Retarding Field Energy Analyzer Optimization And Space Charge Effects

Abstract

In industrial plasma processing for semiconductor fabrication, it is important to understand the characteristic properties of the plasma. The ion energy distribution function (IEDf) is one such property. The IEDf has a direct impact on process outcomes. Retarding field energy analyzers (RFEAs) have been used extensively to obtain IV curves for typical process conditions. These IV curves are often analyzed using an ideal RFEA model when calculating IEDfs. However, several factors can cause a measured IV curve to deviate from an ideal one, especially for higher grid voltages and plasma densities. Three factors to consider are voltage dip within the grid holes 1, electric field non-uniformity due to the probe geometry, and space charge build up between the grids 2. This last factor is a result of high ion flux or larger grid separation caused by high grid voltages. In this study, electrostatic simulations (EM Works) and particle-in-cell (PIC) simulations (XPDP1) were used to parametrize the impact of these factors on IV curves. Electrostatic simulation results led to a RFEA geometric design that minimized vertical electric field variations. The field uniformity was improved by 25x across the sensor area after optimization. In addition, the overestimation of the IEDf due to voltage dip within the grid holes was quantified. A shift of 2–2.5 eV was observed. Computed IEDfs were reconstructed from PIC generated IV curves using regularization methods. These simulations demonstrate how IV curves vary due to space charge build up. Space charge only affected lower energy ions. The specific energy is dependent on the grid separation distance. In this case, the IV curve begins to fall off at a lower voltage with a more gradual slope causing a larger low energy tail. This non-ideality in the curve can be corrected by limiting the flux of ions into the probe or through corrections during the regularization reconstruction. By taking these factors into account, it is possible to optimize a RFEA and modify measured IV curves to better represent an ideal curve.