| 講演要旨： |
In order to meet the scaling trends for the sub-45 nm next technology nodes, it has become necessary to engineer very precisely the doping profiles in the devices. The main characteristics to achieve are very highly doped, very shallow (10-20 nm) junctions with very well controlled lateral extension and abruptness. These are fabricated using advanced doping techniques such (cocktail) ion implantation, plasma immersion, vapor phase deposition,… and ultra fast anneal methods (LTA, flash, RTA, SPER ) , msec anneal for which predictive modeling is still very immature and accurate experimental carrier profile determination is required. As in several of these applications, high depth resolution (<1 nm) needs to be combined with high sensitivity (< ppm) and quantification accuracy (<3%), Secondary Ion Mass spectrometry (SIMS) is quite often the only technique able to fulfill these needs. The success of SIMS appears somewhat surprising as it based on a very crude concept i.e using an ion beam sputtering process to obtain depth information and relying on spontaneous ionization to produce the secondary ions. Nevertheless SIMS has become a very well established technique for dopant profiles in semiconductor materials (in particular Si) with very high sensitivity, depth resolution and quantification accuracy. We will outline these beneficial aspects of SIMS by showing examples of cases whereby films differing only by 0.5 nm in thickness clearly could be identified, and small compositional variations in composition of SiGe-films can be probed. Unfortunately many cases exist where SIMS is limited I performance as well.
As will be shown in this presentation, many of the SIMS properties and limitations can be understood by a careful analysis of the interaction between the primary ion and the substrate focusing on the temporal changes in (matrix) sputter yields, the retention of the primary species and defect-induced migration of constituents. For instance the main reason for the success of SIMS is the enhanced ionization probability (leading to high sensitivity) due to use of the reactive primary ions such as oxygen and cesium, coupled with very low primary beam energy (down to 150 eV). Unfortunately the use of an energetic oxygen and Cs primary ion implies that the target is heavily modified during the analysis which does lead to some transient effects as well as (in a number of cases) unwanted distortions due to ion beam mixing, surface topography (ripples), loss in depth resolution by segregation, changes in ionization probability at interfaces in multilayer systems due to a different retention of the primary species,…. This necessitates the need for very advanced (extremely low energy) protocols as well as the concurrent use of alternative methods like MEIS, H-RBS and H-ERD and emerging concepts like Atomprobe analysis.
|
