Effects of Initial Plasma Properties on Plasma Recovery in Plasma Source Ion Implantation

Summary form only given. An expanded plasma sheath in front of a negatively biased target immersed in low-pressure plasmas recovers to its initial state after turning off the negative bias. This recovery process has been assumed to be governed by the ambipolar diffusion model (Wood, 1993) or Bohm sheath collapse model (En et al., 1995). However, the detailed explanations for plasma recovery process are still unsatisfactory since those models are based on the unrealistic assumption of a step-like density profile at sheath edge. We investigated the plasma recovery process occurring during the pulse-off time both experimentally and numerically. Experiments were carried out with low-pressure argon plasmas generated by an rf inductive heating in the cylindrical quartz vacuum chamber. The dynamic motion of plasma sheath and the propagation of rarefaction wave into the plasma during and after the pulse bias were measured with a positively-biased electrostatic probe. Also, initial plasma properties such as plasma density, electron temperature, and ion flow velocity were measured using various probe methods. For the numerical consideration, an equivalent circuit model was developed to model the pulse system used in the experiment. In the circuit model, the response of plasma sheath was calculated using one-dimensional two-fluid model and inserted into the circuit model as a load element of the entire pulse system. From the experimental and numerical results, it was found that initial plasma properties in front of a target were important in plasma recovery process. Especially, the initial ion flow velocity always existing around an object immersed in plasmas by a formation of presheath or a diffusion of bulk plasma was found to play a crucial role for plasma recovery in low-pressure plasmas. These results indicate that a complete set of initial plasma properties should be considered in the model to describe plasma recovery process properly.