Manufactured products are made from atoms. The properties of those products depend on how those atoms are arranged. If we rearrange the atoms in coal we can make diamond. If we rearrange the atoms in sand (and add a few other trace elements) we can make computer chips. If we rearrange the atoms in dirt, water and air we can make potatoes.
Todays manufacturing methods are very crude at the molecular level. Casting, grinding, milling and even lithography move atoms in great thundering statistical herds. It's like trying to make things out of LEGO blocks with boxing gloves on your hands. Yes, you can push the LEGO blocks into great heaps and pile them up, but you can't really snap them together the way you'd like.
In the future, nanotechnology will let us take off the boxing gloves. We'll be able to snap together the fundamental building blocks of nature easily, inexpensively and in most of the ways permitted by the laws of physics. This will be essential if we are to continue the revolution in computer hardware beyond about the next decade, and will also let us fabricate an entire new generation of products that are cleaner, stronger, lighter, and more precise.
It's worth pointing out that the word "nanotechnology" has become very popular and is used to describe many types of research where the characteristic dimensions are less than about 1,000 nanometers. For example, continued improvements in lithography have resulted in line widths that are less than one micron: this work is often called "nanotechnology." Sub-micron lithography is clearly very valuable (ask anyone who uses a computer!) but it is equally clear that conventional lithography will not let us build semiconductor devices in which individual dopant atoms are located at specific lattice sites. Many of the exponentially improving trends in computer hardware capability have remained steady for the last 50 years. There is fairly widespread belief that these trends are likely to continue for at least another several years, but then conventional lithography starts to reach its limits.
If we are to continue these trends we will have to develop a new manufacturing technology which will let us inexpensively build computer systems with mole quantities of logic elements that are molecular in both size and precision and are interconnected in complex and highly idiosyncratic patterns. Nanotechnology will let us do this.
When it's unclear from the context whether we're using the specific definition of "nanotechnology" (given here) or the broader and more inclusive definition (often used in the literature), we'll use the terms "molecular nanotechnology" or "molecular manufacturing."
Whatever we call it, it should let us
Get essentially every atom in the right place.
Make almost any structure consistent with the laws of physics that we can specify in molecular detail.
Have manufacturing costs not greatly exceeding the cost of the required raw materials and energy.
There are two more concepts commonly associated with nanotechnology:
Clearly, we would be happy with any method that simultaneously achieved the first three objectives. However, this seems difficult without using some form of positional assembly (to get the right molecular parts in the right places) and some form of massive parallelism (to keep the costs down).
The need for positional assembly implies an interest in molecular robotics, e.g., robotic devices that are molecular both in their size and precision. These molecular scale positional devices are likely to resemble very small versions of their everyday macroscopic counterparts. Positional assembly is frequently used in normal macroscopic manufacturing today, and provides tremendous advantages. Imagine trying to build a bicycle with both hands tied behind your back! The idea of manipulating and positioning individual atoms and molecules is still new and takes some getting used to. However, as Feynman said in a classic talk in 1959: "The principles of physics, as far as I can see, do not speak against the possibility of maneuvering things atom by atom." We need to apply at the molecular scale the concept that has demonstrated its effectiveness at the macroscopic scale: making parts go where we want by putting them where we want!
One robotic arm assembling molecular parts is going to take a long time to assemble anything large — so we need lots of robotic arms: this is what we mean by massive parallelism. While earlier proposals achieved massive parallelism through self replication, today's "best guess" is that future molecular manufacturing systems will use some form of convergent assembly. In this process vast numbers of small parts are assembled by vast numbers of small robotic arms into larger parts, those larger parts are assembled by larger robotic arms into still larger parts, and so forth. If the size of the parts doubles at each iteration, we can go from one nanometer parts (a few atoms in size) to one meter parts (almost as big as a person) in only 30 steps.