A method of extracting a feature from a data set includes iteratively extracting a feature from a data set based on a visualization of a residual pattern within the data set, wherein the feature is distinct from a feature extracted in a previous iteration, and the visualization of the residual pattern uses the feature extracted in the previous iteration. Visualizing the data set using the feature extracted in the previous iteration may include showing residual patterns of attribute data that are relevant to target data. Visualizing the data set using the feature extracted in the previous iteration may involve adding cluster constraints to the data set, based on the feature extracted in the previous iteration. Additionally or alternatively, visualizing the data set using the feature extracted in the previous iteration may involve defining conditional probabilities conditioned on the feature extracted in the previous iteration.

A method for determining a contribution of a processing apparatus to a fingerprint of a parameter across a substrate, the method including: obtaining a delta image which relates to a difference between a first pupil image associated with inspection of a first feature on the substrate and a second pupil image associated with inspection of a second feature on the substrate, wherein the first and second features have different dose sensitivities; determining a rate of change of the difference in response to a variation of a dose used to form the first and second features; selecting a plurality of pixels within the delta image having a rate of change above a predetermined threshold; and determining the contribution using the determined rate of change and the delta image restricted to the plurality of pixels.

A dark field digital holographic microscope is disclosed which is configured to determine a characteristic of interest of a structure. The dark field digital holographic microscope comprises an illumination device configured to provide at least: a first beam pair comprising a first illumination beam of radiation (1010) and a first reference beam of radiation (1030) and a second beam pair comprising a second illumination beam of radiation (1020) and a second reference beam of radiation (1040); and one or more optical elements (1070) operable to capture a first scattered radiation and to capture a second scattered radiation scattered by the structure resultant from the first and second illumination beams respectively. The beams of the first beam pair are mutually coherent and the beams of the second beam pair are mutually coherent. The illumination device is configured to impose incoherence (ADI) between the first beam pair and second beam pair.

Systems and methods disclosed are generally related to masklessly developing connections between a chip-group and a design connection point on a substrate. In placement of the chip-group on the substrate, according to certain embodiments the chip-group may be dispositioned relative to an expected position per a substrate layout design, causing a connection misalignment with the design connection point. According to certain embodiments, a machine learning (ML) model is trained on historical and simulated pixel models of chip-group connections and design connection points. Upon determining the chip-group misalignment by a metrology measurement, the trained ML model determines a pixel model to connect the misaligned chip-group, and causes the pixel model to be exposed to a substrate with a digital lithography tool, thereby connecting the dispositioned chip-group to the design connection point.

A process of characterizing a process window of a patterning process, the process including: obtaining a set of inspection locations for a pattern, the pattern defining features to be applied to a substrate with a patterning process, the set of inspection locations corresponding to a set of the features, the set of features being selected from among the features according to sensitivity of the respective features to variation in one or more process characteristics of the patterning process; patterning one or more substrates under varying process characteristics of the patterning process; and determining, for each of the variations in the process characteristics, whether at least some of the set of features yielded unacceptable patterned structures on the one or more substrates at corresponding inspection locations.

Disclosed is an illumination source for generating measurement radiation for an inspection apparatus. The source generates at least first measurement radiation and second measurement radiation such that the first measurement radiation and the second measurement radiation interfere to form combined measurement radiation modulated with a beat component. The illumination source may be a HHG source. Also disclosed is an inspection apparatus comprising such a source and an associated inspection method.

A method for characterizing post-processing data in terms of individual contributions from processing stations, the post-processing data relating to a manufacturing process for manufacturing integrated circuits on a plurality of substrates using a corresponding processing apparatus for each of a plurality of process steps, at least some of the processing apparatuses each including a plurality of the processing stations, and wherein the combination of processing stations used to process each substrate defines a process thread for the substrate; the method including: obtaining post-processing data associated with processing of the plurality of substrates in a cyclic sequence of processing threads; and determining an individual contribution of a particular processing station by comparing a subset of the post-processing data corresponding to substrates having shared process sub-threads, wherein a process sub-thread describes the process steps of each process thread other than the process step to which the particular processing station corresponds.

Methods for training a process model and determining ranking of simulated patterns (e.g., corresponding to hot spots). A method involves obtaining a training data set including: (i) a simulated pattern associated with a mask pattern to be printed on a substrate, (ii) inspection data of a printed pattern imaged on the substrate using the mask pattern, and (iii) measured values of a parameter of the patterning process applied during imaging of the mask pattern on the substrate; and training a machine learning model for the patterning process based on the training data set to predict a difference in a characteristic of the simulated pattern and the printed pattern. The trained machine learning model can be used for determining a ranking of hot spots. In another method a model is trained based on measurement data to predict ranking of the hot spots.

A device manufacturing method, the method comprising: obtaining a measurement data time series of a plurality of substrates on which an exposure step and a process step have been performed; obtaining a status data time series relating to conditions prevailing when the process step was performed on at least some of the plurality of substrates; applying a filter to the measurement data time series and the status data time series to obtain filtered data; and determining, using the filtered data, a correction to be applied in an exposure step performed on a subsequent substrate.

A method of determining a field of view setting for an inspection tool having a configurable field of view, the method including: obtaining a process margin distribution of features on at least part of a substrate; obtaining a threshold value; identifying, in dependence on the obtained process margin distribution and the threshold value, one or more regions on at least part of the substrate; and determining the field of view setting in dependence on the identified one or more regions.