Molecular manufacturing, on the other hand, represents a "bottom-up" technology. Desired products will be built directly by "assembler" machines, molecule by molecule, making larger and larger objects with atomic precision. The results of such processes may also be very small or very large, much as biology builds both micron-sized bacteria and 100-meter tall sequoia trees. However, since assemblers add matter only where it is intended, little need be removed and hence there may be minimal waste during the process. By guiding with precision the assembly of molecules and supra-molecular structures, such a manufacturing system could construct an extraordinarily wide range of products of unprecedented quality and performance.

In 1959, the Nobel physicist Richard P. Feynman¹⁵⁶ observed that:

"The principles of physics, as far as I can see, do not speak against the possibility of maneuvering things atom by atom. It is not an attempt to violate any laws; it is something, in principle, that can be done; but, in practice, it has not been done because we are too big . . . Ultimately, we can do chemical synthesis."

"A chemist comes to us and says, 'Look, I want a molecule that has the atoms arranged thus and so; make me that molecule.^' The chemist does a mysterious thing when he wants to make a molecule. He sees that it has got that ring, so he mixes this and that, and he shakes it, and he fiddles around. And, at the end of a difficult process, he usually does succeed in synthesizing what he wants.^"

"But it is interesting that it would be, in principle, possible (I think) for a physicist to synthesize any chemical substance that the chemist writes down. Give the orders and the physicist synthesizes it. How? Put the atoms down where the chemist says, and so you make the substance. The problems of chemistry and biology can be greatly helped if our ability to see what we are doing, and to do things on an atomic level, is ultimately developed a development which I think cannot be avoided."

Nearly 40 years after Feynman's famous "Plenty of Room at the Bottom" speech, and a decade after Drexler's original proposal⁸¹⁰ for a bottom-up approach to machinebuilding using molecular assemblers, Nobel chemist Richard Smalley also largely agreed that this objective should prove feasible. Noted Smalley:²³⁸⁹ "On a length scale of more than one nanometer, the mechanical robot assembler metaphor envisioned by Drexler almost certainly will work..."

Many skeptical questions arise when one first encounters the ideas of molecular nanotechnology and molecular assemblers (Section 2.4.2). It is useful to keep in mind the proven feasibility of such systems in the biological tradition (Section 1.3.2.1) over billions of years of natural evolution. In one sense, molecular nanotechnology will be a refinement and expansion upon how nature works at the molecular scale. Nature's examples, such as human beings, certainly answer the most basic skeptical questions, such as: Can macroscopic objects be built from molecular scale processes? (Yes, thanks to cellular replication.) Are molecular objects stable? (Of course; the human population alone contains >10²⁹ reasonably stable and quite functional ribosomal "nanomachines"). Observes Nobel chemist Jean-Marie Lehn:⁷⁶⁵ "The chemist finds illustration, inspiration and stimulation in natural processes, as well as confidence and reassurance since they are proof that such highly complex systems can indeed be achieved on the basis of molecular components."

What about quantum effects? The uncertainty principle makes electron positions somewhat fuzzy, but the atom as a whole has a comparatively definite position set by the relatively great mass of the atomic nucleus. The quantum probability function of electrons in atoms tends to drop off exponentially with distance outside the atom, giving atoms a moderately sharp "edge". Mathematically, the positional uncertainty of a single carbon atom of mass m_C = 2 x 10^-26 kg bound in a single C-C bond of stiffness¹⁰ k_C = 440 N/m may be crudely estimated from the classical vibrational frequency n_C = (k_C/m_C)^1/2 = 1.5 x 10¹⁴ Hz. This sets the zero-point vibrational bond energy E_C = h n_C / 2 = 4.9 x 10^-20 J = k_C DX_C² / 2 where h = 6.63 x 10^-34 J-sec (Planck's constant) and DX_C ~ 0.015 nm is the maximum classical amplitude of the bound carbon atom (roughly the same as the 3 dB point for the gaussian wavefunction, notes J. Soreff). Thus DX_C is just ~5% of the typical atomic electron cloud diameter of ~0.3 nm, imposing only a modest additional constraint on the fabrication and stability of nanomechanical structures. (Even in most liquids at their boiling points, each molecule is free to move only ~0.07 nm from its average position.)²⁰³⁶

How about the effects of Brownian bombardment on nano-machines? Describing a nanomechanical component as a harmonic oscillator embedded in a gas, Drexler¹⁰ notes: "At equilibrium, an impinging gas molecule is as likely to absorb energy as to deliver it, and so molecular bombardment has no net effect on the amplitude of vibration. How a system is coupled to a thermal bath can affect its detailed dynamics, but not the statistical distribution of dynamical quantities." Nanomachines must also obey the laws of thermodynamics. Nanorobots cannot be used as a "Maxwell's demon"^2349,2350 -- energy must be expended to do useful work; there can be no free lunch.²⁶¹¹

Will high-energy radiation damage nanomachines? Radiation can break chemical bonds and disrupt molecular machines.⁸ The annual failure rate for a properly designed ~1 micron³ molecular machine (containing many billions of atoms) can be made as low as several percent.¹⁰ This estimate is derived from a consideration of all known thermal, photochemical, radiation, and other damage mechanisms, but the dominant error mechanism that appears difficult to substantially reduce is damage caused by background radiation. Failure rates are not zero, but are nonetheless remarkable by today's standards (Chapter 13), especially given error detection and correction at the module level.

During nanodevice fabrication, the problems of crossbonding and reactive intermediates pose additional constraints that must be dealt with using appropriate assembler designs. Specifically, assemblers with rigid positioning components operating in vacuo can restrict reactive intermediates and molecular tool tips to desired locations with a precision of at least ~0.1 nm (Section 9.3.1.4), or slightly less than an atomic diameter. Within the confines of an evacuated or "eutactic" assembler workspace, the location of every structural atom may be known to within the uncertainty created by thermal noise. Internal structural components may possess relatively high stiffness, so positional uncertainty caused by thermal noise may be small. In the classical analysis, positional variance DX² = kT / k_s, where k = 1.381 x 10^-23 J/kelvin (Boltzmann's constant), T is temperature in kelvins, and k_s is component stiffness, typically 10 N/m for nanometer-scale diamondoid components,¹⁰ hence the standard deviation of component position is ~0.02 nm at room temperature, or ~0.01 nm at cryogenic (e.g., LN₂) temperatures. By employing sufficiently stiff mechanisms the position of every structural atom in the system can be known to within a fraction of an atomic diameter with high reliability and without the need for explicit positional sensing.* Steric tool hindrance -- in molecular manufacturing, the conical volume occupied by a tool tip or manipulator platform -- is a technical design problem that may be overcome by using sufficiently stiff tools or multi-tip tools, working in vacuo. Additionally, while some synthetic methods might involve the manipulation of individual atoms by appropriate tools, such as the hydrogen abstraction tool which removes a single selected hydrogen atom from a diamondoid surface (Section 2.3.3), other reactions will involve small clusters of atoms or larger molecular components (Section 2.4.1).

---

* Medical nanorobots may be built to atomic precision, but once built and deployed such devices often will not perform molecularly precise movements during normal functions and may frequently rely upon sensor-based motion control; Section 9.3.3.

---

What about friction²⁸⁹⁶ and wear among nanomechanical components? Properly designed molecular machines lack wear mechanisms, although other damage mechanisms remain.²²⁴³ Dry bearings with atomically precise surfaces can have negligible static friction³²⁴³ and wear, and very low dynamic friction or drag.¹⁰ Nanomechanical components are better viewed as moving smoothly in a force field than as sliding subject to friction. For example, total drag power dissipated by an isolated molecular sleeve bearing of radius 2 nm spinning at 1 MHz is dominated by band-stiffness scattering, amounting to 0.000004 picowatts (pW) (Eqn. 6.4), very small compared to the 1-1000 pW power draw that will be typical of micron-size medical nanorobots (Section 6.5.3). The strong precise surfaces of nanomachines experience no change during a typical operational cycle, hence zero wear. Within the single-point failure model (Chapter 13), the first step in a wear process (e.g., a dislocated atom) is regarded as fatal, hence cumulative wear plays no role in determining device lifetimes.¹⁰ In eutactic nanomachines, contaminants of all kinds are rigorously excluded. The equivalent of wear particles cannot appear until the device has failed. On this scale, repulsive fields provide "lubrication" and an oil molecule would be a contaminant object, not a lubricant.

The general conclusion is that molecular nanotechnology violates no physical laws. In 1998, it was generally accepted that molecular nanotechnology will be developed, although there remains some disagreement about how long it will take. To progress from today's limited capabilities to the complex nanorobots described later in this trilogy will clearly require a great deal of research and development effort. Still, the proper question is no longer "if" but "when." This Chapter presents a few of the many possible technical paths leading to a working molecular assembler that were being discussed in 1998. The reader is strongly urged to consult the most current literature for the latest results. Section 2.2 briefly summarizes the conventional top-down approaches to nanotechnology. Section 2.3 describes several possible direct bottom-up pathways to molecular manufacturing including biotechnology, supramolecular chemistry, and scanning probes. Gimzewski and Joachim³²⁰⁰ note that "bottom-up approaches to nanofabrication may one day compete with conventional top-down approaches in providing nanotechnologies for the next millenium. Top-down refers to increasing miniaturization through extension of existing microfabrication schemes, [whereas] the bottom-up scenario is one of ever-increasing complexity at the molecular level while maintaining control on an atom-by-atom basis". Section 2.4 surveys molecular mechanical components and design work on molecular assemblers and molecular manufacturing systems from the perspective of 1998.