Hardware and process dependence of electron shading damage in a high density plasma oxide etch tool

Plasma process induced damage is a continuing concern in device manufacture, and has been correlated with reduced product yield and increased process variability. There are two primary gate oxide damage mechanisms during plasma processing: (1) charging due to nonuniform plasmas, and (2) charging due to aspect ratio dependent electron shading. For properly designed high density plasma (HDP) sources, charging due to plasma nonuniformity is minimal, while charging due to electron shading can be severe depending on process conditions. In order to design low damage HDP etch processes, it is necessary to characterize the dependence of electron shading damage on process parameters. In this study, statistical design of experiments (DOE) was used to characterize electron shading damage in a Lam 9100 HDP oxide etch tool. Unlike previous work using one-at-a-time experimental methods, our approach permits observation of the charging damage response over a process window in which source RF, bias RF, pressure, argon flow, and chamber gap heights were simultaneously varied. Regression methods were used to model the charging damage response, and a good fit to the experimental data was obtained. The modeled damage response is a complex function of the input variables, including significant interaction and quadratic terms. In general, hardware and process changes that reduce electron density or temperature were found to reduce damage, in agreement with previous reports. By comparing plasma charging and etch rate response surfaces, we identify process and hardware conditions that result in a low damage, high throughput via etch process