Is it fair or reasonable to examine all the societal and technological change that occurred during the 20th century, and project a roughly equivalent amount of change in this century? Would that be expecting too much change, or perhaps too little? How much power will modern, accelerating technologies possess by the year 2100, 92 years from now?
Today we are pleased to have a guest blog entry from Tihamer Toth-Fejel, a senior research engineer at General Dynamics Advanced Intelligence Systems, a science advisor to the Lifeboat Foundation, and a member of CRN's Global Task Force. He asks:
One of the points to remember is that Moore’s Law is exponential — or even faster, if Ray Kurzweil is correct. So it’s not really a question of what kind of technology did we have 92 years ago...we should ask about 920 years ago, or even 9200 years ago.
In other words, if Freitas, Merkle, and Moriarty succeed next year, all heck will break loose. Not immediately, of course, but diamond is a very useful engineering material. If we could use it to make cars and buildings, we would. Everything depends on availability and cost — and both depend on technology.
Four years from now, the Zyvex-led DARPA Tip-Based Nanofabrication project expects to be able to put down about ten million atoms per hour in atomically perfect nanostructures, though only in silicon (additional elements will undoubtedly follow; probably taking six months each). At a standard Moore’s Law exponential growth rate (doubling time of 18 months), this Patterned Atomic Layer Epitaxy (Zyvex’s approach) will only get us up to 23,058,430,092,136,939,520,000,000 atoms per hour by 2100 — a few hundred pounds worth.
On the other hand, does atomically precise tip-based nanofabrication follow Moore’s Law? If we take Eigler’s 35 xenon-atom IBM emblem in 1990 as the start, then that gives us a doubling time of just over 12 months, and that adds up. By 2012, we’ll be doing about ten times better than Moore’s law. And that is without the big kick.
This is the big kick: What happens when we use probe-based nanofabrication to build more probes?
It’s starting to happen now (see “Thermal Actuated Multi-Probes Cantilever Array for Scanning Probe Parallel Nano Writing System” by Watanabe, Isono, et al). Chad Mirkin, who also has another piece of the DARPA Tip-Based Nanofabrication project, has already used 55,000 dip pen nanolithography tips to make 1,600 100 nm dots in under 30 minutes. (Mirkin is using standard microphotolithography MEMS to make the dots; this is the most conservative approach to productive nanosystems. The others include Structural DNA — Rothemund, Nanorex, et. al — and Schafmeister’s Bis proteins, plus a few more not as promising.)
What happens when productive nanosystems get built, and are used to build better productive nanosystems? The exponential increase in atomically precise manufacturing capability will make Moore’s law look like it’s standing still.
As Tihamer suggests, we seem to be on the cusp of building nanoscale machines that can build other nanoscale machines. Following that "big kick," the only practical limits on how much power the technology can provide will be software design and human imagination.