In 1965, Gordon Moore enunciated a law that, for 50 years, has been one of the mainstays of electronics: Moore’s Law. What is this law? Is it still valid? What is the future of the development of integrated circuits?
At events like CES, held a few days ago in Las Vegas, or Computex, held in June in Taiwan, manufacturers of electronic devices we often surprise with their new products, their big announcements and, of course, their major R & D No one is surprised to see more and more powerful devices that also are extremely small and we offer great autonomy characteristics that always seem to lead to one of the great debates linked to the world of electronics: the limits of Moore’s Law.
Manufacturers like Samsung have developed new chips that have completely changed the manufacturing processes of integrated circuits, processors like the new Intel Haswell, containing 1,400 million transistors into a small cube of just 177 square millimeters and wherein each transistor measures only 22 nanometers, great technological leaps that suggest that maybe soon we will say goodbye to one of the most important tenets of the electronics world.
The Moore’s law is a postulate was enunciated in 1965 and since then has been fulfilled, a law that is often cited lot when new microprocessors are launched and worth knowing in depth:
The complexity of the components has been increased approximately 2 every year. In the short term, it can be expected that this rate will continue or even increase. In the long run, the growth rate is less predictable, although there is no reason to believe that will not remain constant for at least another 10 years. That is, at 1975 the number of components in each low cost integrated circuit will be 65,000. I think that such a large circuit can be built on a single silicon wafer.
Gordon Moore, the great visionary
The head of the title of this law was Gordon Moore an article for the journal Electronics which was published on April 19, 1965. Moore, one of the “8 traitors” who founded Fairchild Semiconductor, enunciated this assumption based on his own observations and the evolution he saw within the production chain Fairchild. Extrapolation was very interesting: every year the number of transistors that were part of an integrated circuit, thereby increasing the capacity of these doubled.
Perhaps it may seem an observation without much importance, but it is noteworthy that Gordon Moore (who later would join Robert Noyce to found Intel in 1968) made six years before the development of the first microprocessor. In fact, at that time his intention was to show that the transistor would be a key component for electronic devices and to validate the business model of Fairchild.
Although formulated empirically, Moore’s Law became a model perfectly validated by industry. The number of transistors per unit area was doubled and was reaching a new technology (with larger scale integration) that left the previous outdated. The pace at the beginning, it was dizzying, but in 1975, Gordon Moore had to readjust its own law to change the factor of time and adjust it to 2 years because the industry was not moving as quickly.
Since the 1975 adjustment, Moore’s Law has remained the gold standard in the development of integrated circuits but obviously increased scale integration (reducing the size of transistors) implies many challenges for manufacturers. Intel Haswell processors are composed of 1,400 million transistors just 22 nanometers in size each, 22 nanometer technology is virtually the ” Great Barrier Silicon “and approached an area where this material could present instabilities if we continue to increase the level of integration.
Silicon limits approaching
Approaching the limits of silicon can be a big problem. While the projections made by Gordon Moore are being met, virtually, from its title in 1965, the limits of silicon have begun to slow the progression of technology. Estimates from the ITRS (International Technology Roadmap for Semiconductors), increasing the scale of integration and therefore the evolution of the chips being produced, since 2010, every 3 years and although the chip capacity has continued increasing, largely has been so through the use of technologies with multiple cores (cores).
Although we have saved so far, the physical limits of silicon, is a scenario that inevitably approaches. During the last LinuxCon, Linus Torvalds spoke precisely of Moore’s law and its possible end, a big challenge for the industry may not be ready but, inevitably, will perhaps in the next 10 or 15 years.
In 2005, self- Gordon Moore stated in an interview that his statement would be valid for 10 or 15 years, therefore, could peaking between 2015 and 2020., at least as relates to silicon will reach a point where if we go down over the size of transistors, they stop working and stop behaving as expected.
New materials and manufacturing processes
The change cycle approaches and industry takes time preparing for it. The end of Moore’s law does not imply the collapse of our technology but will bring about changes in the materials used for the development of electronic components and obviously the chip manufacturing processes.
Research projects such as Samsung and the University of California-Los Angeles (UCLA) while working in memories flash extremely small that may develop from 10 nanometers transistors. To reach the limit of 10 nanometers and therefore overcome the barrier of high current 22, the project has put the spotlight on a new material graphene.
Maybe still is early to say that graphene replace silicon, but it certainly has been postulated as a perfect complement that can stabilize the silicon when physical barriers are overcome. In addition, they are also conducting multiple investigations revolving around the graphene and the development of a new generation of chips of this material to permit work to very high frequency.
In addition to the inclusion of new materials, manufacturing processes will also change. Clearly, the challenge is great but many companies have already implemented these changes, for example, Samsung or Matrix Semiconductor have been working on their “three-dimensional chips” that allow scaling up integration without substantially changing current processes photolithography.
McKinsey recently published a report precisely analyzing the impact of the end of Moore’s law in the integrated circuits industry. From their perspective, McKinsey predicts a transition that will go through changes in manufacturing processes in order to extend the life of silicon. A change that is not considered as profitable for the manufacturer and that could lead to the end of Moore’s Law for the year 2018 and perhaps the start of a new era in electronics.
It’s amazing how a statement, formulated almost 50 years ago, has been able to make the technology that we know today, a law based on the observation that, perhaps, is living his last years.