A review of thin-film aluminide formation

Abstract

With the continuing drive toward greater device densities and finer dimensions in the microelectronics industry, the required proprerties of the metallization layers have become increasingly stringent. Transition metal aluminides are of great interest due to the use of transition metals as diffusion barriers, to suppress hillock formation, and to increase electromigration resistance. This review seeks to collect all the relevant work on transition metal aluminide formation: initial phase formation, temperature of formation, uniformity of growth, and microstructure of the phase formation. Where data are available, the subsequent phases formed, the growth kinetics, and dominant moving species are included. The Ni-Al system is discussed in detail as it is one of the best understood systems and exhibits typical behavior. The Ti-Al, Al(Cu) system and TiW diffusion barriers are also discussed individually due to their technological importance. As has been found for silicide formation, there are some general patterns of behavior with aluminide formation. In general, the initial aluminide phases to grow are the most Al-rich phases: Co 2 Al 9 , Cr 2 Al 13 , HfAl 3 , MoAl 12 , NbAl 3 , NiAl 3 , TaAl 3 , TiAl 3 , WAl 12 , and ZrAl 3 . There are exceptions, Pd 2 Al 3 , Pt 2 Al 3 , and VAl 3 are the initially growing phases, but are not the most Al-rich phases. Where marker experiments were performed, Al has been identified as the dominant diffusing species during the growth of the initial phase. It has been suggested that the Al-rich initial phase results from the greater supply of Al (relative to transition metal) to the growing interface with exceptions caused by complex (and hence difficult to nucleate) phases. The initial reaction temperatures ranged from 225–250°C for Pd 2 Al 3 and Pt 2 Al 3 to 500–525°C for WAl 12 formation. In general, the phase formation is planar, though impurities and grain sizes can modify this. For metals forming high-melting-point compounds, the reaction is more likely to be non-uniform. Though generalized rules have been proposed, there are still many open questions. Our understanding of aluminide formation lags behind that of silicides.