(Invited) Chemical Approaches to Etch and Pattern Copper Films

Abstract

Microelectronic device fabrication has traditionally been performed by selective removal or etching of material from deposited films to create patterns. Plasma etching has been the mainstay of this process since the early 1970s. The introduction of copper (Cu) metallization (~1997) by IBM resulted in the development of additive approaches (e.g., damascene process) due to the inability to plasma etch Cu. Despite attempts to plasma etch Cu as early as 1980, the only viable plasma-based methods in this time frame were inert gas sputtering, where selectivity to underlying layers and low etch rate were problematic, or chlorinated vapors that required elevated temperatures (>200 C) or significant ion bombardment energies to volatilize Cu chlorides. Both approaches required a hard mask rather than photoresist layers to ensure mask stability and precise pattern transfer. Due to the inability to form high volatility Cu halides that could be removed readily by plasma processes, other non-plasma or partial plasma approaches to Cu etching at lower temperatures were investigated. Addition of a UV lamp to facilitate excitation and desorption of volatile Cu chlorides in plasma etching was invoked. Formation and volatilization of copper by oxidation followed by complexation with organic ligands at lower (~150 C) temperatures was first reported in the mid-1990s, but this processes generated isotropic etch patterns; more recent approaches have reported anisotropy. Plasma chlorination followed by a liquid etch to remove the chlorinated Cu was also reported; due to the directionality of the plasma chlorination step, anisotropic profiles resulted. Hydrogen-based etching of Cu films was reported in 2010, initially with a two-step process consisting of plasma chlorination followed by a hydrogen plasma step to remove volatile products. This effort subsequently demonstrated that etching by pure hydrogen in a one-step process was possible albeit with low etch rates; etch rate increases were observed with vapor additives. Pure hydrogen plasmas etched photoresist layers, thereby requiring hard masks, but additives allowed employment of photoresist as a mask layer. Anisotropic etch profiles from these hydrogen-based processes were possible, in part due to additives protecting sidewalls and photoresist from etching and to ion bombardment during etching.This presentation will describe the development of Cu etching and patterning methods beginning in the 1970s. Mechanistic considerations of the various processes reported will be discussed and the current needs and limitations of these approaches indicated.