Modern Consumer-Grade Computer Microprocessor-Silicon Transistors by Sasa Macura
Modern Consumer-Grade Computer Microprocessor-Silicon Transistors by Sasa Macura
Introduction
Current consumer-grade computer microprocessors are sold by two manufacturers. Intel,
who is highly dominant in the market, and AMD, who is the small guy comparatively. What
most people dont realize is that these manufacturers actually contract out the real manufacturing
to one of several companies in the field of Nano manufacturing. In turn each of these companies
may have a completely separate way of building their transistors and chips. Intel for example,
actually owns a fabrication lab for its manufacturing and thus has some great advantages when it
comes to pricing and R&D. On the other hand AMD has a contract with GlobalFoundries to
manufacture its chips.
Currently all transistors are made in one form or another with silicon. The first silicon
transistor was developed by Morris Tanenbaum in 1954. This transistor was developed by
Tanenbaum under the employment of Bell Labs.1 Here we see the transistor from Tanenbaum
next to the one that Texas Instrument released earlier, but developed later:
Fundamental Concepts
There are three fundamental concepts when you talk about consumer-grade transistors.
There is size, function, and materials. In modern technology the size is the hot topic currently.
Everything needs to downsize but become more efficient.
Size
Silicon based transistors have a simple operation in modern-day computers, they
determine when a current is off or on. Essentially all a computer consists of underneath the GUI
(graphical user interface) is 1s and 0s, no matter the operating system or what function it is
trying to carry out. Computers are complicated machines that come down to such a base
function that it is almost laughable. In essence the transistor count determines the speed at which
computers calculate problems and carry out functions. It is imperative to reduce the size of
transistors in order to increase the transistor count per chips. Modern day CPU (central
processing units) have between 1.2-1.4 billion transistors each, which, compares to only 10 years
ago at a count of 230 million.2 This introduces us to a concept known as Moores Law.
Gordon E. Moore is one of the co-founders of Intel. In 1975 he stated that the circuit
density would double every 24 months. Current forecasts predict a slow-down in density due to
problems with the use of silicon. As silicon based transistors become smaller they have an issue
with functionality. The heat output is significantly increased and there can be problems with
precision. Currently the smallest in-use transistor size is 14nm on some of the current Exynos
chipsets. To counter this the Fab labs need to invest in R&D for other materials.
Materials
Modern transistors mainly use silicon as their material. A large reason for the use of
silicon is because it is a very common element in nature, and so it significantly reduces the cost
of manufacturing. Silicon also has the advantage and disadvantage of not being the greatest
conductor of electors. This in turns allows you to easier turn off or 0 during an operation, but
also makes it harder for the 1 to come through quickly.
Some exotic materials are currently being experimented with and have some promising
results. The main issue behind any type of change from one material to another is that there will
need to be a change in technique. The most promising and exciting material currently being
experimented with is graphene. Graphene is a single atom thick layer of carbon. The advantages
to graphene include an increase in heat tolerance and an increase in efficiency. The current test
being ran have shown promising results in the 100GHz range by IBM in 2010.3 Manufacturing
graphene is also pretty simple. It is literally just separating layers from a lump of carbon, people
have demonstrated it with something as simple as tape and graphite from pencil shavings.
Manufacturing this material may be easy but it turns into a problem. Due to the nature of the
graphene, mainly its size, a large amount cannot easily be made.
A promising and very new material that has just recently been discovered is silicene. It is
basically similar to graphene but the structure is made of a single layer of silicon, which is the
traditional material in transistors.4 Advantages of silicene commercialization would include a
very large abundance of raw material due to it being silicon based. As things currently stand
silicene is a very new structure and there is still not much known as far as practicality or even
functionality is concerned.
Function
Although the function of a transistor is to turn a current on and off it is a rather
complicated function to accomplish. In order for a transistor to work there must be some type of
electron movement occurring. The way current transistors achieve this using silicon is by
creating one that has more free electrons and one that has less. The side with less will attract
electrons thus creating an inward flow. The one with more will repel them and will create an
outward flow. These silicon pieces are sandwiched in layers with material to help move the
electrons in or out. This can help create a switch that is either off or on. When layers are made
with oxide and certain metals they are called MOSFETs.
Current Investigations
All the large current manufacturers of both commercial and industrial based micro
processing solutions have money invested in research for the future of silicon based transistors.
Intel having its own manufacturing wing has currently been one of the leaders of the pack, not to
be outdone by Samsung with their moves into the field of 14nm manufacturing. Some of the
competitions does not have the budget to support all the research that needs to be done to update
to a smaller scale but even then the advancements are numerous. These giants can be considered
solely responsible for the furthering of the industry and innovation in new developmental
technology for both testing and constructing such small nanoscale transistors.
The entire industry seems to be moving to a 14nm transistor and we will only see more
development in the coming years. What will be most interesting is how the current generation of
industrial grade hardware will look. In general the industry will have more advanced processing
units then the common public. Current generation consumer hardware is generally based off the
same hardware that was used about two years ago in the industrial sector. Current industrial
machines will run into the proverbial wall due to a lack of space to downsize.
Currently scientists all over the country are trying to find more creative ways to solve our
downsizing problem. The hope is to find the next life changing transistor material in order to
enable things such as implanted health monitors and tools for complex calculations. These
nanoscale improvements will enable us to use technology to an entirely different degree. When
transistors can be made small they will downsize the PCB (printed circuit board) which will in
turn enables us to be non-invasive when it comes to medical diagnostics. For example gold
nanorods could be used to create solvent sensors for water using coatings of polycationic
polymer, diallyldimethylammonium chloride.5
On the consumer computer front improvements have been happening for years, the only
fear currently is hitting a wall with research. It seems obvious by the research currently being
done by the major players in the field that there will probably not be a slow down on the
development of transistor sizes. Intel has already started to use the 14nm and has said that they
are researching even smaller manufacturing techniques.6 These techniques will help usher in a
new generation of faster and more efficient processors but will push us towards the point of
replacement for the basic silicon material used in those transistors. Something exotic will have
to be used for the future to become even smaller.
References
Riordan, Michael. The Lost History of the Transistor. IEEE Spectrum May 2004: 48. Print.
"CPU DB: Recording Microprocessor History." Communications Of The ACM 55.4 (2012): 55-
Johnson, Colin R. "IBM Demos 100-GHz Graphene Transistor | EE Times." EETimes. N.p., 5
Qing-Xiang, Pei, et al. "Effects Of Temperature And Strain Rate On The Mechanical Properties
Of Silicene." Journal Of Applied Physics 115.2 (2014): 1-6. Academic Search Premier. Web. 20
Mar. 2015.
5
Biosensors." Analytical Letters 40.11 (2007): 2067-2096. Academic Search Premier. Web. 20
Mar. 2015.
6
Mujtaba, Hassan. "Intel's 14nm Core M Series "Broadwell-Y" Processors With 4.5W TDP
Detailed - Powers Fanless and 2 in 1 Mobility Devices." WCCFTech RSS. N.p., 05 Sept. 2014.
Web. 23 Mar. 2015.