果冻传媒

To realize further improvements in lithographic approaches, it is important to understand how the pattern formation in EUV lithography works and what new physical phenomena play a role.

To achieve this goal, developed a kinetic Monte Carlo engine that aims to include all the physical information of the photoresist material and the processes, fully creating a digital twin of the EUV pattern formation process. He developed the framework for such a model.

New insights

In the current model, Fernandez Miguez included all quantitative information that is currently available and used placeholder parameters and processes when such information is not yet known.

The simulation model is expected to provide a qualitatively correct view, which can be further refined based on future studies. With the model, he obtained a first set of results and new insights. The model can be used to analyze results of experiments that lead to refinement of our current understanding of the EUV patterning process. The resulting new insights that can then be added to the simulation tool.

High-Performance Computing Cluster

In the thesis, Fernandez Miguez outlines how to run the simulation tool on a High-Performance Computing Cluster.

Subsequently, he presents the results of various types of simulations that help understanding the pattern formation process, like studying the distribution of damages around a point of absorption, simulating electron yield experiments, and studying the amount of material remaining after development.

Metal-oxocore photoresists

All simulations apply to metal-oxocore photoresists. The simulation results for these materials show a fair agreement with published experimental results.

To gather further insights, he performed point-of-absorption simulations on a different material, poly(methyl methacrylate) (PMMA), and compared the results obtained with those obtained for metal oxide resists. In addition, he also varied the rates of the elastic scattering and degradation processes to study the sensitivity of the result to those (less well known) parameters.

The Fireworks Model

Within phenomenological models that describe the patterning process, it is often assumed that the clouds of damaged photoresist molecules that are formed after the absorption of a photon are centered around the point of absorption.

However, Fernandez Miguez found that the KMC simulation results cannot be understood well with such a model. This is explained from a statistical analysis of the locations of the individual damage clouds, which shows that their centers are in general displaced from the point of absorption.

As a result, he developed a new way to describe the stochastic distributions of the location and shape of the damage cloud, the so-called Fireworks Model.

This model allowed him to better understand the KMC results, and to obtain a good agreement between this model and the full KMC results. In the future, the Fireworks Model might be used to replace the kinetic Monte Carlo simulations for some applications and thus save computation time.

To address remaining discrepancies between the experimental results and the full KMC model, he also explored possible additions to the code, like including rotations of the molecules due to thermal effects during an often-used photoresist post-illumination baking step or using more a refined method for modelling the inelastic scattering processes that lead to the cascade of electrons and holes that induce the chemical changes that lead to the finally formed pattern.