Orgin of
03/01/1999
Origin of "stuffed diffusion barriers" (revised)
I must emphasize here that my letter in the August issue (Letters, p. 18) was in direct response to the paper on W/Ti ("High-throughput W/Ti barrier sequential deposition" June, p. 127). With regard to Bill Westwood`s letter in October on p. 18, I was clearly remiss in not citing the work of Holloway and Nelson, who were the first (to our best knowledge) to report the use of oxygen to improve the diffusion barrier properties of a refractory metal, chromium in this case.
Indeed, I did reference their significant work in our 1978 paper on thin solid films, which was done in 1974 at Fairchild without knowledge of Holloway and Nelson`s work. However, I don`t believe that their 1976 paper employed a definitive microanalytical technique that showed the presence of oxygen in the grain boundaries of their chromium films. I believe the grain boundary "stuffing" effect was merely implied from the determination of the (low) energy-of-activation. The presence of 5-8 at% of oxygen, which would be more than sufficient to occupy the grain boundary volume of a refractory metal with small (30-500 nm) grain size, was noted in both the papers I cited in my August letter.
To my knowledge, this was the first microanalytical data to support the "stuffing" model for improving diffusion barrier properties. Dr. Arthur Learn and Marc Nicolet, with his colleagues at Caltech, were major contributors in much of the early work, along with Ian Mitchell and Ian Macintosh at Chalk River Nuclear Labs, who did the critical nuclear activation analyses.
In summary, I do not claim to be the sole originator of "stuffed barriers," but do claim the co-authorship of definitive studies, which clearly demonstrated that grain boundary "stuffing" phenomena were indeed involved in the case of sputtered W/Ti.
Ron Nowicki
Integrated Processing
Sunnyvale, CA
Clarification
The photo of the MEMS-based mass flow controller on p. 44 in the September issue is from Redwood Microsystems Inc.