Diffusion of boron in Mo silicide films

Abstract

The diffusion of boron in MoSi2 films has been studied. The diffusion is determined by a combination of two factors: the low diffusivity in the bulk of grains and the much higher diffusivity in the grain boundaries. The diffusivity in the bulk of grains was 1×10−14 cm2/s and 8×10−13 cm2/s for furnace annealing at 950 °C and halogen lamp annealing at 1060 °C, respectively. The diffusivity in the grain boundaries was over two orders of magnitude higher than that in the bulk of grains. High-temperature annealing of MoSi2 films reduce the solubility and diffusivity of boron in the films because of grain growth.

Ultra Shallow Junction Formation Using Diffusion from Silicides: II . Diffusion in Silicides and Evaporation

“…Here we assume that the evaporation rate is proportional to the surface (soluble) dopant concentration, which is in turn related to the bulk dopant concentration by Eq. [2]. The bound dopant, Q0, does not evaporate.…”

“…The silicide thickness is d, C~t is the average dopant concentration in the silicide bulk after an annealing which allows a thorough diffusion in the silicide, and Qs~ the amount diffused into the Si substrate. We assume, further, that the surface dopant exists within a depth, defined as k, and has an average concentration of Csu~, so that C~ = (1 + ks)C~ [2] and Qss = kksC~, [3] where ks is the surface segregation constant. When Qs~ is negligible compared to QT (Qsi is typically a few percent of QT), QT can be expressed as…”

“…The overall expression for the average dopant concentration in the bulk of the silicide for the case of rapid dopant diffusion can be written according to Eq. [2], [6], [7], and Because of the presence of dopant at the silicide surface and the subsequent dopant loss, the dopant concentration at the silicide/Si interface is much less than the implant dose divided by the siticide thickness. The interface concentration is also lower when diffusion through the silicide is slow.…”

“…If diffusion through the silicide becomes rate limiting, the evaporation rate should be determined by the diffusional flux which can be written as F = D-- [10] CoSi2, which is far more than the amount evaporated. [2], [6], [7], and Because of the presence of dopant at the silicide surface and the subsequent dopant loss, the dopant concentration at the silicide/Si interface is much less than the implant dose divided by the siticide thickness. Arsenic, on the other hand, is almost immobile by bulk diffusion at 900~ and the flux-time product caused by the grain boundary diffusion is estimated to be 4 • 1013 cm -2 from the interface pile-up in Fig.…”

“…They are the evaporation of the dopant from the silicide, the segregation of dopant at the silicide surface, and the slow diffusion of dopant within the silicide. However, grain-boundary diffusion that is usually dominant over bulk diffusion in silicides (2,3) may bring the dopants to the silicidesilicon interface to allow further diffusion into silicon.…”