Still room at the bottom?

November 19, 2005 11:33 PM

In 1959, Richard Feynman gave a now-famous talk at Caltech titled Plenty of Room at the Bottom in which he challenges the scientific community to write the entire Encyclopedia Brittanica on the head of a pin. Has anybody ever done this?

In this speech, Feynman offers a prize of $1,000 to anyone who can write a page of a book at 1/25000 of its original size, which is the reduction required to fit the entire encyclopedia on a pinhead. I understand this prize was claimed in 1985 by a graduate student at Stanford, Tom Newman. But has anybody ever actually gone all-out and put the entire EB on a head of a pin?

posted by sergeant sandwich to science & nature (17 comments total)

Perhaps not the EB exactly, but this will probably get you close, if not past, the data density you need..
posted by Rothko at 11:41 PM on November 19, 2005

fixed link.
posted by ori at 11:51 PM on November 19, 2005

from Rothko's article:
IBM's test platforms can store up to 390 bits per square micron (a micron is a millionth of a metre).

the Stanford project that won Feynman's prize was an image of the first page of Dickens' A Tale of Two Cities in a 6 x 6 micron area, written using electron-beam lithography which routinely writes feature sizes of 10 nm. 390 bits per square micron, while cool and much better than say a hard disk, is a little behind the curve on this stuff.
posted by sergeant sandwich at 12:08 AM on November 20, 2005

sorry, that sounded a bit snarky. what i should have said is that there already exists technology for writing permanent data at a much greater density. however, the project in question is re-writeable data, and is nothing to shake a stick at. apologies if i came off rude sounding!
posted by sergeant sandwich at 12:19 AM on November 20, 2005

How about quantum dots: "The RAM design is based on a simple 2-dimensional grid of identical memory cells, addressed by row using a QCA decoder. The proposed memory has storage capacity > 1.6 Gbit/cm2." About 1600 bits per micron. Still not great, but closer, maybe.

I don't know what the density would be in a qubit system, but with superpositioning the density could go higher than one bit per electron. Assuming we can invent quantum computers to decouple more than a handful of bits.
posted by Rothko at 1:57 AM on November 20, 2005

only vaguely related, but perhaps interesting anyway - in a recent nature there was an article describing how doping in semiconductors is reaching a critical problem. currently doping is achieved by allowing "impurities" to randomly diffuse into silicon, but devices are reaching a size scale comparable to the distance between each atom of the doping substance. that means that the performance of individual devices will start to vary significantly. to correct this they're going to have to start doping silicon in a fixed way, rather than at random. i have no idea how they'll do this (i didn't finish the article...), but it seems like it could drive the development of the same technologies used to "write" these atomic scale texts, since you want to place dopant atoms at specified locations.
posted by andrew cooke at 5:45 AM on November 20, 2005

For maximum density, e-beam is the way to go. Focused Ion Beam would probably work too. Not sure if it would be justified speedwise to write a complete encyklopedia with any of them, as they are very expensive and there's usually a queue.

The superposition of states in a qubit doesn't hold any extra extractable information. That is, we can only read out a qubit as 0 or 1 in the end.
posted by springload at 5:57 AM on November 20, 2005

After reading the essay, it sounds like Feynman's challenge wasn't to store the data of the text in an area the size of a pinhead, but to actually reproduce the text in writing on the head of a pin, i.e., if the text were enlarged, it would be human-readable.

Not actually answering the question: In 1990 IBM printed their logo with 35 xenon atoms on a nickel surface at super-cooled temperatures. This is probably about as small as we're ever going to get for actually printed text-- anything smaller would run into quantum effects.
posted by justkevin at 8:22 AM on November 20, 2005

In this speech, Feynman offers a prize of $1,000 to anyone who can write a page of a book at 1/25000 of its original size, which is the reduction required to fit the entire encyclopedia on a pinhead.

If we assume a 3mm square for each letter, that drops us down to 12 nm square letters, or 4nm pixels for writing with those trendy pixel fonts.

Intel mass-produces CPUs flawlessly with 900 nanometer transistors, billions of them. People have etched things at the atomic scale.

I know, I've read news articles about moving individual atoms around with an electron microscope, but I'm having trouble googling them up.
posted by Paris Hilton at 8:34 AM on November 20, 2005

Paris, that'd be 90 nm. 900 nanometer is ancient. In fact I think some of the Intel mass-produced CPUs are in at least 65 nm.
posted by substrate at 3:53 PM on November 20, 2005

sergeant sandwich: I'm curious about the reason for your question. If no one has done this before, do you intend to be the first? It would certainly be an interesting task, and good PR for the science of small things.
posted by springload at 6:06 PM on November 20, 2005

sergeant sandwich: I'm curious about the reason for your question. If no one has done this before, do you intend to be the first?

well, given that the stanford group did the proof-of-principle with Dickens, i don't think there's any scientific benefit to carrying the project out all the way to the EB. our lab is getting an e-beam litho system soon, so i think it's something we'd be capable of.

honestly, as far as i can tell, the real difficulty here would be scanning all 20,000 pages (or whatever) of the encyclopedia and automating the imaging process. it's more of an engineering/control problem at this point.

i don't see there being much real benefit to it other than PR; it would surely be the sort of thing that one could show in a museum, but is it worth spending half a year on when we could be doing actual science? i am certain feynman would say no.
posted by sergeant sandwich at 6:21 PM on November 20, 2005

EB could probably be convinced to provide digital copies. The PR value for them would be big enough. Then it wouldn't take very long time, and be a good exercise in e-beam manufacturing (Even with good modern machines, the 8nm dots are a stretch). Otherwise one could try it with Wikipedia. That would be kind of a corny statement though, and probably require a cushion of needles.

footnote: I think science for the museum is worth a bit of effort. Scientists should be serious about educating and fascinating the people who are not part of the game. Failure to do so leaves room for crap like intelligent design theories and lack of interest from the society to provide scientific funding.
posted by springload at 7:02 PM on November 20, 2005

I think science for the museum is worth a bit of effort. Scientists should be serious about educating and fascinating the people who are not part of the game.

word - this is a good point. something to put on the to-do list then, after the research i'm supposed to be doing gets done. in the absence of an answer to the actual question, i'm going to mark this as the best one.
posted by sergeant sandwich at 7:40 PM on November 20, 2005

In my copy of "The pleasure of finding things out" (ISBN 0-14-029034-6) the essay has a following note about the prizes. Short summary -

The bet about the small motor was claimed less than a year later by Bill McLellan. It was supposedly disappointing to Feynmann as it used no new technology.

The encyclopedia bet was more or less filled 26 years later, by a grad student named Tom Newman when his supervisor was out of town. He wrote the first page of "A Tale of Two Cities" at a 1/25,000 scale using electron beam lithography.
posted by scodger at 10:21 PM on November 20, 2005

Doh, always read the more inside.
posted by scodger at 10:22 PM on November 20, 2005

Good luck! If you don't get around to it I might look into the possibility myself.
posted by springload at 6:20 AM on November 21, 2005