Another metallurgical phenomenon that is critical to VLSI circuit performance and lifetime involves the interdiffusion between thin-film multilayers; this was pointed up previously in a different context in the section on modulated structures. Of particular note is the undesirable interdiffusion that occurs at aluminum-silicon contacts, such bimetallic systems being rarely at equilibrium. Silicon can dissolve appreciably in solid aluminum at the processing or operating temperatures, causing aluminum to cross the interface into the silicon (often called “spiking”) and thereby degrade or short-circuit the semiconductor junction.73 Although the kinetics of this process can be slowed by prealloying the aluminum with silicon, much effort has been concentrated on the interposition of diffusion-barrier layers.75 Various barriers have been explored, some with considerable promise, but an orderly rationale has not yet emerged. Barriers under investigation are the silicides of Co, Ni, Pd, and Pt; TiC and TiN; and refractory metal combinations, sometimes with noble or near-noble metals. In cases where undue interaction may take place between the barrier and the aluminum or silicon on either side, multiple barriers are of interest. Here again, an underlying hypothesis is that the detrimental interpenetration is dominated by grain boundary diffusion at the low temperatures involved and that this process can be alleviated by a large grain size or by “blocking” the grain boundaries with relatively insoluble solute or impurity atoms.
The point to be stressed for present purposes is that diffusion barriers in VLSI circuits as well as metallization more generally are rich frontiers for metallurgical research. One can sense that this kind of thin-film metallurgy will soon overlap the previously discussed field of nanoscale multilayered structures in which a high degree of coherency and orientation-alignment (approaching the monocrystalline state) is maintained between the metal layers—setting the stage for unexpected atomic-bonding, mechanical, and electrical properties. It can also be anticipated that other phenomena such as phase transformations and interface-related processes in thin-film and layered alloys will be profoundly different from what is now well known about the bulk systems.
Although ferromagnetism is normally regarded as an integral part of solid-state physics, many of the technologically important magnetic materials are metallic alloys and therefore fall within the general scope of metallurgy. Magnetic materials play a crucial role in modern electronic and electrical devices and enjoy a market of about $2 billion per year (growing at the rate of some 12 percent per year) in components that are 10 to 20 times greater in added value.76 For present purposes we shall call particular attention to