For those unfamiliar with Clarke's book, the universal replicator would be able to scan any object with one unit, and then use another unit (or multiple units) to replicate it down the atom. Or, I suppose it could produce an item from a computer blueprint, like the machines now in use.
Here is one of the companies that makes these things, which they call an "FDM Rapid Prototyping 3 dimensional printer": http://www.stratasys.com/ Anyway, what Clarke envisions is FAR more advanced than this, putting it hundreds of years in the future, or perhaps thousands of years. For one thing, today's machines only use a few different materials, whereas the future Universal Replicator would use any atom of any element, putting each atom in the right molecule, and each molecule in the right spot. It would assemble these atoms at lighting speed, so that the object, however large, would be fabricated in a few seconds or minutes. IBM and others have used STM to move and rearrange individual atoms on the surface of materials. That is far different than moving them into a 3-dimensional matrix. Plus, I suppose that at the best possible speed of the IBM device, and even assuming you use parallel STM on the same object, building up a small 3-dimensional object would probably take thousands of years. It may not even be physically possible to move atoms around with this kind of precision and speed, but I think Clarke would blithely assume that some method will be found. As was pointed out here years ago, cellular machinery controlled by DNA is capable of moving individual atoms of different elements into complex shapes with great speed and precision. So perhaps something like this could be done, to manufacture objects from organic materials. But what we really want are objects made of any material, assembled much faster than cellular processes allow. The other day I read about people moving tiny objects with lasers, without physical contact, with a so-called "optical tweezer." I thought "ah, ha! Just what we need for a replicator." But unfortunately the technique works because the lasers warm the air around the object or ablate the material and the reaction pushes the sample around. That does not sound like the level of precision or control we need, plus you can't ablate a single atom. And I suppose we're going to have to do this in a vacuum to avoid contamination. Here is something about that in Popular Science (which beats Wikipedia, okay?): http://www.popsci.com/science/article/2010-09/aussie-researchers-use-working-tractor-beam-move-objects-5-feet - Jed
