Investigation of reconstruction uncertainties and solution multiplicity in high-harmonics generation EUV scatterometry

As the world around us becomes evermore technologically integrated and devices become smaller, faster, and more efficient, so too do our favourite gadgets become more complex. Inside each one of our devices exists a complicated, multi-dimensional maze of components and materials that ultimately work together to bring us the technological wonders of today. These crucial structures are made via a complex flow of lithographic processes in which layers of material are patterned and deposited, removed and polished, implanted and etched away at the scale of only a few nanometres. The precise manipulation and creation of nanoscale structures is what enables new computing powers, more accurate sensors, and faster microelectronics.

These new technologies only work as long as the underlaying structures are near-perfectly made and the complexity and extreme small size of these components provides significant challenges in their characterization and imaging. Recently, a new characterization technique using extreme ultraviolet (EUV) light (with a wavelength in range ~10 - 120 nm) has emerged that has the potential to non-destructively characterize nanostructures with complex inner 3D geometries. This technique, EUV scatterometry, rely on the use of computer algorithms to process patterns of scattered EUV light from a structure or device in order to identify multiple geometric parameters with sub-nanometre accuracy. This technique shows great promise for non-destructive characterization of nanostructures; however, we are only beginning to scratch the surface of their power for characterizing actual microelectronic device structures.

In metrology it is crucial to know how accurate and precise the technique is. Therefore, quantification of reconstruction uncertainties is of utmost importance. The task gets more complex when solution is not unique and a number of model parameters are cross-correlated.

In this project, you will work closely with a diverse team of scientists and experts within imec’s materials characterization laboratory and advanced patterning department to help usher in a new era of EUV-based characterization of semiconductor materials. In your work, you will apply advanced statistical methods like Markov Chain Monte-Carlo to probe and map multidimensional error-parameter spaces in order to identify hidden correlations, multiple solutions and underlying reconstruction uncertainties while evaluating relevant semiconductor structures and devices. The algorithms you develop will push the boundaries of the state-of-the-art in EUV characterization, defining resolution and enabling exotic paradigms such as buried-layer/depth-resolved 3D geometry characterization. These algorithms will then be applied on real-world semiconductor structures to help enable a new paradigm in characterization that supports development of evermore advanced devices.