Abstract:
Methods and metrology modules and tools are provided, which minimize an estimated overlay variation measure at misalignment vector values obtained from a derived functional form of an overlay linear response to non-periodic effects. Provided methods further quantifying target noise due to the non-periodic effects using multiple repeated overlay measurements of the target cells, calculating an ensemble of overlay measurements between the cells over the multiple measurement repeats and expressing the target noise as a statistical derivative of the calculated overlay measurements. Sub-ensembles may be selected to further characterize the target noise. Various outputs include optimized scanning patterns, target noise metrics and recipe and target optimization.
Abstract:
Methods are provided for deriving a partially continuous dependency of metrology metric(s) on recipe parameter(s), analyzing the derived dependency, determining a metrology recipe according to the analysis, and conducting metrology measurement(s) according to the determined recipe. The dependency may be analyzed in form of a landscape such as a sensitivity landscape in which regions of low sensitivity and/or points or contours of low or zero inaccuracy are detected, analytically, numerically or experimentally, and used to configure parameters of measurement, hardware and targets to achieve high measurement accuracy. Process variation is analyzed in terms of its effects on the sensitivity landscape, and these effects are used to characterize the process variation further, to optimize the measurements and make the metrology both more robust to inaccuracy sources and more flexible with respect to different targets on the wafer and available measurement conditions.