This paper reports on boron and arsenic redistribution during the TiSi2 self-aligned silicide process as applied to shallow (<0.2 ~tm) junctions. Dopant loss was seen to occur through evaporation from the silicide surface, segregation into the Ti-rich outer layer which is subsequently removed, and diffusion into the silicide layer. Depending on the silicide and junction annealing temperatures, up to 99% of the implanted dopant dose can be lost via these three mechanisms. Dopant loss is particularly acute when the silicide is formed concurrently with recrystallization/annealing of the junction implant, before the dopant diffuses into the silicon. Germanium preamorphization, to eliminate channeling of the ion implanted dopants, further aggravates the loss of boron but has a slightly beneficial effect with arsenic. Oxide capping of the silicide before annealing reduces dopant evaporation and increases the dopant concentration at the silicide contact, but at the expense of increased junction depth. ) unless CC License in place (see abstract). ecsdl.org/site/terms_use address. Redistribution subject to ECS terms of use (see 155.69.4.4 Downloaded on 2015-06-04 to IP

Different aspects of the reaction between Ti and single‐crystal (100) Si are reported when the reaction is thermally activated by rapid thermal annealing. The sheet resistance variation of the TiSix film was measured for a wide range of annealing temperatures, from 400° to 1100°C, in two different atmospheres of high purity Ar and N2 gas, for bare and As‐implanted Si. At temperatures ⩾800°C, the only silicide formed was TiSi2 . Both cross‐sectional TEM and RBS results indicate that the amount of titanium consumed in the reaction depends strongly on the atmosphere used for silicidation. Argon makes possible the reaction of almost the whole Ti layer with silicon, while nitrogen leaves an unreacted layer of about 15 nm, independent of the thickness of the original film. A two‐step anneal is necessary to eliminate bridging in a patterned oxide structure. No difference was observed in the roughness of the external surface whether the samples were annealed in Ar or in N2 . No significant enhanced diffusion of As into Si was observed near the normalSi/TiSi2 interface after RTA in an Ar ambient, while a redistribution of As into the silicide takes place. The As piles up at the TiSi2 surface, from which it evaporates at high temperatures. Results of furnace‐annealed samples in N2 atmosphere show a pile‐up of As at the normalSi/TiSi2 and the TiSi2/normalambient atmosphere interfaces, where it evaporates from the latter, at high temperatures.