Chemical and particulate contamination removal from patterned and nonpatterned semiconductor surfaces using oscillating flow

As semiconductor device feature size continues to shrink, the removal of nano-scale particles and cleaning of deep submicron trenches presents tremendous challenges in semiconductor manufacturing. The absence of tools to measure contamination in submicron trench makes it difficult to conduct trench cleaning experiments. Physical modeling of the submicron trench cleaning process presents the most effective approach to understand and study the process.; In this work, based on the numerical study of blanket wafer cleaning and the modeling of pulsating flow over large periodic cavities, cleaning of submicron trenches is investigated using numerical modeling. Numerical predictions were verified experimentally. Compared to steady flow cleaning, cleaning using oscillating flow exhibits a significant advantage on patterned (structured) surface cleaning. The results indicate that the cleaning effectiveness is a function of the flow velocity, frequency and cavity width which are represented by Strouhal number. The optimum Strouhal number for trench cleaning was 0.133 for trenches 100 microns and larger. This is also true for deep trenches with high aspect ratio (up to 5:1). However, for smaller trenches, due to the very short diffusion time for microscale trenches, higher frequency consistently gives higher cleaning efficiency. Cleaning using normal flow to the wafer surface shows orders of magnitudes higher cleaning efficiency than parallel flow for steady and oscillating flow cleaning.; Nano-scale particle removal mechanisms by non-contact cleaning technique (such as megasonic cleaning) and contact cleaning technique (such as brush cleaning) were also investigated. Megasonic (high frequency ultrasonic) cleaning is found to be a promising technique for the removal of nano-size particles down to 10 nm. Megasonic waves induced acoustic streaming and thin acoustic boundary layer are essential to the removal of submicron and nano-size particles. As the frequency increases, the acoustic boundary layer thickness decreases, and the streaming velocity increases thereby increasing the drag force and consequently particle removal. Properly utilized cleaning solutions can induce larger repulsive electrostatic double layer force which decreases the total adhesion force. In brush cleaning, direct contact between the particle and the brush is essential to the removal of submicron particles. If the brush does not contact the particles a 100 nm particle can hardly be removed when the brush is 1 micron above the particle. Proposed mechanism results were experimentally verified for megasonic and brush cleaning for particles larger than 100 nm.