3D Transistors: Better Performance at Even Lower Power
Technology has progressed quite swiftly, since one of the greatest inventions-transistor, paving way for more powerful, cost-effective and energy efficient products. For the first time in history, silicon transistors have entered the third dimension with the Tri-Gate transistors. Read on to find out it's features, structure and how it benefits the new-age processors.
So
what comes to your mind when you hear the word 'Intel'? Definitely, a
manufacturer of all your celeron's, pentium's, dual cores, core i's, and
other micro-processor chips, right? The pace dictated by Moore's Law
has required numerous innovations and as a result of which the 'Sponsors
of Tomorrow' introduced a three dimensional transistor technology,
which is basically, Tri-Gate transistors on its 22 nm logic technology.
Important Turning Point in Transistor Technology
In
1947, the first transistor was demonstrated at Bell Laboratories.
Silicon was first used to produce bipolar transistors in 1954, but it
was not until 1960 that the first silicon metal oxide semiconductor
field-effect transistor (MOSFET) was built. The earliest MOSFETs were 2D
planar devices with current flowing along the surface of the silicon
under the gate. The basic structure of MOSFET devices has remained
substantially unchanged for over 50 years.
Continued optimization
and manufacturability studies on 3-D transistor structures was on at
research and development organizations in leading semiconductor
companies. Some of the process and patent development has been published
and publicly shared, and some remained in corporate labs. The
International Technology Roadmap for Semiconductors (ITRS) drives the
research investment interests of the semiconductor industry, which is
coordinated and published by a consortium of manufacturers, suppliers,
and research institutes.
The ITRS defines transistor technology
requirements to achieve continued improvement in performance, power, and
density along with options which should be explored to achieve the
goals. The ITRS and its public documentation captures conclusions and
recommendations regarding manufacturing capabilities like strained
silicon and High-K metal gate, and now the use of 3-D transistor
technologies to maintain the benefits of Moore’s law. Based on documents
produced by the ITRS and an examination of academic papers and patent
filings, research into 3-D transistor technologies has grown
dramatically in the last decade.
The Triggers That Threw Spotlight on 3-D Transistors
Two
important pronouncements that have occurred in the last two years that
have propelled the 3-D transistor structure into the industry spotlight,
and into a permanent place in the technology story of MOSFET
transistors are 1) The first announcement by Intel Corporation on 4th of
May, 2011, about their Tri-Gate transistor design that had been
selected for the design and manufacture of their 22 nm semiconductor
products. 2) The second announcement was the publication of ITRS
technology roadmaps, with contributions from many other semiconductor
manufacturing companies that identified 3-D transistor technology as the
primary enabler of all incremental semiconductor improvement beyond the
20 nm or 22 nm design node.
So Exactly How Small is 20 nano-meter?
Each
and every micro-processor manufactured today is made of millions, or
even billions, of tiny electrical components called transistors. Over
time, in accordance with Moore's law, transistors have been getting
smaller and smaller and because of which, computing and communication
devices continue to get smarter, faster and highly efficient.
Intel
believes that keeping up with Moore's Law has never been exactly easy.
Especially for 22 nanometers, Intel claims that it became clear early on
that continued shrinking was not going to give the expected benefits
without some radical redesign. After a decade of research and
development, taking advantage of the work of Hisamoto and others in
FinFET development and optimization, Intel invented the solution. For
the first time in history, the transistor has officially entered the 3rd
dimension. Image 1 shows what a 3D transistor looks like.
Image 1: A 3D transistor chip
In
fact, there are more than a billion transistors on this single chip
which, unfortunately, are far too small to be seen with the naked eye.
Now imagine yourself 20,000 times smaller! To give you a point of
reference, right now, you are even smaller than a human hair. But you
would actually still be far too large when compared to a 22 nano-meter
transistor for a meaningful imagination. You actually need to be 100
nano-meters tall or about 20 million times smaller than your actual
size. At this scale, you are about the right size to literally see,
demonstrate some of the attributes and functions of a single, modern
transistor out of those millions of transistors inside the 3D transistor
chip. Well, since there's no such shrinking device as yet, let us just
consider the figures below as transistors models and understand.
But first, what are 2-D transistors?
Image 2: Traditional Planar 2-D Transistor
For
the last four decades, planar or 2-D transistors, have been at the core
of transistor design and architecture. In the image-2, we see a form of
silicon that creates a stream (dotted yellow) through which electrons
flow. The gate, which is made of metal over a material with high
dielectric constant, controls the flow of electricity in that stream. It
acts as an ordinary switch, turning flow on and off. That is, if an
ordinary switch had the ability to turn itself on and off over 100
billion times a second! Technically, traditional 2-D planar transistors
form a conducting channel in the silicon region under the gate electrode
when in the “on” state. Talking about states, some key objectives in
transistor design are to have as much current flowing as possible when
in the “on state” for performance, to have as close to zero current
flowing when it is in the “off” state to minimize power usage, and to
switch very quickly between the two states again, for performance.
Now Coming to 3-D Transistors...
Image 3: 22nm Trigate Transistor
As
transistors get ever smaller, one way to achieve this is to get tighter
control, by having the gate wrap around the channel as much as
possible. The animated version of the transistor can be seen in image-3.
With Intel's 3D transistor's architecture, the flat two dimensional
stream has been replaced with one or more three-dimensional fins as
shown in the image-4.
Image 4: 22nm Trigate Transistor
The
control is on all the three sides of each fin, rather than just one, as
in the the Planar 2-D transistor. In simpler terms, the transistor
channel is raised into the 3rd dimension. Current flow is controlled on
three sides of the channel (top,left and right). This is called a
Tri-Gate transistor and its real advantage over Planar is the ability to
operate at lower voltage with lower leakage, providing an unprecedented
combination of improved performance and energy efficiency. This
breakthrough invention allows Intel to create transistors that are
smaller, faster and use less power than ever before, enabling a new
generation of computing technology in every category, from the fastest
super computers to the smallest hand-held devices. Tri-Gate transistors
can have multiple fins (as shown in image 5) connected together to
increase total drive strength for higher performance.
Image 5: Tri-Gate transistors with multiple fins
The Real Deal With 3D
The
3-D geometry and structure of the Tri-Gate transistor provides a host
of important improvements over the planar transistor structure, all
related to the ‘wrap-around’ effect of the MOSFET ‘gate’ around the
source-to-drain ‘channel.’ These advantages manifest in improved
performance, reduced active and leakage power, transistor design
density, and a reduction in transistor susceptibility to charged
particle single event upsets (SEU).
The power advantage results
from the improved control of the channel by the gate’s electric field on
three sides of the fin. As explained by Intel Corporation at their
Intel Developer Forums (2011, 2012), this power advantage is created by
an effectively steeper transistor voltage curve for Tri-Gate
transistors. Transistor designers can take advantage of this steeper
curve with either a significant reduction in leakage current for the
same performance of a planar transistor, or substantially higher
performance (transistor operation speed), or a combination of both.
The Real Advantage of Tri-Gate Transistors
·More than 50% power reduction at constant performance.
·37% performance increase at low voltage.
·Improved performance and efficiency.
"For
years we have seen limits to how small transistors can get," said
Gordon E. Moore. "This change in the basic structure is a truly
revolutionary approach, and one that should allow Moore's Law, and the
historic pace of innovation, to continue." - Gordon E. Moore
"The
performance gains and power savings of Intel's unique 3-D Tri-Gate
transistors are like nothing we've seen before. This milestone is going
further than simply keeping up with Moore's Law. The low-voltage and
low-power benefits far exceed what we typically see from one process
generation to the next. It will give product designers the flexibility
to make current devices smarter and wholly new ones possible. We believe
this breakthrough will extend Intel's lead even further over the rest
of the semiconductor industry." - Mark Bohr, Intel Senior Fellow
FUN FACTS: EXACTLY HOW SMALL (AND COOL) IS 22 NANOMETERS?
The
original transistor built by Bell Labs in 1947 was large enough that it
was pieced together by hand. By contrast, more than 100 million 22nm
tri-gate transistors could fit onto the head of a pin*.
More than 6 million 22nm tri-gate transistors could fit in the period# at the end of this sentence.
A 22nm tri-gate transistor's gates that are so small, you could fit more than 4000 of them across the width of a human hair^.
If
a typical house shrunk as transistors have, you would not be able to
see a house without a microscope. To see a 22nm feature with the naked
eye, you would have to enlarge a chip to be larger than a house. (4)
Compared
to Intel's first microprocessor, the 4004, introduced in 1971, a 22nm
CPU runs over 4000 times as fast and each transistor uses about 5000
times less energy. The price per transistor has dropped by a factor of
about 50,000.
A 22nm transistor can switch on and off well over
100 billion times in one second. It would take you around 2000 years to
flick a light switch on and off that many times**.
It's one thing
to design a tri-gate transistor but quite another to get it into high
volume manufacturing. Intel's factories produce over 5 billion
transistors every second. That's 150,000,000,000,000,000 transistors per
year, the equivalent of over 20 million transistors for every man,
woman and child on earth.
*A pin head is about 1.5 mm in diameter.
#A period is estimated to be 1/10 square millimeter in area.
^A human hair is about 90 microns in diameter.
(4)The smallest feature visible to the naked eye is 40 microns.
**Assumes a person can flick a light switch on and off 150 times per minute.
Table Courtesy – Intel's Press Material on 22 nm 3-D transistor technology
Tri-Gate Devices Now in Production
Image 6: 22nm Manufacturing Fabs
The
advanced state of semiconductor manufacturing at very small geometries
(40 nm, 28 nm, 22 nm or 20 nm and beyond) requires research and
development expenditures that now limit this technology to a handful of
companies with capital expenditure capabilities in the billions of
dollars. As a result, only a handful of manufacturers are able to
capitalize on the known advantages of 3-D transistor technology. Intel
Corporation is the only company to have made this
design and manufacturing transition in 22 nm technology, and can provide
data on the overall maturity and manufacturability of Tri-Gate
transistors on a mass production scale. This data, as of the first
quarter of 2013, includes 100 million units of Tri-Gate transistorbased
products.
Image 7: Gates and Fins of 22 nm 3-D transistor
Several known issues and characteristics of the 3-D gate structure
have been acknowledged and addressed to achieve manufacturing and
design maturity with the technology. These include the modeling of new
parasitic capacitance values not modeled in traditional planar designs,
layout dependent effects, and the use of double-patterning techniques
using current lithographic equipment to form closely spaced fins. A
great deal of publicity and user education is underway in 2013 by
companies like Cadence and Synopsys revolving around the impact of
Tri-Gate rules and flexibility in the design of future semiconductor
products.
Impact on FPGA and Other Semiconductor Device Performance
Let's
see how this three dimensional technology will provide a significant
boost in the capabilities of high-performance programmable logic.
The
primary advantage of Tri-Gate technology to FPGA-based electronic
product designer is the continuation of Moore’s Law in the steady march
of improvements in transistor density, performance, power, and
cost-per-transistor. This sustains an industry of consumer electronics,
computing platform development, software complexity advances, memory and
storage growth, mobile device creativity and development, and business
automation and productivity.In addition, control over the static and
active power dissipation of semiconductors improves tremendously with
this technology. For users of FPGAs, this makes programmable logic that
advances to 14 nm technology and beyond both power competitive with ASIC
and ASSP design solutions on available competing design nodes, with
even more significant advantages in programmability, performance,
flexibility, Open Computing Language (OpenCL™) software design entry,
and integration of DSP, transceiver, hardened processor, and
configurable I/Os.
Image 1: A 3D transistor chip
Image 2: Traditional Planar 2-D Transistor
Image 3: 22nm Trigate Transistor
Image 4: 22nm Trigate Transistor
Image 5: Tri-Gate transistors with multiple fins
More than 6 million 22nm tri-gate transistors could fit in the period# at the end of this sentence.
A 22nm tri-gate transistor's gates that are so small, you could fit more than 4000 of them across the width of a human hair^.
If a typical house shrunk as transistors have, you would not be able to see a house without a microscope. To see a 22nm feature with the naked eye, you would have to enlarge a chip to be larger than a house. (4)
Compared to Intel's first microprocessor, the 4004, introduced in 1971, a 22nm CPU runs over 4000 times as fast and each transistor uses about 5000 times less energy. The price per transistor has dropped by a factor of about 50,000.
A 22nm transistor can switch on and off well over 100 billion times in one second. It would take you around 2000 years to flick a light switch on and off that many times**.
It's one thing to design a tri-gate transistor but quite another to get it into high volume manufacturing. Intel's factories produce over 5 billion transistors every second. That's 150,000,000,000,000,000 transistors per year, the equivalent of over 20 million transistors for every man, woman and child on earth.
*A pin head is about 1.5 mm in diameter.
#A period is estimated to be 1/10 square millimeter in area.
^A human hair is about 90 microns in diameter.
(4)The smallest feature visible to the naked eye is 40 microns.
**Assumes a person can flick a light switch on and off 150 times per minute.
Table Courtesy – Intel's Press Material on 22 nm 3-D transistor technology
Image 6: 22nm Manufacturing Fabs
Image 7: Gates and Fins of 22 nm 3-D transistor
No comments:
Post a Comment