Center for Responsible Nanotechnolgy (CRN)

Results of Our Ongoing Research

These pages, marked with GREEN headings, are published for comment and criticism. These are not our final findings; some of these opinions will probably change. LOG OF UPDATES

CRN Research: Overview of Current Findings

Powerful Products of Molecular Manufacturing

Overview: Even a primitive diamond-building nanofactory can create products vastly more powerful than today's versions. Electrical power can be converted to motion, and vice-versa, with 10 times the efficiency and about 10⁸ (100,000,000) times more compactly. Computers can be 10¹² times smaller and use 10⁶ times less power. Materials can be about 100 times stronger. This means that most human-scale products would consist almost entirely of empty space, reducing weight, material requirements, and cost. Most of the rest of the product would be structural, easy to design. Even the simplest products could be software-controlled at no extra cost. Manufacturing of prototypes would be quite rapid—a few minutes to a few hours. Because manufacturing and prototyping are the same process, a successful prototype design could immediately be distributed for widespread use. A designer working with a few basic predesigned blocks could design, build, and test a simple product in less than a day. Products with complex interfaces to humans or to the environment—information appliances, automobiles, aerospace hardware, medical devices—would be limited by the time required to develop their software and test their functionality. However, in some fields the high time and money cost of manufacture slows other parts of the development cycle; this effect would disappear. An explosion of new, useful products could rapidly follow the widespread availability of a nanofactory.

A nanofactory is a good way to build large, useful MNT products.	Early efforts toward a molecular manufacturing capability will probably produce something like the nanofactory described in CRN's paper, "Design of a Primitive Nanofactory". A nanofactory is a system for combining nanoscale parts to make large-scale objects. Starting with a basic molecular nanotechnology (MNT) self-duplicating fabricator, a nano-scale device that can copy itself from simple chemicals, a personal nanofactory (PN) can be built in just a few weeks. Then it can start making duplicate PNs, and then products. The first nanofactory will probably be almost as simple as the one analyzed in that paper. The products it produces may not be the best possible, but this analysis is useful to show that MNT products can be at least this powerful. The nanofactory uses a very simple and limited method of building products. Functionality must be contained within small blocks, which must then be fastened together by the trillions to make a product. In this design, the blocks are only 200 nm (nanometers) on a side. However, this is big enough to hold a small CPU or a motor. Larger functionality can be split between blocks—power, force, and signal can all be transferred between the blocks by simple, efficient interfaces. A product that needs a supercomputer or a powerful motor can combine smaller computers or motors. This splitting will add some inefficiency, but not enough to seriously impair the product. As analyzed below, MNT-built machines will be many times better than conventional machines, and products made by dividing the machines among nanoblocks will still be many times better.
Products can be 5 times as strong, 10 times as efficient, and millions of times as compact — or better.	MNT-built diamond products can be at least ten times stronger than steel, 100 times stronger than aluminum or plastic. (Mechanical joints between blocks cost some percentage of pure diamond strength, so nanoblock products may not be as strong as pure diamond). This means that space used for structure in today's products can be left empty or filled with functional machinery. Assuming that computation power scales linearly with speed and transistor count, the "mobile" version of the Pentium 4 uses about six million times more power per computation than a rod-logic nanocomputer. The fastest computer in the world, as of this writing, is the NEC Earth Simulator. It includes 640 8-processor nodes using 20 kW apiece, for a total of 13 MW (ignoring the large crossbar switch). It also includes 10 TB of RAM, and fills a large building. Assuming that the Earth Simulator's power consumption per operation is comparable to the Pentium 4, a comparable massively parallel nanocomputer would require 2 watts. The CPUs would require a volume of 8 million cubic microns, and the memory an additional 3 million cubic microns. The entire computer could fit into a cubic millimeter, so is trillions of times more compact. Motors can be built as small as 50 nm wide. At that scale, they will be electrostatic rather than electromagnetic. When running at high speed, such a motor can convert more than 500,000 watts of power per cubic millimeter at better than 99% efficiency. An electrostatic motor does not require any current flow to maintain magnetic fields, so it becomes even more efficient at low speed and high torque. The motor also functions as a DC generator without modification. Like computers, motors will require essentially no volume in human-scale products. Power can be transferred efficiently by rotating mechanical rods: a rod 100 microns in diameter can transmit 1,000 watts.
Products can be designed in days and distributed in hours.	In many ways, MNT product design will be similar to software design. Products will be designed in a CAD (computer-aided design) system and built in a fully automated factory. Design of the user interface will usually be far more difficult than design of the internal workings. The structure can be specified by simple volume-filling operations; the function (computation and power use) can be likewise simple, but since functional components will be small enough to fit almost anywhere in the product, designers will be tempted to include complicated user interfaces. The time to build a prototype will be measured in minutes or hours, rather than the weeks it can take to produce a mechanical prototype with today's processes. A designer may work on a design, send it to the personal nanofactory, go to lunch, test the product, decide it's not quite right, make a change, and build a second version before quitting time. This will cause significant improvements in the process of developing new products. The process will be far easier, faster, and cheaper than it is today. Once a product is designed, it can be put into production immediately. The same manufacturing process that built the prototype can be used, without modification, to build as many copies as desired. For products designed in terms of volume-filling and functional specification, the product data file can be pretty small—perhaps under a gigabyte for simple products—so it can be transferred to PNs anywhere. Not only the design cycle, but the production and distribution time, will shrink to less than a day.
Products can even be pre-designed.	CAD software for product design can be developed long before personal nanofactories become available. Once design software is available, product designs can be created. Thousands of products can be designed in advance for a hypothetical PN. Most product designs will use "virtual materials" consisting of arrays of nanoblocks, since 200 nm is far smaller than necessary for the precision of most designs. When the real nanofactory is developed, the virtual materials will be translated into nanoblocks. All predesigned products would then be available for immediate production. It's unclear whether this would be worth doing, but it certainly could be done. One benefit is that MNT product designers could be trained in advance. Regardless of whether products are pre-designed before PNs arrive, or designed rapidly after they arrive, it is clear that a nanofactory can rapidly be put to use, replacing a large fraction of the industry, materials, and transportation required for today's industrial infrastructure.

DEVIL'S ADVOCATE —

Won't a nanoblock design be too limiting in what you can build?

A nanoblock-based design throws away almost all possible structures and a large fraction of the theoretical maximum performance. "Virtual materials" design throws away still more. But what's left is still revolutionary. Computer programmers do the same thing. There are eleven simplifications, or levels of abstraction, between the silicon and the screen. The plan described here uses just three levels of abstraction: diamond chemistry, nanoblocks, and virtual materials.

You're making this sound too easy. It can't be that easy in real life.

Using the full capabilities of molecular nanotechnology would be extremely hard. But once we can build a fabricator, the rest—including the personal nanofactory—can use the easiest and most reliable fraction of the possibilities. By the time we can build a fabricator, we'll be able to build thousands of diamond shapes. Nanoblocks only need hundreds of shapes to make thousands of blocks. Products only need hundreds of different types of nanoblocks. Just pick the best designs and throw away the rest.