Stimuli-Responsive CeO2 Removal Via Surface Redox Modulation during Shallow Trench Isolation (STI) Post-Chemical Mechanical Planarization (p-CMP)

Abstract

As transistor miniaturization continues to extend Moore’s Law into the future of technology, the demand for high-efficiency device manufacturing processes has increased drastically in recent years. More specifically, a critical step in preparing integrated circuits (ICs) and logic devices is Chemical Mechanical Planarization (CMP), which relies on a delicate balance of chemical and mechanical parameters to achieve angstrom-level surface uniformity and ultimately allows for increased transistor packing density. A sub-area of CMP that has gained significant attention is Shallow Trench Isolation (STI), which involves the isolation of electrically active components by removing the bulk oxide (i.e., tetraethyl orthosilicate (TEOS)) overburden from the deposition process. During polishing, chemical slurries comprised of CeO2 nanoparticles, rate enhancers, selectivity and rheology modifiers, and pH adjusters are utilized to activate the oxide layer and consequently remove material. The STI removal mechanism, known as the chemical tooth model, involves the nucleophilic attack of CeO2 nanoparticles at surface defect states (i.e., oxygen vacancies (Ovacs)) and resultant dative Ce-O-Si bond formation, which makes the surface redox state of CeO2 (i.e., Ce3+/Ce4+) a key driver for oxide-nanoparticle interactions. It is well known that Ce3+ nanoparticles contain a higher concentration of Ovacs, which enhances the material removal rate (MRR). However, the hard adsorption properties of CeO2 have resulted in the demand for a more efficient post-CMP (p-CMP) cleaning process for TEOS that promotes increased CeO2 removal without inducing secondary defects (i.e., scratching, dishing, corrosion, etc.). Current industry-standard methods involve using a polyvinyl alcohol (PVA) brush to transport harsh chemistries to contaminated surfaces under high shear force (SF). Additionally, the use of megasonic energy is an emerging non-contact cleaning mode that relies on generating reactive oxygen species (ROS) (i.e., cavitations) to induce particle removal. Previous work has shown that supramolecular additives improve overall particle removal efficiency (PRE) when implemented in traditional PVA brush cleaning due to their ability to encapsulate and charge-flip CeO2 nanoparticles while reducing SF. As a result, this research investigates a combinatorial approach to PVA brush cleaning utilizing megasonic energy to release Fenton catalysts from polymer capsules, which aid in the production of ROS through H2O2 decomposition. Initial results have shown that, by increasing the concentration of ROS during cleaning, a Ce3+ à Ce4+ transition occurs on the surface of adsorbed CeO2, enhancing overall PRE.