SAN FRANCISCO With the embrace of other materials such as silicon germanium (SiGe) and silicon-on-insulator (SOI) for IC fabrication, the semiconductor industry has over the past decade entered into the "super-silicon era," when silicon on its own is no longer adequate to address the requirements of electronics, according to Lewis Counts, vice president of analog technology and a Fellow at Analog Devices Inc.
"I submit that over the last 10 years, we've moved to the super-silicon era," Counts told an audience of hundreds who attended his keynote presentation at the International Solid State Circuit Conference (ISSCC) here Monday (Feb. 12).
Counts praised the invention of the planar process by Jean Hoerni and his colleagues at Fairchild Semiconductor in the 1950s, calling it "one of the most profound achievements of the 20th Century" and saying that it established silicon as the key material in electronics. But, he said, far more complex processes have evolved in recent years to keep pace with more demanding system requirements. Today, he noted, processes ranging from SiGe BiCMOS to 100V double-diffused metal oxide semiconductor (DMOS) are required to optimally support many systems.
Counts said that it is likely that other new structures and materials will be introduced into semiconductor manufacturing that will drive device performance beyond levels that can currently be envisioned.
"Can we continue to improve speed and performance without reducing dynamic range for analog and mixed-signal devices? I believe we will be able to do that, but not without fundamental changes to the structure of devices," Counts said.
Counts defined the five biggest challenges facing analog/mixed-signal chip design: dynamic range, bandwidth, absolutes (voltage, amps, seconds and Kelvins), power efficiency and complexity. He said that the lines between analog and digital design were blurring, but that the two remain distinct and separate disciplines. "But at the highest levels, design is design," he added.
Counts also touched on the challenges associated with integrating multiple radios in a cellular handset, estimating that future phones will have at least six radios to support communication standards such as Wi-Fi, Bluetooth, WiBree, Ultra-Wideband (UWB), WiMax, Near Field Communication (NFC) and RFID. This undertaking will be challenging technically, Counts said, and also from a business perspective, since successfully integrating these radios will need to be done while keeping costs and power consumption to a minimum.
In general, the business of handsets is getting more complex as the devices evolve to include more capabilities, Counts said. "We expect one thing to do everything in less than 2 watts," Counts said. "It's almost unfair to call it a cell phone anymore."
Counts said that designers don't want to create multiple antennas for each radio, but that some form of adaptive antenna is going to be needed. This adaptive antenna could perhaps be a MEMS switch, he speculated.
Counts also noted that transistors are rapidly replacing inductors in radio design. The use of inductors in radios dates back to inventor Guglielmo Marconi, he said, but they take up too much space and pose a barrier to miniaturization. Since the 1980s, he said, transistors have been replacing inductors wherever possible. But the use of inductors in radios will not completely disappear, Counts said, because inductor-based oscillator circuits require less power than those based on transistors, due to phase noise issues.
"The number of inductors per chip won't go to zero," Counts said, "but will certainly be minimized."
Counts also sat down with EE Times editor-in-chief Brian Fulller to discuss where he thinks the industry is headed. See the video interview here.