Still Room at the Bottom

Fifty years ago today, Russian cosmonaut Yuri Gagarin became the first human to leave Earth and enter space. (He was perfectly qualified for the job: he was short, and was willing to sit there and do nothing as he was hurled like a cannon ball into space.) If sputnik awoke the world to the technical possibilities of space, Gagarin awoke our sense of awe and adventure for space. I grew up in the sixties thinking that almost anything was possible, and that our future would be filled with bigger and better things. Flying into space implied that no barrier was too high to be surmounted by human ingenuity and effort.

But fifty years later the promise of space travel remains mostly promise. When I watched the 1968 film 2001: A Space Odyssey in 2001 (didn’t we all), I was struck by how little of our early vision for space exploration had actually come about. There is a simple lesson here that is very easy to forget: scaling up is hard. To build a building twice as tall requires more than twice as much steel and concrete. Launching twice the payload into space requires more than twice the rocket power. The scaling is superlinear, and that doesn’t make for good economics (or good physics). In our gravity-constrained world, bigger is sometimes better, but it is always much, much harder.

At the same time that most of us earthlings were swooning over the first manned space flight, a handful of engineers at Fairchild Semiconductor were working out the kinks on a much smaller project – connecting four transistors together on one slab of silicon to make the first commercial integrated circuit. Not too many people noticed this innovation at the time, let alone appreciated its significance. There would be no ticker-tape parades (though there would eventually be quite of few millionaires among this talented group, and even a few billionaires). But something important had begun, and the promise of the silicon IC revolution has exceeded all expectations.

(It’s interesting to note that much of the early work on integrated circuits was funded by the Apollo program in its desire to miniaturize electronics destined for space.)

And so another simple lesson is learned: scaling down doesn’t behave like scaling up. Not to say that making something smaller is necessarily easier, but smaller mean less – less material, less energy, less space. The scaling works in our favor. Of course, there are limits, and those limits become something close to insurmountable when the dimensions of the device reach atomic scales. But the room between the macroscopic dimensions of our everyday objects and the microscopic dimensions of the atomic scale is something like 6 or 8 orders of magnitude. As Richard Feynman famously said, there’s plenty of room at the bottom. Semiconductor technology has been steadily mining this room at the bottom, shrinking features from 25 microns to 25 nanometers in the last 50 years.

Is there still room at the bottom? I think so. CMOS transistors may only last for another factor of two of shrinking (or less), but other devices will allow dimensions closer to one or a handful of nanometers. And we have not yet begun to think of all the possible things we can make with a vast toolbox of micro- and nanofabrication technologies. (Alas, the phenomenal success of the CMOS transistor has probably crowded out a wide range of other useful devices.) So while the way we have scaled in the past (think Moore’s law) may not last, there is still plenty of room for innovation at the bottom. I suspect that my young children will one day marvel at the progress in scaling down during their lives, while wondering whatever happened to the promise of space travel.