MemberSeptember 9, 2021 at 4:49 pm
When two solids are combined in just the right proportions, changes in chemical bonding may produce an alloy that melts at a temperature far lower than either can melt by itself. Such an alloy is called eutectic, Greek for “good melting.” The eutectic alloy of gold and silicon – 81 percent gold and 19 percent silicon – is especially useful in processing nanoscale semiconductors such as nanowires, as well as for device interconnections in integrated circuits; it liquefies at a modest 363˚ Celsius, far lower than the melting point of either pure gold, 1064°C, or pure silicon, 1414°C.
The rapid growth and evolution of the strange circles at 600˚C is caught in successive frames under an electron microscope (structures at right are already in place as the sequence begins). Underlying a thin layer of gold (mottled gray), a weak spot opens in the silicon dioxide barrier, allowing pure silicon in the substrate to react with the gold. A pool of molten eutectic quickly spreads (dark gray). When it becomes large enough, surface tension ruptures the liquid, pulling the eutectic aside to surround a cleared zone of silicon dioxide, now barren except for a central square of gold and silicon. Time from first to last frame is just 2.8 seconds; the area covered in each image is about 40 by 80 micrometers.
“Gold-silicon eutectic liquid can safely solder chip layers together or form microscopic conducting wires, by flowing into channels in the substrate without burning up the surroundings,” says Berkeley Lab’s Junqiao Wu. “It’s particularly interesting for processing nanoscale materials and devices.” Wu cites the example of silicon nanowires, which can be grown from beads of eutectic liquid that form from droplets of gold. The beads catalyze the deposition of silicon from a chemical vapor and ride atop continually lengthening nanowire whiskers.
Understanding just how and why this happens has been a challenge. Although eutectic alloys are well studied as solids, the liquid state presents more obstacles, which are particularly formidable at the nanoscale because of greatly increased surface tension – the same surface forces that make it difficult to form ultra-thin films of water, for example, because they pull the water into droplets. At smaller scales the ratio of surface area to bulk increases markedly, and nanoscale structures have been described as virtually “all surface.”
These are the conditions that the team led by Wu, who is a faculty scientist in Berkeley Lab’s Materials Sciences Division and a professor in the Department of Materials Science and Engineering at the University of California at Berkeley, set out to examine, by creating the thinnest possible films of gold-silicon eutectic alloys. The researchers did so by starting with a substrate of pure silicon, on whose flat surface an extremely thin barrier layer (two nanometers thick) of silicon dioxide had formed. On this surface they laid layers of pure gold, varying the thickness from one trial to the next between just a few nanometers to a hefty 300 nanometers. The silicon dioxide barrier prevented the pure silicon from mixing with the gold.
The next step was to heat the layered sample to 600 °C for several minutes – not hot enough to melt the gold or silicon but hot enough to cause naturally existing pinholes in the thin silicon dioxide layer to enlarge into small weak spots, through which pure silicon could come in contact with the overlying gold. At the high temperature, silicon atoms quickly diffused out of the substrate and into the gold, forming a layer of eutectic gold-silicon alloy nearly the same thickness as the original gold and spreading in a virtually perfect circle from the central pinhole.
When the circular disk of eutectic alloy got large enough it suddenly broke up, disrupted by the high surface energy of the gold-silicon eutectic liquid. The debris was literally pulled to the edges of the disk, piling up around it to leave a central denuded zone of bare silicon dioxide.