Cloud and big data applications are putting new challenges on systems, at the same time that underlying silicon chip technology is reaching its limits. Bandwidth to memory, high speed communication and device power consumption are becoming increasingly challenging to improve upon. So, IBM Research is putting $3 billion into solving this “chip grand challenge” and expects not only to push silicon tech beyond seven nanometers and improve upon today’s systems, but to eventually build systems based upon non-traditional architectures that are much more efficient than today’s machines.
The Smarter Planet blog caught up with Supratik Guha, IBM Research’s director of Physical Science, to find out what it means – and what’s required – to go beyond silicon.
Smarter Planet: A nanometer is 100,000 times smaller than the width of a human hair. How do we work at those dimensions?
Supratik Guha: We are able to work at these dimensions because of the fundamental materials science and physics that has been carried out in research labs worldwide. Over the past 30 to 40 years, we’ve been trying to measure, control, deposit, and pattern materials at nanometer scales. The 1980s to 1990s were an era where many new measurement techniques were developed. For example, near field techniques such as the atomic force microscope, and new methods in electron microscopy, and medium energy ion spectroscopy. Without these measurement techniques we would not be able to develop the kinds of technologies needed today at such precision.
Then over the last 20 years, methods were developed for depositing nanometer thick layers over contoured surfaces, such as atomic layer deposition. Now, there are new methods being developed to pattern or position at nano dimensions, such as directed self assembly. And over the years, lithographic techniques have become increasingly sophisticated and able to pattern to smaller dimensions.
SP: What materials and other techniques are being experimented with as possible silicon replacements?
SG: About a decade ago scientists worldwide started looking for a replacement for silicon, anticipating that the day would come when devices would be so small that silicon would no longer be an appropriate material to use in semiconductors. Many different materials were, and are being investigated. Of the different options, IBM is looking into are carbon nanotubes, III-V semiconductor compounds and a radically new approach called piezotronics which you can think of as a solid state relay based switch.
None of these technologies are ready for “prime time” as there are clear technical challenges to overcome before we can consider them legitimate contenders. Some of them, like piezotronics, are at very early stages of feasibility demonstration.
SP: What are the advantages of these materials, and why do we see them as sustainable replacements for silicon?
SG: Let’s look at carbon nanotubes. They have three properties that distinguish them. One is that charge carriers move very fast within them, so we can therefore expect fast devices. Secondly, they are shaped like a tube and therefore they also possess the perfect electrostatics for ultra-small devices. And third, they have a band gap that would allow operation at 0.3-0.4 volts.
Carbon nanotube transistors would provide a factor of about a 5-10X improvement in figure of merit over silicon at around 5 nm.
SP: How will system architectures need to change in order to integrate with, and take advantage of non-silicon chips?
SG: A non-silicon chip for a new, non-conventional architecture is a field that is very much in early stages. It is an important field, but one that is quite undefined, today. Scientists are working in this because of the promise of either extremely high power efficiency benefits, or the ability to solve problems that are unsolvable or very difficult to solve today.
Read more about IBM Research’s $3B investment, here.
Download the infographic, here.