The main objective of this paper is a series of experiments designed to elucidate the origin of large fluctuations in the growth stress of aluminium rich TixAly-alloy films found in earlier experiments. The internal stress of the various alloy films was measured in situ under UHV-conditions during as well as after deposition, using a cantilever beam technique. In parallel experiments the microstructure of the different films was investigated by TEM.

The experiments were started assuming that the fluctuations in the stress vs. thickness curve of aluminium rich TixAly-alloy films indicate unstable films. In order to test this hypothesis, we have performed annealing experiments with alloy films of which the aluminium content was varied from 50% to 75%. The films were formed by evaporating the alloy components from two separate evaporation sources onto alumina substrate films at 27°C substrate temperature.

The growth stress of the Ti50Al50-alloy film exhibits a large tensile stress, which increases linearly with film thickness and a negligible stress change after deposition. Increasing the aluminium content lowers the tensile stress built up and the average stress in the Ti25Al75-alloy film is compressive but stress changes after deposition are still very small. This experimental evidence is interpreted to indicate that layer by layer growth is the representative growth mode for these films. The origin of the growth stress of these films is assumed to be the grain boundary relaxation mechanism proposed by R.W. Hoffman. The differences in the growth stress in films with varying aluminium content are assumed to originate from changes in the magnitude of the attractive forces in the grain boundaries, defects etc‥

Annealing these films gives rise to irreversible tensile stress changes. These stress changes in the Ti25Al75-alloy film are about four times that measured during annealing of the Ti50Al50-alloy film. A pronounced tensile stress feature in the temperature range between 230°C and 260°C is interpreted to be due to segregation of aluminium in the alloy film and to the alloy film surface. This aluminium segregation is negligible in alloy films with an aluminum content lower than 67%. With the experimental evidence presented in this paper we are able to show that aluminium segregation is the reason for the large fluctuations in the stress vs. thickness curves of aluminium rich alloy films when the two alloy components are evaporated simultaneously from two separate evaporation sources.