Titanium and nickel silicide formation after Q-switched laser and multiscanning electron beam irradiation

The use of Q-switched ruby laser and multiscanning electron-beam annealing to produce the reaction of thin Ti and Ni films deposited onto silicon single crystals has been studied. Rutherford Backscattering (RBS), 16O(d, p)17O* nuclear reaction, scanning electron microscopy (SEM) observation, and x-ray diffraction were used to characterize the reacted layers. It was found that laser annealing produces a reaction only at the metal-semiconductor interface: the reacted layers are not uniform in composition and more similar to a mixture than to a well-defined phase. On the contrary, the silicide layers produced by multiscanning e beam result from the solid-state reaction of the whole metal film and have a layered structure with well-defined phase composition and sharp interfaces both between the silicide phases and the underlying semiconductor in Ti/Si system. It was observed that the TiSi2 growth mechanism during e irradiation cannot be explained with the parabolic ''diffusion controlled'' mechanism operating in the standard furnace annealing. All our observation seems to indicate that the growth mechanism is a ''nucleation controlled'' process, in which the growth speed of the disilicide is limited by the speed of ejection of oxygen from a TiSi2 layer. In a Ni/Si system, only the NiSi phase could be obtained as a very uniform layer after the e-beam irradiation; the impossibility of obtaining the Ni2Si phase indicates that, in these conditions, the ''first-phase nucleation law'' is no more valid