Search

Silicon dioxide is the most important gate oxide dielectric material, and its surface planarization is an important factor in the continuous shrinking of the size of integrated circuits. Developing a novel chemical mechanical polishing auxiliary technique to achieve efficient polishing of oxide wafers is a significant challenge. In this study, a novel in-situ micro-nano bubbles auxiliary method, which used a nano-hydrophobic film to prepare a gas supersaturated slurry that can generate instant in-situ micro-nano bubbles during polishing, was developed for oxide wafers. Compared with conventional slurries, this auxiliary method increased the material removal rate by about 30% and ensured the low surface roughness of the wafer of 0.14 nm. The existence of micro-nano bubbles and the multi-size distribution and zeta potential of the micro-nano bubbles were measured by optical microscope and dynamic light-scattering methods. Then, to investigate the auxiliary mechanism of this in-situ micro-nano bubbles, different analytical methods, such as oxide thickness measurement system, scanning electron microscopy, Fourier transform infrared spectroscopy, and atomic force microscopy, were used to characterize the sample. The results indicated that during the polishing process, the micro-nano bubbles generated in situ have a good cleaning effect. In addition, the free radicals generated after in-situ micro-nano bubbles burst simultaneously modified the surfaces of the wafer and the pad. The improvement of the adsorption between the wafer and abrasives promoted the chemical reaction between them. After conditioning, the rougher and cleaner pad had a higher mechanical effect. Most importantly, this in-situ micro-nano bubbles auxiliary method has the potential to be applicable to all slurries.

The material loss caused by bubble collapse during the micro-nano bubbles auxiliary chemical mechanical polishing (CMP) process cannot be ignored. In this study, the material removal mechanism of cavitation in the polishing process was investigated in detail. Based on the mixed lubrication or thin film lubrication, bubble—wafer plastic deformation, spherical indentation theory, Johnson—Cook (J—C) constitutive model, and the assumption of periodic distribution of pad asperities, a new model suitable for micro-nano bubble auxiliary material removal in CMP was developed. The model integrates many parameters, including the reactant concentration, wafer hardness, polishing pad roughness, strain hardening, strain rate, micro-jet radius, and bubble radius. The model reflects the influence of active bubbles on material removal. A new and simple chemical reaction method was used to form a controllable number of micro-nano bubbles during the polishing process to assist in polishing silicon oxide wafers. The experimental results show that micro-nano bubbles can greatly increase the material removal rate (MRR) by about 400% and result in a lower surface roughness of 0.17 nm. The experimental results are consistent with the established model. In the process of verifying the model, a better understanding of the material removal mechanism involved in micro-nano bubbles in CMP was obtained.

In this paper, tribological effects on the material removal rate (MRR) are investigated using two types of slurry abrasives (i.e., ceria and silica) and three types of pads. From a micro to macro viewpoint, the physico-chemical interaction between the pad-wafer interface and abrasive was studied based on multi-asperity contact. First, the relative contact pressure in high or low proportions among all asperities was calculated using a contact mechanics model. The results indicate that for the commercial slurry used in the semiconductor industry, relatively high contact pressure due to the asperities can be transferred to the ceria abrasives, and only a low contact pressure can be supported by these silica abrasives. Furthermore, it was confirmed that the two different slurry abrasives showed opposite behaviors according to the real contact area of the pad asperities. That is, as the pad texture changed from ‘Desired pad’ to ‘Smoothed pad’, the MRR of the ceria increased by 47%, while the MRR of the silica decreased 59%. Lastly, the Stribeck curve, which is suitable for the CMP process, was re-analyzed to clarify the abrasion mechanism under the pad asperity scale. The results showed that two-body abrasion may be maintained for the ceria CMP, and transition is possible from two-body to three-body abrasion at the silica CMP as the contact area increases. The present paper is expected to clarify the tribological mechanism of the oxide CMP and can provide further insight into optimizing and selecting consumables such as conditioner, pad, and slurry.