can be understood that while neither lens model in the lens model can "sense" the other lens model, the matching model or algorithm senses both to provide co-optimization adjustments.

[0095] Alternatively, the matching model or algorithm can skillfully operate on a corresponding second model for each lens. Each of these second models (e.g., an ANN) is dedicated to its corresponding lens model and (as already described) tuned for the specific mask used. Thus, each second model is unaware of the other one or more second models. The matching model can skillfully operate on these second models to tune them in a co-optimized manner, providing a co-optimized tuning for each second model (e.g., a co-optimized tuning of model weights / coefficients). For example, the matching model can receive trained second models as input and attempt to find weights for a new or tuned one or more second models (one or more neural networks). The purpose of this new neural network is to match the scanner when it is generated. Ideally, the new neural network is capable of matching the performance of the matched scanners, such that layers can be exposed on any of these scanners without performance differences. In this embodiment, the new neural network can be trained using the same data used to train the earlier (unmatched) neural network.

[0096] To provide a specific example, suppose, five different lenses of the same type. Theoretically, given the same mask, dose, and irradiation curve, these five lenses should then behave in very similar ways (the only difference being the physical differences between the five lenses). The same type of lens setup is performed using the same dose, mask, and irradiation curve. This means that, using this data, it is possible to train a neural network capable of predicting the behavior of these lens types for this dose, mask, and irradiation curve. Now imagine these five lenses are installed in a wafer fab. Using production data (separated into different transmission factors, doses, and irradiation curves), it is possible to reinforce (i.e., further refine) the neural network capable of predicting the behavior of such lenses to obtain different neural networks, each trained to a specific combination of transmission factors, dose, and irradiation curves. If, when using the same five machines, it is desired to expose layers with some differences in transmission factor, dose, or irradiation profile (e.g., for a mask with a new transmission factor but exposed with the same dose and irradiation profile), then the matching model can use a neural network generated for those lenses to produce another neural network that performs well for the new transmission factor, dose, and irradiation profile, without needing to train such a new neural network on data. In other words, if a library of neural networks for transmission factor, dose, and irradiation profile exists, it is possible to generate a new neural network (for the new transmission factor, dose, or irradiation profile) without any training data.

[0097] This matching model can be used only in the initial matching step to adjust the lens model or the second model based on an initial adjustment wafer exposed using a product mask. During such a matching step, the matching model can receive the corresponding model weights of the model it is adjusting, as well as wavefront data, to evaluate the matching. Based on this, optimization can be performed by adjusting the model weights to improve the matching performance for the specific mask.

[0098] Once the machine is matched, there is no longer a need for a matching model / algorithm, and the adjusted lens model or second model is used during production.

[0099] In cases where the second model is adjusted by the matching model, these second models will have been trained or optimized to adjust their corresponding lens models (in real time) to maximize yield (or optimize other parameters of interest) based on wavefront data measured during production for masks exposed on both (or both) of the two scanners.

[0100] Similar techniques can be used to match two (or more) different layers exposed on the same scanner to improve overlap. A test batch of layers can be exposed while recording the wavefront before exposing each wafer for each layer. Optimization can then refine the model weights used for the improved dynamic lensing behavior on different layers to improve, for example, overlap or other parameters of interest.

[0101] It should be noted that in all embodiments described above using raw wavefront data, Zernike decomposed data is alternatively used within the scope of this disclosure. The advantage of raw data is that it contains all Zernike coefficients, while the decomposed data contains only a finite number of Zernike coefficients (typically 64 or 100). Therefore, all optimizations described above (lens setup, determining optimal focal length, machine matching, etc.) can be performed using either raw data or Zernike decomposed data. Better optimization can be expected using raw data. This is because (roughly speaking) in the case of raw data, a decision about which Zernike coefficient(s)(s) to consider is made during optimization (posterior). If only Zernike decomposed data is used, a decision about which Zernike coefficient(s)(s) to consider is made at the outset (a priori).

[0102] Further embodiments are disclosed in a subsequent list of the following numbered aspects:

[0103] 1. A method for determining corrections for a photolithography process performed by a photolithography apparatus, the method comprising:

[0104] A first heating model is obtained, which models the effect of heating the component during the photolithography process;

[0105] Obtain the measured wavefront data related to the objective lens of the lithography apparatus, measured during the lithography process; and

[0106] The first heating model and the measured wavefront data are used to determine the correction for the photolithography process to correct the heating.

[0107] 2. The method according to aspect 1, comprising:

[0108] Calculate the characteristics of the pattern forming apparatus used in the photolithography process;

[0109] The first model uses the calculated diffraction pattern to determine the correction.

[0110] 3. The method according to aspect 2, wherein the characteristics of the pattern forming apparatus include one or more of the following: diffraction pattern, lens heating curve, and transmittance curve of the objective lens.

[0111] 4. The method according to aspect 2 or 3, wherein the step of calculating the characteristics of the pattern forming apparatus uses pattern forming apparatus data as obtained from the description of the pattern forming apparatus.

[0112] 5. The method according to aspect 2 or 3, wherein the step of calculating the characteristics of the pattern forming apparatus uses the measured wavefront data to calculate the diffraction pattern, the measured wavefront data depending at least in part on the layout of the pattern forming apparatus.

[0113] 6. The method according to aspect 5, wherein the measured wavefront data for calculating the characteristics of the pattern forming apparatus includes raw wavefront sensor data, the raw wavefront sensor data including a pixelated image.

[0114] 7. The method according to any of the foregoing aspects, wherein the measured wavefront data relates to the difference wavefront after the exposure step and before the exposure step.

[0115] 8. The method according to any of the foregoing aspects, wherein the step of determining the correction includes:

[0116] Obtain the second model;

[0117] The second model is trained using the measured wavefront data and / or characteristics of the pattern forming apparatus to adjust the first model used in the lithography process.

[0118] 9. The method according to aspect 8, wherein the second model adjusts the first model for the photolithography process being performed.

[0119] 10. The method according to aspect 8 or 9, wherein the second model is a machine learning model.

[0120] 11. The method according to aspect 10, wherein the second model is a neural network.

[0121] 12. The method according to any of the foregoing aspects, wherein the first model is a model based on physical properties.

[0122] 13. The method according to any of the foregoing aspects, wherein the first model is universal for multiple different lithography processes.

[0123] 14. The method according to aspect 13, wherein the plurality of different photolithography processes are associated with a plurality of different pattern forming apparatuses.

[0124] 15. The method according to any of the foregoing aspects, wherein the first model is a lens heating model.

[0125] 16. The method according to aspect 15, wherein the first model is further enhanced by a third model, the third model being trained to adjust the first model for the specific projection lens used.

[0126] 17. The method according to aspect 16, wherein the third model is a machine learning model.

[0127] 18. The method according to aspect 17, wherein the third model is a neural network.

[0128] 19. The method according to aspects 16, 17 or 18 includes an initial step of training the third model for the particular projection lens based on training data from lens heating tests.

[0129] 20. The method according to any one of aspects 1 to 14, wherein the first model is one of a pattern forming apparatus heating model or a substrate heating model.

[0130] 21. The method according to any of the foregoing aspects, wherein the correction includes feedforward correction.

[0131] 22. The method according to any of the foregoing aspects, including using the correction to correct the photolithography process.

[0132] 23. An apparatus capable of operating to perform the method according to any of the foregoing aspects, comprising:

[0133] A wavefront sensor, operable to measure the wavefront data; and

[0134] A processor device, which is operable to perform other steps of the method.

[0135] 24. A photolithography apparatus, comprising the apparatus according to aspect 23.

[0136] 25. The photolithography apparatus according to aspect 24, comprising:

[0137] An irradiation system used to provide measurement illumination;

[0138] The projection optics;

[0139] A pattern forming apparatus support for supporting a pattern forming apparatus; and

[0140] Substrate support for supporting a substrate.

[0141] 26. A method for adjusting a photolithography process for a particular patterning apparatus, the method comprising: obtaining wavefront data associated with an objective lens of a photolithography apparatus, the wavefront data being measured after exposure of a pattern on a substrate using the particular patterning apparatus; determining a pattern-specific wavefront contribution based on the wavefront data and a reference wavefront, the pattern-specific wavefront contribution being associated with the particular patterning apparatus; and using the pattern-specific wavefront contribution to adjust the photolithography process for the particular patterning apparatus.

[0142] 27. The method according to aspect 26, wherein the adjustment is based on raw wavefront data including a scalar representation of the wavefront error as a mapping over the entire exit pupil plane of the objective lens.

[0143] 28. The method according to aspect 26 or 27, wherein at least some of the wavefront data are related to the target wavefront.

[0144] 29. The method according to aspect 28, wherein the target wavefront is a difference target wavefront relating to the difference between a wavefront prior to exposure of the substrate and a wavefront after exposure of the substrate.

[0145] 30. The method according to any one of aspects 26 to 29, wherein the pattern-specific wavefront contribution includes the difference between the measured wavefront of the wavefront data after the exposure and the reference wavefront.

[0146] 31. The method according to aspect 30, wherein the reference wavefront includes a measured wavefront that has been measured prior to the exposure.

[0147] 32. The method according to any one of aspects 26 to 31, wherein the adjustment step includes adjusting a first model associated with a component of the lithography apparatus using the wavefront data, such that the first model is adjusted for the specific pattern forming apparatus.

[0148] 33. The method according to aspect 32, wherein using the wavefront data includes using the wavefront data to train a second model and using the trained second model to adjust the first model.

[0149] 34. The method according to aspect 33, comprising:

[0150] The pattern-specific wavefront contribution is used to calculate the characteristics of the specific pattern-forming apparatus used in the photolithography process; and

[0151] The characteristics of the specific pattern forming apparatus are used in the step of adjusting the first model using the second model.

[0152] 35. The method according to aspect 34, wherein the characteristics of the particular pattern forming apparatus include one or more of the following: diffraction pattern, lens heating curve, and transmittance curve of the objective lens.

[0153] 36. The method according to any one of aspects 33 to 35, wherein the adjustment step is performed in real time on the production substrate using the wavefront data related to the exposure of the production substrate.

[0154] 37. The method according to any one of aspects 33 to 36, wherein the second model is a machine learning model.

[0155] 38. The method according to aspect 37, wherein the second model is a neural network.

[0156] 39. The method according to any one of aspects 33 to 38, wherein the method comprises:

[0157] Obtain the matching model or algorithm; and

[0158] There exists at least one additional first model and at least one additional second model, each of the at least one additional first model being associated with a different lithography device, and the at least one additional second model being used to adjust its corresponding first model; and

[0159] The adjustment steps include: for the specific patterning apparatus, performing a joint optimization adjustment of each of the first model and / or its corresponding second model, using the matching model or algorithm and the pattern-specific wavefront contribution, in order to maximize the yield of the substrate exposed by each lithography device in the lithography apparatus or with respect to the performance parameters of the yield.

[0160] 40. The method according to aspect 39, wherein the matching model or algorithm includes an evolutionary algorithm or a genetic algorithm.

[0161] 41. The method according to any one of aspects 32 to 38, wherein the method comprises:

[0162] A matching model or algorithm is obtained; wherein there exists at least one additional first model, each of the at least one additional first model being associated with a different lithography apparatus; and the adjustment step includes using the matching model or algorithm and the pattern-specific wavefront contribution to perform a co-optimization adjustment of each of the first models for the specific pattern forming apparatus.

[0163] 42. The method according to aspect 41, wherein the common optimization adjustment is performed in terms of maximizing the yield of the substrate exposed by each of the lithography apparatuses in the lithography apparatus, or the performance parameters relating to the yield.

[0164] 43. The method according to any one of aspects 32 to 42, comprising obtaining a layer matching model or algorithm;

[0165] Among them, there are at least two patterning apparatuses to be exposed on the same device to expose different layers of the same device on a common substrate; and

[0166] The adjustment step includes using the layer matching model or algorithm and the pattern-specific wavefront contribution to perform a common optimization adjustment of each of the first models in the first model for the pattern forming apparatus.

[0167] 44. The method according to aspect 41, 42 or 43, wherein the common optimization adjustment is performed in terms of maximizing the yield of the common substrate or the performance parameters with respect to the yield.

[0168] 45. The method according to any one of aspects 41 to 44, wherein the matching model or algorithm and / or the layer matching model or algorithm comprises an evolutionary algorithm or a genetic algorithm.

[0169] 46. ​​The method according to any one of aspects 32 to 45, comprising using the first model and the wavefront data to determine a correction for the photolithography process.

[0170] 47. The method according to aspect 46, wherein the correction includes feedforward correction.

[0171] 48. The method according to aspect 46 or 47, including using the correction to correct the photolithography process.

[0172] 49. The method according to any one of aspects 32 to 48, wherein the first model is further enhanced by a third model, the third model being trained to adjust the first model from a general model for the type of said component to a specific model for said particular component.

[0173] 50. The method according to aspect 49, wherein the third model is a machine learning model.

[0174] 51. The method according to aspect 50, wherein the third model is a neural network.

[0175] 52. The method according to aspect 49, 50 or 51, including an initial step of training the third model for the particular component based on training data associated with the particular component.

[0176] 53. The method according to any one of aspects 32 to 52, wherein the first model is a model based on physical properties.

[0177] 54. The method according to any one of aspects 32 to 53, wherein the first model is one of the following:

[0178] A lens heating model, wherein the component is a lens system or a lens of the lens system; a pattern forming apparatus heating model, wherein the component is a pattern forming apparatus; or

[0179] Substrate heating model, wherein the component is a substrate.

[0180] 55. The method according to any one of aspects 32 to 54, wherein the adjustment step includes determining a process-specific initial lens correction for the first model, the process-specific initial lens correction adjusting the first model to an initial state specific to the pattern forming apparatus.

[0181] 56. The method according to aspect 55, wherein determining the process-specific initial lens correction includes performing optimization on the wavefront data using the pattern-specific wavefront contribution to optimize the wavefront to a target wavefront.

[0182] 57. The method according to aspect 56, wherein the optimization step is performed using an evolutionary algorithm, a genetic algorithm, or a second model, the second model including a machine learning model.

[0183] 58. The method according to aspect 56 or 57, wherein the optimization step further uses alignment data and / or level measurement data for each field to determine the optimal lens setting for each field.

[0184] 59. The method according to aspects 56, 57 or 58, wherein the wavefront data includes wavefront data related to exposures performed using a warm lens at multiple different focus levels; and the optimization step further includes using the pattern forming apparatus to determine an optimal focal length value for the exposure.

[0185] 60. The method according to any one of aspects 55 to 59, wherein the method comprises determining characteristics of the pattern forming apparatus based on the pattern-specific wavefront contribution, and

[0186] The characteristics of the pattern forming apparatus are used in the step of determining the initial lens correction specific to the product.

[0187] 61. The method according to any one of aspects 55 to 60, wherein the determination of process-specific initial lens correction is performed in an initial setup phase, and the method further comprises:

[0188] The patterning apparatus is used on the production substrate during exposure:

[0189] Monitor wavefront data associated with the production substrate to monitor whether the wavefront data has drifted from the optimized state achieved using the process-specific initial lens correction; and

[0190] Corrections are determined based on the monitoring steps when deemed necessary.

[0191] 62. The method according to any one of aspects 26 to 54, wherein the wavefront data includes wavefront data related to exposures performed using a warm lens at multiple different focus levels; and the adjustment step includes using the pattern forming apparatus to determine an optimal focal length value for the exposure.

[0192] 63. The method according to aspect 62, wherein determining the optimal focal length value includes performing optimization on the wavefront data using the pattern-specific wavefront contribution to optimize the wavefront relative to the focal length.

[0193] 64. The method according to aspect 63, wherein the optimization steps are performed using an evolutionary algorithm, a genetic algorithm, or a machine learning model.

[0194] 65. The method according to aspects 62, 63 or 64, wherein,

[0195] The adjustments are performed in real time based on wavefront data associated with the production substrate during the setup phase and / or during exposure on the production substrate using the patterning apparatus.

[0196] 66. The method according to any one of aspects 62 to 65, comprising monitoring whether the optimal focal length has drifted during production by:

[0197] Three or more wavefront sensor measurements are performed after the substrate is exposed, one of the measurements is at the current optimal focal length, one of the measurements is at the current optimal focal length with a focal length drift in a first direction, and one of the measurements is at the current optimal focal length with a focal length drift in a second direction.

[0198] Comparing these three measurements can determine whether the optimal focal length has drifted; and

[0199] If the optimal focal length has drifted, the optimal focal length value is adjusted to the updated current optimal focal length value.

[0200] 67. An apparatus capable of operating to perform the method according to any one of aspects 26 to 66, the apparatus comprising:

[0201] A wavefront sensor, operable to measure the wavefront data; and

[0202] A processor device, which is operable to perform other steps of the method.

[0203] 68. A photolithography apparatus, comprising the apparatus according to aspect 67.

[0204] 69. The photolithography apparatus according to aspect 68, comprising:

[0205] An irradiation system used to provide measurement illumination;

[0206] The projection optics;

[0207] A pattern forming apparatus support for supporting a pattern forming apparatus; and

[0208] Substrate support for supporting a substrate.

[0209] 70. A computer program comprising instructions for performing the method according to any one of aspects 26 to 66.

[0210] 71. A processor device capable of operating to execute a computer program according to aspect 70.

[0211] 72. A processor device operable to adjust a lithography process for a particular patterning apparatus, the device being configured to: acquire wavefront data associated with an objective of the lithography apparatus, the wavefront data being measured after exposure of a pattern on a substrate using the patterning apparatus; determine a pattern-specific wavefront contribution based on the wavefront data and a reference wavefront, the pattern-specific wavefront contribution being associated with the patterning apparatus; and adjust the lithography process for the patterning apparatus using the pattern-specific wavefront contribution.

[0212] While specific reference can be made to the use of embodiments of the invention in the context of optical lithography, it should be understood that embodiments of the invention can be used in other applications (e.g., imprint lithography) and are not limited to optical lithography where circumstances permit. In imprint lithography, the morphology in the patterning apparatus defines a pattern generated on a substrate. The morphology of the patterning apparatus can be pressed into a resist layer supplied to the substrate, where the resist is cured by applying electromagnetic radiation, heat, pressure, or a combination thereof. After the resist has cured, the patterning apparatus is removed from the resist, thereby leaving a pattern in the resist.

[0213] The terms “radiation” and “beam” as used herein encompass all types of electromagnetic radiation, including ultraviolet (UV) radiation (e.g., having wavelengths of 365 nm, 355 nm, 248 nm, 193 nm, 157 nm, or 126 nm) and extreme ultraviolet (EUV) radiation (e.g., having wavelengths in the range of 1 nm to 100 nm), as well as particle beams, such as ion beams or electron beams.

[0214] The term "lens" can refer to any optical component or combination thereof of various types, including refractive, reflective, magnetic, electromagnetic, and electrostatic optical components, when context permits. Reflective components are likely used in equipment operating in the UV and / or EUV range.

[0215] Therefore, the breadth and scope of the present invention should not be limited by any of the exemplary embodiments described above, but should be defined only by the appended claims and their equivalents.

Claims

1. A method for adjusting a photolithography process for a specific pattern forming apparatus, the method comprising: Wavefront data associated with the objective lens of the lithography apparatus is obtained, the wavefront data being measured after exposure of a pattern on a substrate using the specific patterning apparatus; A pattern-specific wavefront contribution is determined based on the wavefront data and a reference wavefront, the pattern-specific wavefront contribution being related to the pattern-forming apparatus. as well as The pattern-specific wavefront contribution is used to adjust the photolithography process for the specific pattern forming apparatus.

2. The method according to claim 1, wherein, The adjustment is made based on wavefront data, which includes a scalar representation of the wavefront error across the entire pupil plane of the objective lens.

3. The method according to claim 1 or 2, wherein, At least some of the wavefront data are related to the target wavefront.

4. The method according to claim 3, wherein, The target wavefront refers to the difference between the wavefront before exposure of the substrate and the wavefront after exposure of the substrate.

5. The method according to claim 1, wherein, The pattern-specific wavefront contribution includes the difference between a first measured wavefront, measured after the exposure, and the reference wavefront.

6. The method according to claim 5, wherein, The reference wavefront includes a second measurement wavefront that has been measured prior to the exposure.

7. The method according to claim 1, wherein, The adjustment step includes using the wavefront data to adjust a first model associated with a component of the lithography apparatus in order to adjust the first model for the specific pattern forming apparatus.

8. The method according to claim 7, wherein, Using the wavefront data includes using the wavefront data to train a second model and using the trained second model to adjust the first model.

9. The method according to claim 8, further comprising: The characteristics of the pattern-specific wavefront contribution used in the photolithography process are calculated. as well as The characteristics of the specific pattern forming apparatus are used in the step of adjusting the first model using the second model.

10. The method according to claim 9, wherein, The characteristics of the specific pattern forming apparatus include one or more of the following: diffraction pattern, lens heating curve, and transmittance curve of the objective lens.

11. The method according to claim 8, wherein, The adjustment steps are performed in real time on the production substrate using the wavefront data associated with the exposure of the production substrate.

12. The method according to claim 8, wherein, The second model is a neural network.

13. The method according to claim 8, wherein, The method includes: Obtain the matching model or algorithm; and There exists at least one additional first model associated with different lithography devices, and at least one additional second model for adjusting the corresponding first model; and The adjustment steps include: for the specific patterning apparatus, performing a joint optimization adjustment of each of the first model and / or its corresponding second model using the matching model or algorithm and the pattern-specific wavefront contribution, in order to maximize the yield of the substrate exposed using each of the lithography apparatuses in the lithography apparatus or the performance parameters related to the yield.

14. The method according to claim 7, wherein, The method includes: Obtain the matching model or algorithm; Among them, there exists at least one additional first model associated with different lithography equipment; and The adjustment step includes: for the specific pattern forming apparatus, performing a joint optimization adjustment of the first model and the other first model using the matching model or algorithm and the pattern-specific wavefront contribution.

15. A computer program comprising instructions for performing the method of claim 1.

Images