Portland, Ore. A transmission electron-microscope procedure from IBM Corp. lets researchers create real-time videos of liquid deposition processes to explore their mechanisms at work.
The experimental technique allows researchers to quantify electrodeposition fine-tuning instead of depending on trial and error.
Before the IBM breakthrough, nucleation, growth and coalescence in electrodeposition processes could only be observed indirectly, by measuring the current transient and analyzing with electrochemical models. But orders of magnitude of difference were found between the parameters obtained from models of current-transient analysis and those confirmed by post-growth microscopy, IBM said.
"There are a lot of reactions which take place in a liquid environment," said Frances Ross, a researcher at the Thomas J. Watson Research Center (Yorktown Heights, N.Y.) and winner of this year's Burton Medal for contributions to microscopy. "For instance, our technique will be useful for engineers who want to make videos of materials being deposited on the electrodes of rechargeable batteries."
Much to their frustration, semiconductor engineers perfecting copper-on-silicon processes have had to rely on slow imaging techniques, such as atomic-force microscopy, which takes as long as 30 seconds per still image. Fast acquisition methods that work only at step edges have been developed, but they run at only a few frames per second, too slow for electrodeposition, which occurs in milliseconds. In addition, they do not work for the three-dimensional growth of copper on silicon, IBM said.
Now, using Ross' imaging technique, chip engineers can make 30-frame-per-second movies of the three-dimensional growth of copper on silicon using a conventional transmission electron microscope.
When chip makers went from aluminum to copper interconnects, they found the best way to deposit copper was not sputtering or evaporation but electrodepositiona liquid process in which an entire wafer is immersed in a bath of copper sulfate and sulfuric acid and voltage is applied. The copper grows in both the trenches and on the surface of the wafer. Later, a mechanical polishing step removes all copper except for that in the trenches.
"Engineers have discovered many ways to optimize copper growth, using various chemical additives to the copper sulfate and sulfuric acid bath, but even now it's not clear how they work. . . . That's what gave us the motivation to do our experiment," said Ross.
By directly observing additives under various conditions and parameter settings, Ross hopes to come up with confirmed rules that tell engineers what it is about the surface that causes copper to begin growing there.
"With that knowledge we may be able to design surfaces that have nucleation sites in the places where we want them, so that we have perfect control of the film that gets grown," said Ross.