Investigation on multilayered chemical vapor deposited Ti/TiN films as the diffusion barriers in Cu and Al metallization J.Detailed analytical transmission electron microscopy investigations were performed on a wellknown diffusion barrier system for very-large-scale integration metallization. It will be demonstrated that interfacial reactions are of great importance for the barrier mechanism. Both Ti and TiN act as diffusion barrier for the semiconductor and the metallization, respectively. For an aluminum-based metallization, TiN has a "spongelike" function due to its ability to absorb several amounts of aluminum at elevated temperatures and therefore inhibits diffusion towards the substrate. Ti acts for silicon as a compound forming barrier according to Nicolet's classification [in

No abstract

Defects in reactively sputtered titanium nitride diffusion barriers in submicron devices were investigated. Four different failure mechanisms could be extracted, two originating from the production of the layers, two related to foregoing process steps. The latter are contact hole overetch and topology inside the contact hole, Le., edges, created by deviating isotropic etch ratios of different dielectric layers in a post-treatment after the contact etch. The insufficiencies related to the preparation of the TiN films are microcracks due to excessive stress incorporated and encroachment caused by a Ti-rich titanium nitride on the contact hole walls, proved by an Auger electron spectroscopy analysis. Accidental accumulation of two or more of these defects in a production line can cause the barrier to fail. The process of contact degradation was investigated by transmission electron microscopy and energy-dispersive x rays. It was found that after aluminum penetration the destruction of the contact area proceeds via a ternary Alx Tiy Si z reaction until aU titanium is consumed, followed by the growth of a pure Al spike and the fatal breakdown of the contact.

Abstract. 2014 ECOSS (Electron Compton Scattering on Solids) is a powerful method in order to obtain the momentum density distribution of conduction electrons. Apart from the Bragg-Compton channel coupling which produces a strong and often irregular background, the main problem in ECOSS has been the low count rate at scattering angles where the impulse approximation is valid. Simulations showed that the Gatan 666 PEELS should be capable of producing spectra of sufficient accuracy within a dwell time of some 100 s, when dark current and flat field corrections are applied. Prelimi [5][6][7][8]. In the past decade the increasing availability of PEELS systems [9] has offered new possibilities for previously unfeasible energy loss experiments such as ECOSS.As yet, the indirect detection of electrons via a YAG scintillator seems to be the method of choice for PEELS [10,11]. As a detector, linear PDAs [10,11,15] or two-dimensional CCDs [12,13,16] [14]. Eventually, for the resolution given above, (0394E = 5 eV, 039403A9 = 9 x 10-8sr) (2) and a scattering angle of 34 mrad at 300 kV, we calculate a partial cross section in the Compton maximum: Acr = / / ~203C3 ~E03B403A9dE d03A9 = 3 x 10-31 M2 /atom. (1) The overall momentum resolution Aq in an experiment is governed not only by the energy resolution given in equation (2)

No abstract