By Bob Haavind, Editorial Director, Solid State Technology
Enormous potential markets are emerging for semiconductors, said Jai Hakhu, corporate VP, Intel, at The ConFab. However, serving them will require continued innovation, and an intense focus on efficiency by the industry. Chips will have to provide multiple functions at lower price points to capitalize on the opportunities, he stated.
Hakhu cited several segments where there could be huge growth, including mobile applications, the digital enterprise, emerging nations such as China and India, and digital health devices.
In developing nations, he said, kids want more powerful mobile multifunction computers for under $1000.
In the health area, there are 200,000 hospitals with 18 million beds, but their methods are antiquated with medical records in file folders taking weeks to analyze and ship around. Hakhu said he knows of only two hospitals that are fully automated, in Germany and Japan.
He expects these huge growth markets to continue the advance of semiconductor products as long as the industry can continue to track Moore’s Law.
Hakhu showed how Gordon Moore, with just a few data points, had accurately predicted exponential growth that has continued for over 40 years now. “If Gordon Moore’s paper had been in economics, he’d have won the Nobel Prize by now,” he commented.
But tough challenges must be met to keep the trend going, according to Hakhu. He showed, for example, that the industry must continue to shrink features, holding power down while boosting performance. He showed that high-k dielectrics will have to replace SiO2 to allow this. One reason is that two or more processors must now be put on a chip to maintain performance, but the chip size can’t grow, or the economics of wafer processing will be disrupted.
The shrink will also require EUV lithography, but the industry will have to figure out how to process 120 wafers/hour with EUV. Also, as the elements used in chipmaking go from about six in the 1980s to almost the whole periodic table in the future, extensive materials R&D will be needed. He sees universities playing a key role, but he opposes businesses being set up out of university research projects.
Interconnects will also prove difficult, and he’s not sure what solution will emerge. Although air gaps have been used in gallium arsenide chips, it was only for a couple of layers. He can’t see air gaps for the 8-9 layers needed for advanced CMOS chips.
Hakhu also commented that research can often be effective in generating technical papers, but to be useful the results must be transferable into manufacturing.