IEEE Spectrum October, 2016 - 12

were connected so they would
become part of the same transistor. Stacking the nanowires
boosts the amount of current
that can pass through a given
area of transistor-and thus
how much can pass through
a given area on a chip.
The great advantage of this
approach, explains Imec
logic program manager Naoto
Horiguchi, is that the fabrication process is quite close to
that used to make the FinFET
transistors in today's highperformance chips. "This is
kind of a natural extension
from FinFET to the next generation," he says.
Imec's is the first demonstration of stacked-nanowire
devices at such a small scale,
says Michael Guillorn of IBM's
Thomas J. Watson Research
Center in Yorktown Heights,
N .Y. G u i l l o r n h a s b e e n
involved with IBM's gate-allaround device research program since 2009.
IBM presented results at
last year's VLSI event that
explored the limits of performance for a single layer of
horizontal nanowires at an
advanced chip manufactur-

"This is kind
of a natural
extension
from FinFET
to the next
generation"
-Naoto
Horiguchi,
Imec

OCT 2016

NORTh AMERICAN

ing level. Because resistance
increases as a channel narrows, a common concern has
been that nanowires would
too strongly resist the flow of
current, Guillorn says. But the
IBM group's results suggest
that is not the case. They also
suggest that more ribbonlike
nanowire structures-wide
and flat instead of square or
round in cross section-could
help improve performance.
There is still plenty of work
to be done. One focus at Imec,
Mertens says, is on ways to
better isolate the gate that
surrounds the channel of
the nanowire from the source
and drain regions. Because
the source and drain are
so close to the gate in the
small device, they tend to
exchange energy via capacitive coupling.
Guillorn says researchers
are, one by one, "knocking
things off the list of reasons"
why nanowires might not be
desirable. Assuming engineers work out all the kinks
in manufacturing and performance, the devices could
potentially enter chipmakers'
production lines in three to
six years, he says. The
t i m i n g w i l l h i n ge o n
h ow l o n g c o m p a n i e s
believe they can extend
the FinFET, the cadence
at which they decide to
introduce future generations of chips, and how
aggressive they want to be
with the underlying technology. "Early adoption of
gate-all-around," Guillorn
says, "is a great way for a
semiconductor company
to differentiate itself."
-r achel courtl a nd

SPECTRUM.IEEE.ORG

The InTerneT of
fewer ThIngs
early predictions of 50 billion
connected devices by 2020 are
being scaled back
If you follow discussions about the Internet of Things, you've probably heard this
stunning prediction at least once: The
world will have 50 billion Internet-connected devices
by 2020. Ericsson's former CEO, Hans Vestburg, was
among the first to toss out that number, when he
gave a 2010 presentation to shareholders. The following year, Dave Evans, who worked for Cisco at the
time, published the same prediction in a white paper.
Today, that figure has arguably done more than any
other statistic to set sky-high expectations for potential IoT growth and profits. Remarkably, those projections weren't even close to the highest at that time.
A 2012 IBM forecast predicted 1 trillion connected
devices by 2015. "The numbers were getting kind of
crazy," recalls Bill Morelli, director of the IHS Markit
division that handles IoT and digital security.
Both Ericsson and Evans have since lowered
their expectations for 2020: Evans, a cofounder of
Stringify, who now serves as its chief technology
officer, says he expects to see 30 billion connected
devices by then; Ericsson figures on 28 billion by
2021. Other firms have adopted similar tones: IHS
Markit projects 30.7 billion IoT devices for 2020;
Gartner expects 20.8 billion by that time (excluding smartphones, tablets, and computers); and
International Data Corp. anticipates 28.1 billion
(again, not counting those devices).
That's likely because it's the third quarter of 2016,
and we're nowhere near 1 trillion IoT devices-
or even 50 billion for that matter. The true total
is somewhere between Gartner's estimate of
6.4 billion (excluding smartphones, tablets, and
computers), and IHS Markit's estimate of 17.6 billion
(with all such devices included).
But the popular 50 billion figure continues to
be widely cited. Even Evans is a bit surprised by
its staying power. "I think people do tend to latch
onto numbers that seem really hard to fathom," he
says. "Fifty billion is pretty staggering."
Peter Middleton, a research director at Gartner

http://SPECTRUM.IEEE.ORG

Table of Contents for the Digital Edition of IEEE Spectrum October, 2016