© 1999 Robert A. Freitas Jr. All Rights Reserved.

Robert A. Freitas Jr., Nanomedicine, Volume I: Basic Capabilities, Landes Bioscience, Georgetown, TX, 1999

2.2 Top-Down Approaches to Nanotechnology

Nanotechnology was first proposed by Nobel physicist Richard Feynman in December 1959, in a talk¹⁵⁶ in which he also issued a seemingly "impossible" challenge to build a working electric motor no larger than a 1/64th-inch (400-micron) cube, backed by a $1000 prize to spur interest in the new field. Just 11 months later, engineer William McLellan had constructed a 250-microgram 2000-rpm motor out of 13 separate parts and collected his reward.^300,355 (McLellan's entire motor is only as big as the period at the end of this sentence.)³⁵⁵ In 1995, a $250,000 Feynman Grand Prize (Section 2.4.2) was made available, this time sponsored by the Foresight Institute, for any engineer who could build the first programmable nanometer-scale robotic arm. How long will it take for history to repeat?

In his famous 1959 talk, Feynman proposed the prototypical top-down strategy for building complex nanomachinery -- essentially a completely teleoperated machine shop, including mills, lathes, drills, presses, cutters, and the like, plus master-slave grippers to allow the human operator to move parts and materials around the workshop.

To build a nanomachine using Feynman's scheme, the operator first directs a macroscale machine shop to fabricate an exact copy of itself, but four times smaller in size. After this work is done, and all machines are verified to be working properly and as expected, the reduced-scale machine shop would be used to build a copy of itself, another factor of four smaller but a factor of 16 tinier¹⁵⁶ than the original machine shop. This process of fabricating progressively smaller machine shops proceeds until a machine shop capable of manipulations at the nanoscale is produced. The final result is a nanomachine shop capable of reconstructing itself, or of producing any other useful nanoscale output product stream that is physically possible to manufacture, using molecular feedstock.

As Feynman describes the process:¹⁵⁶

"When I make my first set of slave hands at one-fourth scale, I am going to make ten sets. I make ten sets of hands, and I wire them to my original levers so they each do exactly the same thing at the same time in parallel. Now, when I am making my new devices one-quarter again as small, I let each one manufacture ten copies, so that I would have a hundred hands at the 1/16th size....If I made a billion little lathes, each 1/4000th the scale of a regular lathe, there are plenty of materials and space available because in the billion little ones there is less than two percent of the materials in one big lathe. It doesn^'t cost anything for materials, you see. So I want to build a billion tiny factories, models of each other, which are manufacturing simultaneously, drilling holes, stamping parts, and so on."

"As we go down in size, there are a number of interesting problems that arise. All things do not simply scale down in proportion. There is the problem that materials stick together by the molecular (van der Waals) attractions. There will be several problems of this nature that we will have to be ready to design for....[But] if we go down far enough, all of our devices can be mass produced so that they are absolutely perfect copies of one another. We cannot build two large machines so that the dimensions are exactly the same. But if your machine is only 100 atoms high, you only have to get it correct to one-half of one percent to make sure the other machine is exactly the same size -- namely, 100 atoms high!"

Progress is being made on Feynman's top-down approach in a relatively new engineering field known as Micro Electro-Mechanical Systems or MEMS, originally an extension of chip-etch technology rather than micromanipulation. Conventional purely-electronic devices fabricated on silicon chips with ~0.2 micron features must be coupled to external systems to produce mechanical effects, but MEMS devices integrate mechanical components directly with electrical circuitry. Thus while microelectronic chips merely route electrons, microelectromechanical systems give the electronics immediate access to control applications, enabling single microdevices to interact directly with the physical world.

Generic microsystems research has proven to be a rewarding activity for two decades, with each successively smaller motor or machine widely publicized.²⁸⁶⁵ The field gathered momentum after the first working micromotors were demonstrated in the late 1980s by groups at Berkeley and MIT. By 1990, tiny electrostatic motors with 100-micron rotors displayed operating speeds of ~250 Hz⁵⁵⁶ and operating lifetimes of ~10⁶ revolutions.²³⁶³ A new Journal of Micromechanics and Microengineering appeared in 1991, followed by the Journal of Microelectromechanical Systems in 1992 and the first MEMS Application Symposium in Tokyo in June 1996. By the mid-1990s, Sandia's Microelectronics Development Laboratory,²³⁵⁶ one of the most prominent vendors of micromachines, could mass-produce planar micromechanical devices smaller than ~1 mm², incorporating motors with gears and cogs each no thicker than a human hair and turning at speeds of >4000 Hz. Wobble micromotors such as those fabricated in silicon by the University of Utah's micromachine lab demonstrated very little friction or abrasion, with measured operating lifetimes in excess of 300 million revolutions.²³⁶⁶ (Diamond coatings could further reduce friction by an order of magnitude over silicon,²⁵⁴⁵ and the wear rate of polycrystalline diamond is known to be ~10⁴ times lower than for silicon.)²⁸⁵²

MEMS research has produced multi-micron-scale accelerometers,^2382-2384 diverse microsensors (e.g., blood pressure microsensors attached to cardiac catheters), microscale cantilevers and jointed crank mechanisms,²³⁸⁰ 5-micron barbs,²³⁷¹ flow microvalves and pressure microtransducers,^544,2373 micropistons and micropumps,²³⁷³ microgear trains,²³⁷⁹ microactuators⁵⁴⁵ and piezodriven micromotors,²³⁸¹ micromirrors and microshutters,^{546,1062,1974} Fresnel lens microarrays,²³⁷⁴ microgyroscopes,¹³⁸³ a 2-mm long combustion chamber suitable for turbine use²³⁷⁷ (making possible a 1 cm³ gas-turbine generator that would deliver 50 W of power or 0.2 N of thrust),²³⁷⁸ and multidevice microsystems²³⁷² that were available customized or off-the-shelf in mass quantities by 1998. Laser beams²³⁷⁵ and ion beams²³⁷⁶ were used to carve ~300-micron-diameter gears with 50-micron wide teeth in solid diamond. Microelectrodeposition and microcontact printing readily produced topologically complex 3-D microstructures with 1-10 micron feature sizes.²³⁶⁴ Colloidal templating yielded 0.71-micron diameter hollow silica spheres with wall thicknesses from tens to hundreds of nanometers,²³⁶⁸ 10-25 nm trenches,²³⁵⁷ or 4-nm thick graphite sheet layers tiled on spherical surfaces ~0.2 microns in diameter.²³⁶⁹ A CAD/CAM desktop micromanufacturing system invented at Lincoln Laboratory in Lexington, MA, machined three-dimensional structures, including spheres, in silicon, using computer-controlled laser/chemical etching of cubic-micron voxels at the rate of ~20,000 voxels/sec.²³⁶⁷

By 1998, smaller (~100 nm) accelerometers were being developed by such companies as Analog Devices in the U.S. and Daimler-Benz in Germany, and electronic control systems such as microswitches and microrelays were available commercially from several companies including Integrated Micromachines of California. Micromachining was a well-developed engineering discipline.²⁶⁹⁶ Westinghouse Science and Technology Center used MEMS sensors to build a miniaturized mass spectrometer. Micro-opto-mechanical systems were under development. Chemists were proposing¹²¹ and building^1222,1228 Integrated Chemical Synthesizers consisting of millimeter- to micron-sized chemical reactors, designed for specific applications, along with associated devices for moving reactant and product streams (pumps), mixing reactants (reaction chambers), analyzing streams, and separating products (separation chambers), constructed conventionally using silicon substrates and photolithographic techniques to form the desired structures but also utilizing the high selectivity of biological molecules such as enzymes and antibodies to fabricate sensors. MEMS had also led to the development of microgrippers that could manipulate individual 2.7-micron polystyrene spheres, dried red blood cells of similar size, and various protozoa.¹²⁶⁷ Market Intelligence Research Corp. of Mountain View, CA, estimated a total worldwide budget for micromachine research and development of $995 million in 1992 and over $3 billion by 1998.³⁵⁷ Systems Planning Corporation of Arlington, VA, projected that the microsystems industry could be worth more than $14 billion annually by 2001.¹²⁵⁹

Some MEMS research followed surprising directions. For example, in the mid-1990s MEMS engineers at the University of Tokyo Department of Mechanical Engineering fabricated an 8-hinge, 8-plate wing microstructure capable of independent flight through the air.^{1573-1576,2353} The structure was fabricated by traditional silicon micromachining techniques and was then released from the wafer surface, with the initially flat sections lifted into flight position using a micromanipulation system with glass probes several microns in diameter. Each wing was ~500 microns square; applying 300 VAC at 10 KHz caused resonant vibration and the flapping amplitude of the wings exceeded 30°.²³⁵³ In another 4-wing configuration,¹⁵⁷⁴ the flapping resonance frequency was ~150 Hz and the device freely flew up and down a thin glass pole when placed in an alternating electromagnetic field. Another bi-wing design made of magnetized nickel coated silicon measuring 2000 x 10,000 microns successfully flew without power supply cables and guides in the manner of a butterfly or mosquito, when an alternating magnetic field strength of >400 oersted and 12 Hz was applied;¹⁵⁷⁶ a similar device with 320 x 800 micron wings also flew in a 300 Hz magnetic field.¹⁵⁷⁵ "If we release the silicon mosquito from the silicon chip, it flies off and we cannot find it again," complained researcher H. Miura. "It's very small, like dust." MEMS "gnat robots" have also been described.¹²⁵⁰

Using electron-beam (e-beam) lithography, in 1997 researchers at the Cornell University Nanofabrication Facility built what they believed were the world's smallest mechanical devices, including a 4-micron-wide Fabry-Perot interferometer and, just for fun, the world's smallest guitar, carved out of crystalline silicon and no larger than a single human cell.²³⁵⁴ The "nanoguitar" was 10 microns long and 2 microns wide, with six strings each ~50 nm (~200 atoms) wide that would probably resonate at >10-100 MHz if plucked. Presentation of the nanoguitar enthralled the public, surprising at least one Cornell researcher (in another lab) who remarked that "anyone with a sufficiently advanced e-beam lithography system could make it. I make 50 nanometer silicon objects routinely. Doesn't seem like a big deal to me, but evidently most people have no idea what we're capable of these days."

Direct-write e-beam lithography uses a tightly-focused beam of electrons, steered by electromagnetic deflectors, to trace out patterns on a wafer resist surface. The electrons chemically alter the resist, which is then etched away, leaving the desired pattern. By 1998, e-beam spot sizes had been focused to less than 1 nm, and 5-nm beam widths were routine, although forward scattering of the electrons imposed a limit of resist exposure resolution of ~10 nm for e-beams²³⁵⁸ and ~5 nm for focused ion beams.¹²⁵⁹ E-beam writing times were very slow -- for example, 80 hours were required to carve a single photolithography mask for a 1G DRAM, though much of this time was required for proximity calculations by the AI (artificial intelligence) software [E.A. Rietman, personal communication, 1999] (a better approach is to use a neural net).³⁰¹² By 1998, an electron interferometer containing atomically smooth mirrors spaced a few atomic layers apart had been fabricated,³¹⁸³ and in 1999, careful electron-energy-loss spectroscopy experiments revealed that silicon dioxide must be at least 4 atoms (~0.7 nm) thick to act as a conventional electrical insulator.³²⁹⁵

Atom lithography, employing a low-energy neutral atomic beam (usually Na or Cr) cooled and collimated by optical molasses, then concentrated at the nodes of the standing wave formed by the superposition of a laser beam and its reflection, was predicted also eventually to allow surface deposition of 5-10 nm features.^{2355,2370,2838-2841} The first directional atomic beam (of sodium) with a ~0.002 radian beamspread and ~femtogram pulses (~26 million atoms, or ~(100 nm)³, per pulse) was demonstrated in early 1999, with one commentator noting that "the longer dream...is atomic holography...[that] could combine beams of atoms to build a 3-D solid object".³¹⁷³ At the same time, another group achieved a continuous stream of rubidium atoms lasting up to ~0.1 sec in duration, with a beam radius potentially as small as ~1 nm.³²⁰⁵ Current-carrying wires have been used to guide cold lithium atoms "just as optical fibers guide light";³¹⁸² the first curved focusing atom mirror also was demonstrated in early 1999.²⁹²⁵ By 1998, the speculative possibility of quantum-mechanical micromanipulation also had been described.⁶

Other approaches offered comparable results. For instance,engineers at the University of Minnesota NanoStructure Laboratory used nanoimprint lithography (NIL) in 1996 to etch out patterns of lines, grooves and circles as small as 25 nanometers in a polymer,²³⁵⁷ and by 1998 NIL had been demonstrated in nanoscale quantum-wire, quantum-dot, and ring transistors with feature sizes below 10 nm.²³⁵⁸ Nanochannel array glasses containing 33-nm capillaries arranged in a 2-D hexagonal close packing configuration at a number density of ~300/micron² had been available since 1992.²³⁶⁵

Thus by 1998, top-down MEMS techniques could in theory already make nanoscale (though not atomically precise) parts. Might these techniques also produce assemblies of such parts, creating complex machines? Very simple mobile robots of ~1 cm³ volume were commonplace,^2361,2362 so for a more challenging demonstration of MEMS' ability to manufacture complete working microrobots, in 1994 Japanese researchers at Nippondenso Co., Ltd. fabricated a 1/1000th-scale working electric car.^2351,2352 As small as a grain of rice, the microcar was a 1/1000-scale replica of the Toyota Motor Corp's first automobile, the 1936 Model AA sedan (Fig. 2.1). The tiny vehicle incorporated 24 parts, including tires, wheels, axles, headlights and taillights, bumpers, a spare tire, and hubcaps carrying the company name inscribed in microscopic letters, all manually assembled using a mechanical micromanipulator of the type generally used for cell handling in biological research (Chapter 21). In part because of this handcrafting, each microcar cost more to build than a full-size modern luxury automobile.

The Nippondenso microcar was 4800 microns long, 1800 microns wide, and 1800 microns high, consisting of a chassis, a shell body, and a 5-part electromagnetic step motor measuring 700 microns in diameter with a ~0.07-tesla magnet penetrated by an axle 150 microns thick and 1900 microns long. Power was supplied through thin (18 micron) copper wires, carrying 20 mA at 3 volts. The motor developed a peak torque of 1.3 x 10⁶ N-m (mean 7 x 10⁷ N-m) at a mean frequency of ~100 Hz (peak frequency ~700 Hz), propelling the car forward across a level surface at a top speed of 10 cm/sec. Some internal wear of the rotating parts was visible after ~2000 sec of continuous operation; the addition of ~0.1 microgram of lubricant to the wheel microbearings caused the mechanism to seize due to lubricant viscosity. The microcar body was a 30-micron thick 20-milligram shell, fabricated with features as small as ~2 microns using modeling and casting, N/C machine cutting, mold etching, submicron diamond-powder polishing, and nickel and gold plating processes. Measured average roughness of machined and final polished surfaces was 130 nm and 26 nm, respectively. The shell captured all features as small as 2 mm on the original full-size automobile body. Each tire was 690 microns in diameter and 170 microns wide. The license plate was 10 microns thick, 380 microns wide and 190 microns high. Nippondenso subsequently used similar manufacturing techniques to build a prototype of a capsule intended to crawl through tiny pipes in a power plant or chemical plant like an inchworm, hunting for cracks. In 1999, three Japanese electronics companies announced the creation of a 0.42-gram, 5-mm long "ant size" robot reportedly able to lift 0.8-gram loads and to move at ~2 mm/sec, as part of the government's ongoing Micro Machine Project.³²⁵⁸

By 1998, micromanipulators with high placement precision had been demonstrated in various laboratories around the world. For example, the Spider-II micromanipulation robot employed bimorphic piezoactuators within a 260-micron cubic work volume and a three-axis gripper placement accuracy of 8 nm.²³⁵⁹ A microassembly robot using stick-and-slip actuators had 5-nm resolution over a 200-nm scanning range with a maximum speed of 4 mm/sec and 0.4% repeatability at 10 KHz, delivering a maximum driving force of 155 mN and rigidity of 6.3 N/micron.²³⁶⁰

In terms of system size, micromachined MEMS devices (typically ~10^-13 m³ in volume) lie exactly intermediate between the world (~10^-4 m³) and the nanoscale world (~10^-22 m³). The medical nanorobots described in this book generally will have dimensions, precisions, and component part sizes ~1000-fold smaller than the Nippondenso microcar, which itself was, in turn, ~1000-fold smaller than a macroscale automobile. It is difficult to see how MEMS techniques involving statistical materials deposition and removal could fabricate parts to better than ~10 nm feature sizes or tolerances, or could position parts during assembly to much better than ~10 nm spatial resolution, nor does it appear likely that MEMS techniques, on their own, can manufacture or position any structure to atomic precision. However, this does not rule out the possible use of MEMS techniques to fabricate crude useful parts for subsequent "finishing" to atomic precision by other techniques, or to facilitate the assembly of nanoscale components (of atomic precision or otherwise) that have been fabricated or subassembled by other means (Section 2.3).

Last updated on 6 February 2003