On-product eBeam overlay topic/category AM: Advanced metrology

The semiconductor industry is facing constant overlay challenge posed by tighter overlay error budget and the increase of multi-patterning layers in 14nm node and beyond. While the allowed total measurement uncertainty (TMU) for Overlay metrology is already extremely low due to the ever tightening overlay budget, discrepancies between optical overlay results and electrical test data raise the inaccuracy concerns which are largely induced by the imperfections and asymmetry of the overlay marks located in scribe line. The process impact in the representative overlay targets is also a poor representation of real indie device, which further exaggerate the inaccuracies. Current optical overlay tools have specification for sensitivity to On-Product Overlay (OPO), which is a representation of measurement accuracy but the lack of reference to validate said measurement outcome calls for an indirect method to confirm real overlay. Improvement in overlay control is crucial to accelerate yield learning in high-volume manufacturing (HVM) fabs. In this work, inline eBeam-based Overlay metrology via Scanning electron microscope (SEM) is developed as reference metrology for inline optical overlay to characterize in-die, across-device, on-product overlay behavior at multi-patterning post-etch steps in order to address challenges present in OPO. This method incorporates high resolution imaging, physical measurement, and a Center of Gravity (Cog) algorithm to archive sub-nm precision while indirectly obtaining true overlay. Random Located eBeam Overlay (RLEO) scheme is designed to ensure process compatibility, by selecting small sets of 2D structures spread across the field. These small regions of interest are then measured with much higher resolution under eBeam rather than visible light, currently utilized by the state of the art overlay tools. Another On-product Overlay qualification strategy is demonstrated and by combining the advantages of eBeam and Optical Overlay to balance the precision, accuracy, and productivity, thus improving on-product optical overlay performance. Methods described above, when applied to optical overlay marks, can also assist in singling out the best performing overlay targets (i.e. blossom (BLO), diffraction based overlay (DBO), advanced imaging metrology (AIM), or scatterometry overly (SCOL)) and measurement conditions for a given layer via optical vs eBeam overlay correlation (R
²
). Future work will investigate advanced applications of RLEO method to address challenges presented by four components of the overlay budget calculation; mask, scanner, process, and inline optical overlay, also offers techniques to isolate and control these hidden overlay factors unearthed by eBeam Overlay (EBO).