Paper PS2-WeM5
Quantitative Characterization of Ions and Si Surface Interactions - Estimation of Plasma-Induced Defect Generation Probability

Wednesday, October 17, 2007, 9:20 am, Room 607

The plasma damage induced by ion bombardment has become one of crucial issues from the viewpoints of the physical thickness in scaled devices. Quantitative analysis of the plasma-induced defects is requisite for understanding the mechanism and realizing high performance devices. In this article, ions and Si surface interactions during plasma processing are quantitatively analyzed by novel techniques, providing the defect (charge trapping site) density and defect generation probabilities. Samples were exposed to two different plasma sources, DC and ECR with various biasing, gas mixtures and process time. N-type Si wafers with the low resistivity of 0.02 Ωcm were mounted on the stage and exposed to plasma sources. Combined with plasma diagnostics, two optical analyses, spectroscopic ellipsometry and photoreflectance (a modulation spectroscopy) were conducted to identify the damaged layer thickness with the 4-layer (air/layer-1/layer-2/Si substrate) model, the mechanical strain developing in the vicinity of the surface determined by the Si transition energy change, and the defect density (charge trapping site) by a novel method based on the surface potential change calculation. Also the surface layers were evaluated by a resistivity measurement. We have observed the characteristic structure change in the damaged layers with the relaxed mechanical strain (approximately 0.1 %) and the charge trap site generation with significant densities of 10¹² cm^-2, along with plasma exposure time. Hence we have finally determined the defect generation probabilities per an impinging ion as 10^-3 - 10^-5 s^-1 in the present ion energy and plasma density ranges. The difference is attributed primarily to that in the measured bias voltage (-300 V for DC and -50 V for ECR) dominating the energy of ions accelerated in the sheath, although the ion energy distribution function has to be taken into account for further discussion. This ion energy effect is confirmed from the difference in damage-layer thickness by ellipsometry as well as from the etching simulation. The calculated defect densities in the damaged layer are considered to affect device performances in terms of the increase in power consumption by plasma-induced junction leakage. The obtained defect generation probability is a key parameter for understanding the mechanism of ions and Si surface reactions as well as plasma process designs.