Gallium arsenide backside through chip via hole integration using inductively coupled plasma etching.

摘要

GaAs backside via hole etching with Cl₂ based chemistry and subsequent process integration was studied using Inductively Coupled Plasma (ICP) etching system. Etching characteristic was examined by varying four majors process parameters such as pressure, gas composition, bias RF power and source RF power. Enhanced etching rate compared to Reactive Ion Etching was observed at pressure range less than 14 mT and it is attributed to the higher ion density in the ICP chamber. The amount of reactive Cl neutrals and ions is the controlling factor for the etching mechanism. Etching rate does not depend on the ion bombarding energy, suggesting that by-product desorption may not be the rate limiting step of etching process. Vertical sidewall profile was observed and it is attributed to higher density ion and lower process condition. Sidewall morphology was strongly dependent on the Cl₂ percentage. More chemically driven etching conditions produce more pitted surface. Surprisingly, vertical grass was observed at the bottom of via hole at certain process and thinning condition. Grass formation was found to strongly depend on the surface condition affected by the thinning process. Defects induced during thinning process would introduce grass and enough chemical polishing after mechanical thinning would prevent forming grass. At more physically driven etching conditions, grass formation is less favorable. A grass forming model was suggested. An initial grass forming stage, grass may be formed due to selective etching around defect sites. Faster lateral etching around defect sites generated isolated island structures. As etching proceeds, isolated area becomes smaller and incubates tip of grass. During etching, initial grass grows over time and it is attributed to the field effect. When tip of grass reaches a certain critical size, it may produce high field to deflect the arriving ions. Grass grows since etching is less effective at the tip compared to flat base area. After subsequent metallization, DC via hole resistance was evaluated. It is estimated less than 1 ohm, which has negligible effect on the circuit performance.