How Close is Molecular Computing?
August 20, 2012
•IT and Computer Science
• 2 Comments
By Bruce Harmon, Ph.D., CTU Doctoral Chair of Computer Science
While it may seem like science fiction, modern advances in semiconductor manufacturing meant to support denser integrated circuits, or chips, will soon have us trying to imprint transistors and interconnect of a size less than ten nanometers. According to the International Technology Roadmap for Semiconductors, known throughout the world as the ITRS, this scale will be near the size of individual molecules. This progress raises the question: could molecules be made to serve as transistors and interconnect?
This question first sparked my interest about twelve years ago when colleagues of mine at HP Laboratories began to publish results on the subject. Having worked with the research team just two years prior, I traveled to Palo Alto, California, to meet with the scientists and better understand their work. They were able to demonstrate that a rotaxane molecule could be electrically induced to one of two states: the basis for an element of memory, and one that had rudimentary interconnect.
Keeping Up with Moore’s Law
The motivation for using molecules as transistors is rooted in the extraordinary progress made in semiconductor manufacturing over the last forty years. Ever since Gordon Moore of Intel made his famous prediction, dubbed Moore’s Law, that we would double the density of transistors on integrated circuits every two years, the exponential growth in circuit density has led to the ubiquity of chips that power our everyday devices. From tablets and computers, to smart phones and highly complex automobiles, chips are the lifeline of modern electronics.
Attempts to stay on track with Moore’s Law have become increasingly difficult when attempting to use photolithography to imprint circuit features on silicon that are a fraction of the wavelength of the illumination source. Resulting diffraction has challenged manufacturers and led to sharp increases in the cost of modern manufacturing. While we seem close, the promise of extreme ultra-violet illumination sources needed to imprint at a molecular scale is still outside our reach.
Looking Ahead
Faced with inherent challenges of photolithography, researchers all over the globe have been seeking suitable molecules that could be made to self-assemble. The most promising to date has been the carbon nanotube, an array of carbon atoms, each sharing four outer electrons with its neighbor and bound into an elongated cylinder about 5 – 15 times its diameter in length. Each molecule would be composed of between 20 – 60 atoms and has been shown to control current much like a transistor. The interconnect could be formed using graphene, a thin ribbon just one atom thick and extracted from graphite; which has proven to be exceptionally strong structurally and an excellent conductor.
While we are many years from solving all the problems necessary for molecules to serve as the basis for electronic circuits, we are off to a promising start. Like microchips today, we can expect extremely small computing engines to allow for molecular computing to become ubiquitous, as well. They will be interconnected through wireless technology, communicating with each other and humans. The adoption of mobile computing and communications will skyrocket further, changing all of our lives for the better.
Have a question or prediction about molecular computing? Tweet @CTUtech, or leave a comment below.
Bruce Harmon, Ph.D., is the University Doctoral Chair of Computer Science at Colorado Technical University.
Image credit: The Harvard Independent