Abstract
The sacrificial layer removal is one of the important processing steps in surface micromachining. There are a number of reports on the study of phosphosilicate-glass (PSG) etching in microchannels with HF-based solution. However, there are no consistent conclusive results about the etching front shapes, advancing speed, and factors contributing to the etching behavior. We report here our recent results on the in-situ observation of etching front during PSG removing in HF solution, emphasizing on factors such as property of structure layer, liquid/solid contact property, shape and size of the etching window, and bubble formation. Two-dimensional etching fronts are also investigated. Based on this observation, a new technique to reduce the sacrificial layer removing time is proposed. The fabrication processes of the test structure are described in the following. After the standard wafer cleaning, APCVD PSG layer was deposited at 400oC with a gas flow of 90 sccm silane, 60 sccm oxygen, and 9 sccm phosphine, followed by an annealing in N2 at 950oC for 30 minutes. The PSG layer was then patterned and etched. A transparent layer of 0.3μm LPCVD nitride or 0.2μm LPCVD poly-si was deposited; etching windows were opened by RIE. Finally, measuring scales were created by a shallow RIE etching on the surface. The phenomena of sacrificial layer wet etching were monitored and recorded by a video system fitted with microscope. The shapes of etching fronts change with time and display quite differently for samples with nitride or poly-si as the channel covering material. It was reported that notch-shaped etching front occurs only in undoped sacrificial oxide films, and was attributed to residual stress in the sidewalls of the glass films.We propose that the surface tension force of the etching liquid along the sidewalls pulls the etching front forward along it faster relative to the center, resulting in the notch-shape. The surface tension force dominates especially in thin sacrificial layer and hydrophilic surface, causing faster etching rate and smaller etching front angle.Capitalized on this in-situ observation of etching study, we propose a new strategy to reduce sacrificial layer removing time by depositing a thin layer of hydrophilic material such as nitride buffered between sacrificial PSG and structural material to be released. This idea is proved to be effective as observed. It is also found that bubbles always formed at the etching front, and ejected out via the etching window from time to time. Etching rate increases once the bubble formed. These bubbles complicate the analysis of etching phenomena.