|author||K Eric Drexler|
|publisher||John Wiley & Sons|
|summary||Dr Nano answers his critics with a technical treatise on nanotech.|
About the reviewer
Chris Worth is a web creative director and nanotech junkie based in Paris. You'll find other ramblings on technology, literature, and red hot asian babes at chrisworth.com. He's looking for geeks to build a subversive website for fun and profit, supported by some of the world's top creatives and assorted rich bastards; email him at firstname.lastname@example.org if you're interested.
The Scenario: bringing researchers together
So you think you know what nanotech is, huh? Maybe you read a book by Neal or Greg or William and dreamed of custom-built computing molecules blanketing cities a billion deep, of patterned flesh singing a song of networked biosentience, of hundred-storey polycarbon structures reaching skywards into the electric neon night. Maybe the concept seduced you into Unbounding the Future and its Lilliputian expeditions across molecular landscapes, or you notched up to Engines of Creation and its talk of assemblers and replicators in pages nude of math. I read them too. And they're good, believe me. But to really know nanotech, to bite through the soft pop-sci underbelly and champ down on its hard skeleton of applied physics, you've got to read Nanosystems .
Nanosystems: the first technical treatise on nanotechnology
Nanosystems, by K Eric Drexler, is the real deal: the first textbook on molecular nanotechnology. It's full of greek equations and exponential graphs and globular diagrams that'd scare your chemistry professor, walled in by dense paragraphs of dry prose that'll make your teeth itch. But somehow it's readable - because the book has a broader purpose that goes beyond Potential Energy Surfaces or spatial Fourier transforms or Born-Oppenheimer approximations. That purpose is to bring together researchers from different fields, to show them how their expertise fits into the broad patchwork of nanotechnology. And that means it's readable for any motivated geek, because Drexler assumes no in-depth knowledge of any one field; concepts are explained from first principles and many equations are derived step-by-step. In a nutshell: if you get C, you can get Nanosystems.
So that's the purpose of Nanosystems: to bring disparate researchers into a single conceptual framework and make nanotech a collaborative effort. But just what is nanotech? First, let's define what it isn't - because nanotech discussions often give out more heat than light. Like transgenic crops and human cloning, vast swaths of the argument would disappear if everyone understood the principles.
Nanotech: so what the hell is it?
First, it's not necessarily about small things; the nano prefix refers to precision at the molecular scale, not the size of the finished article. A rocket motor built bottom up from component atoms one by one is molecular nanotechnology; a train of tiny gears built top-down by hewing away at a silicon surface is not. Second, nanotechnology won't turn lead into gold; elements are defined by atomic nucleii, and nanotech isn't interested in the nuclear forces. Third, it isn't a cure for all the world's problems; hatred and bigotry are separate issues no technology can solve. Fourth, there won't be any day when sci.nanotech explodes with cries of "it's here!"; since it'll be the result of research across multiple disciplines, nanotech will arrive in fits and starts.
And finally, on the biggest misunderstanding of all: no, nanotech isn't impossible. The laws of physics don't prevent nanotech happening; in fact, they emphatically make it possible. (Mr Heisenberg isn't half the troublemaker you think he is.) Yes, there's a tad too much hero worship and holy rollerism surrounding the good-natured and approachable Dr Drexler. And that's given rise to some negative column inches by Scientific American's Gary Stix and Nature's David Jones (neither of whom backed up their assertions). But catcalls and hype don't change basic physical principles; nature doesn't give a damn how loud we shout. And since Nanosystems's first printing in 1992, even Drexler's most loudmouthed critics haven't found any showstopping fault with it.
But back to what matters: what is nanotech? Fundamentally, it's about that bottom-up capability: getting every atom where you want it. Once you can get every atom where you want it, you can build machine systems from the bottom up with atomic precision. Once you can build bottom up, you can build machine systems capable of making perfect copies of themselves, as ribosomes do with DNA. And once your machine's made a perfect copy of itself, you can tell those two to build another, and those four to build four more, and so on, meaning that in a day or two you've got enough to start doing serious work. That bottom up capability of "molecular manufacturing" - which Drexler defines as "the construction of objects to complex atomic specifications using sequences of chemical reactions directed by nonbiological molecular machinery" - would lead to a new world of wealth and abundance. And Nanosystems is about reaching it.
The book's structure
Inside the blue-white cover with tantalising schematics of a molecular sorting rotor, atomic-scale bearing, and a robot arm with the 50 nanometer legend, the book's 556 pages split into three parts: "Physical principles", "Components and systems", and "Implementation strategies". What it does, what to do it with, and how to get there, backed up by 450 equations. Resist the urge to skip chapters until you've skimmed the whole book once; it has a developing structure that rewards a bit of linearity. The preface - with its famous first line "Manufactured products are made from atoms, and their properties depend on how those atoms are arranged" - sets the scene, with notes on why it's reasonable to predict tomorrow's technology with today's. ("Our ability to model molecular machines has far outrun our ability to make them...") But the meat starts with the intro:
"The following devices and capabilities appear to be both physically possible and practically realizable:
Programmable positioning of reactive molecules with ~0.1nm precision
Mechanosynthesis at >10^6 operations/device.second
Mechanosynthetic assembly of 1kg objects in <10^4 s
Nanomechanical systems operating at ~10^9 Hz
Logic gates that occupy ~10^-26m (~10^8 micro^3)
Logic gates that switch in ~0.1ns and dissipate <10^-21J
Computers that perform 10^16 instructions per second per watt
Cooling of cubic-centimeter, ~10^5W systems at 300K
Compact 10^15 MIPS parallel computing systems
Mechanochemical power conversion at >10^9W/m^3
Electromechanical power conversion at >10^15W/m^3
Macroscopic components with tensile strengths >5*10^10GPa
Production systems that can double capital stocks in <10^4s"
Yeah, I was drooling too. And just a few pages further in Drexler whacks us with a nanomechanical product: a bearing with shaft and sleeve in 6- and 14-fold prime symmetry to keep it turning. It's made of carbon with the odd silicon and oxygen atom to round it out, dangling bonds capped with hydrogen, and is made of just 206 atoms. Of course it can't be built yet, but the mind boggles anyway. As it should: this diagram is a teaser for the whole book.
The rest of the intro is comparisons: how conventional solution-phase chemistry and mechanosynthetic chemistry are different, how characteristics of different approaches differ at the nanoscale, how the carbon structures described in Nanosystems are just a subset of all covalently-bonded structures, and the scope of the book. Read this: there's no sci-fi here, no what-ifs, no assuming-thats. Nanosystems is about what's possible given today's understanding of how molecules behave - as such, it's more conservative than many papers you'll see in Nature.
Chapter by chapter
Part I - Physical Principles - is the hardest, squashing a physics course into 230 pages. Ride the hump, guys; no pain, no gain. Chapter 1 takes you down into the molecular world, exploring where classical physics scales down and where it doesn't; chapters 2 and 3 get down and dirty with how molecules are shaped and how they behave when pushed. Chapter 5 is for Heisenberg fans, explaining how thermal uncertainty's a far bigger problem than quantum uncertainty at these scales, while 6 and 7 explore how nanomachine designs will be debugged, going into problems of error-checking and heat death. So far, so painful. Work with it.
It's not until chapter 8 that Drexler starts talking about "real" nanotech: mechanosynthesis. This is 1AM stuff when you know you should be putting the book down for the night but can't. You'll be reaching for the Jolt without caring about work tomorrow. There's still plenty of alkenes and alkynes and tensile bond cleavage and Pi-bond torsion talk here, but the graphs stop for a moment as Drexler deals with what later got called "fat finger" and "sticky finger" problems - how to make your reactive tool molecule slim enough to cause one reaction with a target molecule without it getting the wrong one, and how to make sure the reaction happens when you want it to. And this chapter introduces carbon, everyone's favourite element.
Carbon is one seriously cool atom. Tetrahedral covalent carbon - diamond - is a hundred or so times stronger than steel, and its components atoms are everywhere. They do have to be joined together in a precise pattern; that's why diamond is rare today, and why p.241 includes a diagram of adding two ethyne molecules to another hydrocarbon to model a step in diamondoid formation. Peppered with other common elements like oxygen, fluorine, chlorine, hydrogen, silicon, sulfur, phosphorus, and nitrogen, carbon can be assembled into tough, stiff structures with almost any mechanical or electronic property we want. And carbon molecules are surprisingly easy to model accurately on a computer. That's why Nanosystems devotes itself principally to carbon structures.
On to part II, Components and Systems. Chapter 9 kicks off with the difference between housings and moving parts, and answers one criticism levelled at Drexler: you can't extrapolate to the nanoscale from the macroscale. With a grab-bag of molecular rods and strained-shell carbon bearings, Drexler shows where we can and where we can't. Chapter 10 does the same for moving parts, salting in what happens when two structures start interacting with each other: there are some tasty diagrams of molecular gears, rollers, belts and cams here, but watch out for the graphs and equations.
By chapter 11 the components start coming together as complete systems instead of odd toys, worm gears inserted between tube sections and drive rings threaded onto toroidal housings. Some of the drawings look clunky and Victorian to our silicon-bred eyes, until you realise the transistors we know and love are huge rough-hewn logs at this scale and gravity and friction aren't problems in the same way. Nanosystems is about mechanics, not electronics, but a funky electrostatic motor on p.337 blurs the line: at these sizes both approaches are elegant.
It's at chapter 12 that Drexler gets around to computers. Shapes reminiscent of Babbage engines and Jacquard looms parade across the pages in diagrams of rod-logic gate and register apparatus. (Yes, this is the chapter that inspired a scene in The Diamond Age.) Neal Stephenson got it wrong: this is unlikely to be how we'll build tomorrow's PCs, because Nanosystems is an exploration of engineering techniques, not a recommendation to Intel. The chapter pivots on a finite-state machine built with nanomolecular AND/OR rod logic, with text stating a million-transistor CPU would fit inside a 400nm cube, run at 1GHz, and perform at 10^16 instructions per second. Nanoelectronic designs will be many orders of magnitude faster, but they're outside the scope of this book.
Chapter 13 starts the segue into part III, chunking up to how all these nanomachines can be linked into a complete machine system. A sorting rotor extracts the right molecules from a mix with precisely-shaped reactants attached to a cam; a set of them washes a mix progressively cleaner and cleaner (more feedstock for Neal Stephenson's Diamond Age.) Molecular conveyor belts grab molecules from a toothed gear and take them elsewhere. But the chapter's wow-factor (wow being a relative term in Nanosystems) is the nanomanipulator, a squat robot arm of four million atoms, over a hundred moving parts yet just a hundred nanometers tall. It can pitch, roll and yaw in all six degrees of freedom, snaking up and down and round and round with a train of drives and clutches spliced together with worm gears and intersegment bearings. Imagine this arm reaching out and bonding to a single atom with a reactive tip, rotating that atom away from its surface and depositing it elsewhere. Remember that image, because it's at the core of what nanotech is.
Building on this, chapter 14 describes an exemplar molecular manufacturing system: the holy grail. Another chunk up, it gloms together all the machines described already, into a complete factory for building nanomachines. From single atoms to different parts to convergent assembly to parallel construction, the factory masses less than a kilogram. With a few simple instructions, millions of interacting nanomachines will build products in minutes, blocks of molecular sorting rotors, conveyor belts, and assemblers individually unaware of the big picture but working in parallel like any anthill or beehive. Open another can of Jolt, because you're on the home stretch now.
Part III - on Implementation Strategies - tacks away from what we can build and talks about how to build the things that build them. It turns out there's more than one way to do it. In chapters 15 and 16 Drexler discusses a range of cool STM and AFM scopes for pushing and shoving atoms around, and suggests ways reactive tips on the scanning needle could play with them; since Nanosystem's publication this has started happening in several labs. Biomolecular selfassembly and protein folding are other possible paths to those first primitive tools that can bootstrap us up to covalent-carbon nanotech. Talk of cyclic backbones, crosslinking and rigidity will answer a lot of critics' questions, with a forward- and backward-chaining analysis (a la computer science) "indicates that feasible developmental pathways link our present technology base to the technology base described in Part II." And there, save for a couple of appendices on methodology and related research, the book ends.
So drop the Jolt and fall asleep, because then you can dream - dream of nanotech's infinity of possibilities. And then we can start talking about it. Talking about it the way we talk about Linux, informed by sound technical issues instead of hype and soundbites. Because Nanosystems is a subversive book, subversive the way strong crypto and open source are subversive: developing thanks to the hacker ethic, developing to liberate the masses instead of control them. Published anywhere else, this review'd probably scare people off. But to you, it probably sounds like a challenge. So read Nanosystems. Imagine how ten thousand hyperlinked Slashdotters with a strong understanding of nanotech could influence this technology... and have so much damn fun doing it.
So go on, geek: read Nanosystems . I dare you.
FOOTNOTE: About the Foresight Institute
After first reading Nanosystems in 1996 I became a member of the Foresight Institute, which Eric Drexler and Chris Peterson founded to spread information about nanotech. Foresight works quietly and cost-effectively to influence public policy towards safe, informed development of molecular nanotechnology. (As Gayle Pergamit, Drexler and Peterson's technical writing collaborator, says, it's amazing what two people and a letter to the right office can achieve.) At the conferences it runs for its members you can rub shoulders with writers like Greg Bear, David Brin and Gregory Benford, Valley legends like Doug Engelbart, hackers the stature of Raymond and Gilmore, Old Media types from the New York Times and San Jose Mercury, real nanotechies like Ralph Merkle of Zyvex and Josh Hall of IMM, and of course Drexler and Peterson themselves. And this would take you through one bagel at breakfast. Thanks to Foresight I've learned a lot, made some excellent contacts, and several strong friends. You can learn more at www.foresight.org.
Table of Contents
1. Introduction and Overview
- 1.1 Why molecular manufacturing?
- 1.2 What is molecular manufacturing?
- 1.3 Comparisons
- 1.4 The approach in this volume
- 1.5 Objectives of following chapters
Part I2. Classical Magnitudes and Scaling Laws
- 2.1 Overview
- 2.2 Approximation and classical continuum models
- 2.3 Scaling of classical mechanical systems
- 2.4 Scaling of electromagnetic systems
- 2.5 Scaling of classical thermal systems
- 2.6 Beyond classical continuum models
- 2.7 Conclusions
- 3.1 Overview
- 3.2 Quantum theory and approximations
- 3.3 Molecular Mechanics
- 3.4 Potentials for chemical reactions
- 3.5 Continuum representations of surfaces
- 3.6 Conclusions
- 3.7 Further readings
- 4.1 Overview
- 4.2 Nonstatistical mechanics
- 4.3 Statistical mechanics
- 4.4 PES revisited: accuracy requirements
- 4.5 Conclusions
- 4.6 Further Reading
- 5.1 Overview
- 5.2 Positional uncertainty in engineering
- 5.3 Thermally excited harmonic oscillators
- 5.4 Elastic extension of thermally excited rods
- 5.5 Elastic bending of thermally excited rods
- 5.6 Piston displacement in a gas-filled cylinder
- 5.7 Longitudinal variance from transverse deformation
- 5.8 Elasticity, entropy, and vibrational modes
- 5.9 Conclusions
- 6.1 Overview
- 6.2 Transitions between potential wells
- 6.3 Placement errors
- 6.4 Thermomechanical damage
- 6.5 Photochemical damage
- 6.6 Radiation damage
- 6.7 Component and system lifetimes
- 6.8 Conclusions
- 7.1 Overview
- 7.2 Radiation from forced oscillations
- 7.3 Phonons and phonon scattering
- 7.4 Thermoelastic damping and phonon viscosity
- 7.5 Compression of potential wells
- 7.6 Transitions among time-dependent wells
- 7.7 Conclusions
- 8.1 Overview
- 8.2 Perspectives on solution-phase organic synthesis
- 8.3 Solution-phase synthesis and mechanosynthesis
- 8.4 Reactive species
- 8.5 Forcible mechanochemical processes
- 8.6 Mechanosynthesis of diamondoid structures
- 8.7 Conclusions
Part II9. Nanoscale Structural Components
- 9.1 Overview
- 9.2 Components in context
- 9.3 Materials and models for nanoscale components
- 9.4 Surface effects on component properties
- 9.5 Shape control in irregular structures
- 9.6 Components of high rotational symmetry
- 9.7 Adhesive interfaces
- 9.8 Conclusions
- 10.1 Overview
- 10.2 Spatial Fourier transforms of nonbonded potentials
- 10.3 Sliding of irregular objects over regular surfaces
- 10.4 Symmetrical sleeve bearings
- 10.5 Further applications of sliding-interface bearings
- 10.6 Atomic-axle bearings
- 10.7 Gears, rollers, belts, and cams
- 10.8 Barriers in extended systems
- 10.9 Dampers, detents, clutches, and ratchets
- 10.10 Perspective: nanomachines and macromachines
- 10.11 Bounded continuum models revisited
- 10.12 Conclusions
- 11.1 Overview
- 11.2 Mechanical measurment devices
- 11.3 Stiff, high gear-ratio mechanisms
- 11.4 Fluids, seals, and pumps
- 11.5 Convective cooling systems
- 11.6 Electromechanical devices
- 11.7 DC motors and generators
- 11.8 Conclusions
- 12.1 Overview
- 12.2 Digital signal transmission with mechanical rods
- 12.3 Gates and logic rods
- 12.4 Registers
- 12.5 Combinational logic and finite-state machines
- 12.6 Survey of other devices and subsystems
- 12.7 CPU-scale systems: clocking and power supply
- 12.8 Cooling and computational capacity
- 12.9 Conclusion
- 13.1 Overview
- 13.2 Sorting and ordering molecules
- 13.3 Transformation and assembly with molecular mills
- 13.4 Assembly operations using molecular manipulators
- 13.5 Conclusions
- 14.1 Overview
- 14.2 Assembly operations at intermediate scales
- 14.3 Architectural issues
- 14.4 An examplar manufacturing-system architecture
- 14.5 Comparisons to conventional manufacturing
- 14.6 Design and complexity
- 14.7 Conclusions
Part III15. Macromolecular Engineering
- 15.1 Overview
- 15.2 Macromolecular objects via biotechnology
- 15.3 Macromolecular objects via solution synthesis
- 15.4 Macromolecular objects via mechanosynthesis
- 15.5 Conclusions
- 16.1 Overview
- 16.2 Backward chaining to identify strategies
- 16.3 Smaller, simpler systems (stages 3-4)
- 16.4 Softer, smaller, solution-phase systems (stages 2-3)
- 16.5 Development time: some considerations
- 16.6 Conclusions
- A.1 The role of theoretical applied science
- A.2 Basic issues
- A.3 Science, engineering, and theoretical applied science
- A.4 Issues in theoretical applied science
- A.5 A sketch of some epistemological issues
- A.6 Theoretical applied science as intellectual scaffolding
- A.7 Conclusions
- B.1 Overview
- B.2 How related fields have been divided
- B.3 Mechanical engineering and microtechnology
- B.4 Chemistry
- B.5 Molecular biology
- B.6 Protein engineering
- B.7 Proximal probe technologies
- B.8 Feynman's 1959 talk
- B.9 Conclusions