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

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

2.3.3 Scanning Probe Technology

The third general pathway leading to molecular manufacturing involves a technology known as scanning probe microscopes (SPMs).^{1093,2728-2730} The first of the SPMs was the Scanning Tunneling Microscope (STM) developed in the late 1970s and early 1980s by Gerd Karl Binnig and Heinrich Rohrer at an IBM research lab in Zurich, Switzerland,^2731-2735 earning these scientists, along with Ernst Ruska, the 1986 Nobel in Physics. The STM was initially used as an imaging device, capable of resolving individual atoms by recording the quantum tunneling current that occurs when an extremely sharp conductive probe tip (usually tungsten, nickel, gold, or PtIr) is brought to within about one atomic diameter of an atom, and then adjusting the position of the tip to maintain a constant current as the tip is scanned over a bumpy atomic surface (Fig. 2.24). A height change as small as 0.1 nm can cause tunneling current to double. The tip is connected to an arm that is moved in three dimensions by stiff ceramic piezoelectric transducers that provide subnanometer positional control. If the tip is atomically sharp, then the tunneling current is effectively confined to a region within ~0.1 nm of the point on the surface directly beneath the tip, thus the record of tip adjustments generates an atomic-scale topographic map of the surface. STM tips can scan samples at ~KHz frequencies, although slower scans are used for very rough surfaces. In some modern STMs (e.g., the DI Nanoscope), the sample is moved while the tip is held stationary.

A major limitation of the STM was that it only worked with conducting materials such as metals or semiconductors, but not with insulators or biological structures such as DNA.^1066,2775 To remedy this situation, in 1986 Binnig, Quate and Gerber developed the Atomic Force Microscope (AFM)⁴⁴⁵ which is sensitive directly to the forces between the tip and the sample, rather than a tunneling current. An AFM can operate in at least three modes. In "attractive" or non-contact mode (NC-AFM, 0.01-1 N/m force constant), the tip is held some tens of nanometers above the sample surface where it experiences the attractive combination of van der Waals, electrostatic, or magnetostatic forces. In "repulsive" or contact mode (CAFM, 0.01-1 N/m force constant), the tip is pressed close enough to the surface for tip and sample electron clouds to overlap, generating a repulsive electrostatic force of ~10 nN, much like the stylus riding a groove in a record player. There is also intermittent-contact mode (IC-AFM, 0.01-1 N/m force constant), which is sometimes called "tapping" mode. In any of these modes, a topographic map of the surface is generated by recording the up-and-down motions of the cantilever arm as the tip is scanned. These motions may be measured either by the deflection of a light spot reflected from a mirrored surface on the cantilever or by tiny changes in voltage generated by piezoelectric transducers attached to the moving cantilever arm. Typical AFM cantilevers have lengths of 100-400 microns, widths of 20-50 microns, and thicknesses between 0.4 to several microns. AFM tips may be positioned with ~0.01 nm precision, compressive loads as small as 1-10 pN are routinely measured,^{433,450,2757,2758} and the tips may be operated even in liquids.²⁸⁹⁸

S. Vetter notes that STM technology has also improved, reaching resolutions of ~0.001 nm in the z direction (vertical) and ~0.01 nm in the xy plane, well beyond atomic resolution. The STM remains the instrument with the best resolution. The conducting surface limitation has been overcome in some cases by coating the target with an extremely thin conducting layer, developing tips with multiple electrodes, and coating a conducting substrate with a sample so thin as to allow enough conduction even if the sample is characterized as an insulator in bulk.

By 1998, the growing family of SPMs included at least forty types of instruments and techniques that relied on interactions between a scanned surface and a nearby probe. Different instruments measured different forces and thus could be used to characterize different properties of the surface.²⁷⁵⁸ For example, friction force microscopes (FFMs), magnetic force microscopes (MFMs), shear force microscopes (ShFMs), scanning capacitance microscopes (SCMs), scanning conducting ion microscopes, chemical force microscopes,^2755,2756 and electrostatic force microscopes (EFMs) measured frictional drag or other binding forces. Magnetic resonance force microscopes (MRFM)^2776,2929 used a field generated from a small magnet mounted on the tip of the cantilever arm to probe nuclear magnetic moments in a small region on the surface of the sample, imaging atom types and even detecting the spin of a single electron. By the mid-1990s, AFMs were already a $100 million/year industry²⁷⁷⁷ and SPMs generally were an off-the-shelf technology costing up to $50,000-$500,000 for complete systems, with the whole industry worth up to ~$0.5 billion annually. (Low-performance "homebrew," student,²⁷¹² and science-fair STMs have been built for as little as $50.²⁷⁷⁸)

How might SPMs be used for molecular manufacturing? In the crudest approach, SPMs might be employed as nanoscale milling machines to carve out "nanoparts" from appropriate substrates. For example, Kim and Lieber¹⁶ used an AFM to perform nanomachining operations on a molybdenum trioxide crystal -- applying a 100 nN load at the tip, they milled a triangular-shaped 50-nm planar chunk from the crystal, then slid the part 200 nm across the work surface by pushing it with the AFM tip. Sheehan and Lieber^1737,2779 milled a 50-nm rectangular part and two other parts with concavities complementary to the rectangle by mechanically etching a MoO₃ layer deposited atop an MoS₂ surface by pressing down with an AFM tip at a 50 nN load. Like a numerical-controlled machine tool, the AFM motions could be programmed to perform a series of steps, like automatically carving the chemical formula "MoO₃" into the crystal.²⁷⁷⁹ Nanoparts with features as narrow as 10 nm can be nanomilled in a limited set of materials [P.E. Sheehan, personal communication, 1995]. In theory, large arrays of independently-controlled SPM tips (see below) could be operated in parallel to carve out great numbers of similar objects simultaneously, though such objects would of course not be atomically identical without further "finishing." AFM tips have also been used to bulldoze nanoscale lines and rectangular <10-nm features into nonconducting photoresist,^2752-2754 perform "dip-pen" AFM nanolithography,²⁷⁸⁰ form grooves,²⁷⁸¹ and fabricate a single-electron transistor via an STM nano-oxidation process.²⁷⁸² A carbon nanotube affixed to an AFM tip was used as a pencil to write 10-nm-wide features onto silicon substrates at ~0.5 mm/sec:²⁷¹¹ An electric field issues from the nanotube, removing hydrogen atoms from a hydrogen monolayer atop a silicon base; the exposed silicon surface oxidized, producing narrow SiO₂ tracks.

Perhaps more significantly, SPMs have been employed in an increasingly sophisticated manner to manipulate individual atoms and small clusters of atoms. The possibility of modifying surfaces with scanning probe tips was evident from the earliest years of STM research, since inadvertent contact between tips and surfaces routinely caused such modifications,¹⁰ and the possibility of transferring material between tip and sample had already been discussed.²⁷⁸³ Suggestions for controlled surface modification soon appeared,²⁷⁸⁴ and in 1985 Becker and Golovchencko²⁷⁸⁵ used voltage pulses on an STM tip to pluck a single germanium atom from the {111} surface of a sample. In 1988, J. Foster and colleagues²⁷⁸⁶ at IBM Almaden pinned small organic molecules to a graphite surface by applying a small electrical pulse through an STM. Additional pulses enlarged or erased this molecular feature, and often split the organic molecule into smaller pieces, though not in a controlled manner. Other workers used an STM to dump small clusters of gold atoms on a platinum surface,^2736-2739 clusters of copper atoms on a gold surface,²⁷⁸⁷ and molecules of tungsten carbide from a supply flowing over a surface.²⁷⁸⁸ Others reported positioning carbon monoxide molecules and platinum atoms on platinum surfaces,²⁷⁸ moving clusters and single atoms of silicon across a silicon surface at room temperature,^2746-2751 and fabricating via STM a "molecular corral"²⁷⁸⁹ or "quantum corral"²⁷⁹⁰ -- a ring of atoms so small that the enclosed electrons were forced to exhibit quantum behavior. In 1999, Tomanek and Kral³²⁶⁶ proposed an atomic fountain pen in which a carbon nanotube filled with atoms is electrically induced to release these atoms, one by one every 15 microsec, onto the work surface.

More precise control was achieved by Eigler and Schweizer²⁷⁸ at IBM Almaden in 1989, when they used an STM to position 35 individual xenon atoms on a nickel surface to spell out the corporate logo "IBM" (Fig. 2.25). To accomplish this, a bias voltage was applied to weaken the adsorption of each atom to its nickel substrate, then each atom was dragged, one by one, to the desired locations at a speed of ~0.4 nm/sec²⁷⁹¹ to build a meaningful pattern. It took 22 hours²⁷⁹¹ to make the entire logo, or ~38 min/atom. The experiment had to be carried out at 4 K (i.e., liquid helium temperatures) because the arrangement would have been unstable at room temperature. Numerous similar examples of "atomic graffiti"⁷⁷⁵ followed the IBM experiment, including atoms patterned in the shape of a world map;²⁷⁹² the word "atom"²⁸⁹⁵ and "nanoworld",²⁸⁹⁷ spelled out with atoms, in Japanese; the word "Peace," Einstein's famous equation "E = mc²," and even a reproduction of the well-known Einstein portrait with the scientist's tongue sticking out;³⁰⁰ a sketch of a "molecule man" using 28 individual CO molecules on a platinum surface;²⁷⁹² gold nanoparticles spelling out "USC"²⁷⁹³ and "Zyvex";²⁷⁹⁴ and sulfur atoms removed from a surface to make letters 2 nm high.²⁷⁹⁵ In 1996, Gimzewski's group¹⁷³⁵ used an STM tip to manipulate individual buckyballs along terraces on a grooved copper plate, making a "molecular abacus."

Some variant of an SPM would appear to be the most promising tool for direct molecular manipulation. Once a tip lies within ~1 nm of a surface, the potential barrier can be lowered sufficiently for atoms to be induced from the surface by field evaporation;²⁷⁹⁶ an applied tip voltage ionizes the atom which can then be guided around by the tip. Depending on voltage and separation, atoms, atom clusters, or molecules can be pried out, pushed in, or nudged around on a surface. By 1998, techniques for direct atomic manipulation proceeded at room temperature and ~1000 times faster than the original IBM logo had been assembled. For example, the Nanomanipulator,^2716-2724 an interactive haptic control system created by a group at the University of North Carolina at Chapel Hill, allows near-real-time manipulation of individual gold atoms across a surface using a hand-held master-slave controller that drives an STM probe while the position of the atom is displayed on a monitoring screen visible to the user. Partially funded by NSF, Silicon Graphics, and TopoMetrix,³¹⁵⁶ the Nanomanipulator is a fully-integrated system that enables investigators to "feel" the interatomic forces as the user pushes atoms around on a surface.²⁷²⁴ In one demonstration, a TV-watching user herded a gold atom, slipping and sliding, into a slot in the planar workplace in about 1 minute; when force feedback was turned off, movements tended to run wild. IBM Almaden has also experimented with removing individual atoms from metal surfaces using an STM hooked into a Virtual Reality Dataglove apparatus.

To manufacture atomically precise parts, it also will be necessary to manipulate covalent bonds at the probe tip. STMs have broken and created chemical bonds; ~1 volt pulses were used to pull atoms out of crystals, binding them to the tip, and then to reinsert them back into the crystal.^2749,2758 In 1995, the first demonstration of catalysis on a nanometer scale was reported by scientists at the Molecular Design Institute at LBL.²⁷⁹⁷ They used an AFM modified to function like an ultrafine-point pen for catalytic calligraphy to change the chemical composition of a material surface one molecule at a time. A surface was prepared as a self-assembled monolayer (SAM) of alkylazide molecules capped with a crown of three nitrogen atoms, then platinum-coated chromium was deposited onto an AFM silicon tip just a few atoms wide. The SAM was soaked with a hydrogen-containing solvent, then scanned by the AFM over a 100 micron² area, with the platinum catalyzing a covalent bonding reaction in which hydrogen was added to the azides, transforming them into amines as revealed by selective fluorescent tags.

Tip technology is an important and fast-growing subspecialty of SPM research. In 1990, Drexler and Foster²¹⁷ suggested the use of custom-made, synthetic proteins to be mounted on the point of an SPM tip. Modified antibodies or specially designed proteins could serve as simple, first-generation "grippers" for binding and manipulating specific molecules, bringing them into position to react with other molecules in a precise and selective way. Functionalized SPM tips have been created to exploit antibody-antigen recognition²⁷⁹⁸ and in the context of chemical force microscopy.^2755,2756 A major convenience for molecular manipulation is that the same instrument that does the chemistry and the molecular orienting can also be used to inspect the results.³²²

In 1996, Smalley's group at Rice University²⁷⁹⁹ attached a single nanotube to the pyramidal silicon tip of an AFM and showed that it was quite robust and could image the bottom of deep trenches inaccessible to conventional tips. By 1998, progress had been made in functionalized carbon nanotubes for use on AFM tips. Most notably, Wong and colleagues²⁸⁰⁰ prepared nanotube tips by oxidation in air at 700°C, burning off all but 2% of the original material and leaving the ends covered with carboxyl (COOH) groups whose chemistry is rich and well understood. Four different kinds of tips were created: (1) the original carboxyl tip, which is acidic; (2) an amine-terminated tip (made by forming an amide bond to one of the amine groups in ethylenediamine (H₂NCH₂CH₂NH₂)), which is basic; (3) a hydrocarbon-terminated tip (made by forming an amide bond to benzylamine (C₆H₅CH₂NH₂)), which is hydrophobic; and (4) a biotin-terminated tip (made by forming an amide bond to a biotin derivative), which shows specific binding to streptavidin. AFM contact forces between tips and selected substrates were shown to be sensitive to pH and to the chemical details of the substrate in ways consistent with the tips' intended chemistry.

Wong's tips had three closely related advantages over previous techniques. First, when a functional group is attached to the apex of an Si₃N₄ or SiO₂ tip, these groups usually adhere to the sides of the tip as well; upon using the tip as a tool, there is a constant hazard that contact with the sides of the tip will alter the workpiece in unwanted places. Unlike these tips built with bulk techniques, the ends of Wong's tips are very different from their sides -- the carboxyl groups are attached only on the ends of the nanotube, not on the sides. Second, Wong's tips have lateral dimensions set by the nanoscale dimensions of nanotubes, not by top-down fabrication techniques. Note the authors: "...the small effective radius of nanotube tips significantly improves resolution beyond what can be achieved using commercial silicon tips....we have recently demonstrated that lateral resolution of <3 nm can be achieved by using COOH-terminated single-walled nanotubes tips".^2740-2742 Third, single-walled nanotube tips are much closer to yielding truly single atom tips with controlled chemistry than any other alternative. J. Soreff notes that a (10, 10) nanotube with a 1.4-nm diameter has just 20 atoms at an open end. Even given a statistical distribution of tubes with varying numbers of carboxyl groups attached to their ends, one should be able to build a ligand which covers the whole end of the tube, yielding a method for ensuring that just one molecule of known structure and orientation was present at the end of the probe.

Theoretical studies have employed molecular dynamics simulations and other techniques to investigate the behavior of mechanosynthetic tool tips.^{10,2760,2761,2764,3250} For example, Drexler¹⁰ proposed using an acetylene radical to perform selective abstractions of hydrogen from a diamond surface (Fig. 2.26A), and this proposal was further studied in Musgrave's ab initio quantum chemistry analysis²⁷⁶² and Sinnott's molecular dynamics modeling of the reaction at room temperature on a diamond {111} surface;^2763,2764 other H-abstraction studies have been done.^10,2907 One possible structure for a hydrogen abstraction tool might be similar to an anthracene whose end had been modified as shown in Figure 2.26B; prior to use, the structure must be activated by removal of the terminal hydrogen.¹¹⁹⁹ Brenner and colleagues²⁷⁶⁴ describe further work modeling the parallel abstraction of several hydrogen atoms from the diamond {111} surface. Such reactions are similar to those involved in the well-studied chemical vapor deposition of diamond,^2765-2769 but adding only positional control and applied mechanical force; more detailed theoretical studies of diamond mechanosynthesis reactions have begun.^2770,3250

R. Merkle²⁶⁰² has proposed a simple mechanosynthetic system capable of fabricating a large class of useful stiff hydrocarbons. Merkle's system includes positionally-controlled hydrogen abstraction and deposition tools (to remove or add hydrogen atoms to a workpiece), carbene and dimer deposition tools (to add one or two carbons to a workpiece), a silicon radical tool (to remove a carbon), and four other specialized tools. Each proposed tool has a molecular handle structure that could be permanently affixed to a suitably functionalized AFM tip (or other appropriate mechanosynthetic nanomanipulator). As each operation is performed on the butadiyne* (a.k.a diacetylene,²⁸⁰¹ C₄H₂) hydrocarbon feedstock in an inert (e.g., vacuum or noble gas) environment, one or more tools are brought to the workpiece at the necessary position and angle, and any necessary force is applied via the AFM. Interestingly, except for a small number of "vitamin parts"¹¹⁵ involving transition metals, Merkle's system can in theory synthesize all of its own mechanosynthetic tools, thus establishing net positive production of system components and illustrating a limited subset-level "quantitative parts closure" as first defined in 1982 by Freitas and Gilbreath¹¹⁵ in connection with self-replicating machine systems (Chapter 14). Merkle acknowledges that several of the proposed reactions involve the simultaneous coordinated action of two, three, or even four positionally controlled tools, but suggests that two 6-degree-of-freedom manipulators plus some combination of appropriate jigs and fixtures should be sufficient in a more parsimonious system. However, steric hindrance of multiple tools at one site also was not explicitly addressed. Another potential problem to be overcome is loss of positional registration as the various functionalized tool heads are changed out at the AFM tip; such tips might be paired with a second positionally-invariant metrology tip on each tool head, allowing precise calibration of workspace location before proceeding back to the workpiece.

* Experimentalists are warned that butadiyne, chemically distinct from the more common molecule butadiene (C₄H₆), can polymerize explosively under some conditions.²⁸⁰¹

Others have suggested fabricating nanoparts by building up structures by stacking layers of individual atoms, perhaps constrained by sacrificial joiners and scaffolding, and honed by differential etching operations.²⁸⁰²

Assuming that "nanoparts" can be fabricated to atomic precision, can SPMs also be used to assemble these nanoparts into working nanomachines? Some preliminary work has been accomplished. For instance, arrays of nanoscale holes have been synthesized that could serve as "workpiece holders." In one experiment, C₆₀ buckyballs were scattered about on a prepared surface, then an AFM pushed the individual spheres into their own little holes, securely seating them pending further processing.²⁶³⁰ The C₆₀-impregnated perforated film was so durable that the buckyballs remained trapped after several months of storage. Similarly, Jung and colleagues in Gimzewski's group²⁸⁰³ synthesized 4-legged porphyrin-based²⁸⁰⁴ molecular "nanoparts" that were specifically designed for easy positioning on a copper surface. Pushing with an STM at room temperature, the authors displaced the ~1.5-nm wide nanoparts in predefined directions and rotated the 4-legged molecules at will, arranging, for example, the parts into a precise hexagonal configuration on the copper surface. The STM has been used to induce, and to view, the rotation of an individual oxygen molecule trapped on a platinum surface, as it is intentionally and stably rotated into any one of three distinct orientations.²⁸⁰⁵ The forces required to slide, or alternatively to roll, a carbon nanotube across a graphitic surface have been directly measured by AFM.²⁷¹⁷

The first known instance of a crude but purely mechanical "nanopart" assembly operation was performed in 1995 by Paul E. Sheehan and Charles M. Lieber.¹⁷³⁷ Two nanoscale parts with rectangular slots and a third 50-nm rectangular sliding "latch" member were milled from a MoO₃ crystal using an AFM, then the rectangular "latch" member was slid repeatedly from one slot to the other using the same AFM, making a three-nanopart reversible mechanical latch. Noted the authors: "The lateral force needed to break the latch, 41 nN, was large considering the small latch contact area, which suggests that relatively robust assemblies can be created with such devices. Such a reversible latch could serve as the basis for mechanical logic gates....Our results...demonstrate the ability to machine complex shapes and to reversibly assemble these pieces into interlocking structures."

In 1998, manipulation of nanoscale parts in the vertical dimension (e.g., normal to the surface plane) had only just begun.^2743-2745 The first three-dimensional structure built out of single "nanoparts" was demonstrated in 1997 by Requicha's group at the USC Laboratory for Molecular Robotics. First, an AFM was used to push 5-nm gold nanoparticles in the third dimension up and over surface protrusions or obstructions -- for example, a 5-nm gold particle was shoved up onto a 2-nm-high protrusion, then slid back off again onto the surface.²⁸⁰⁶ In another series of experiments conducted in air at room temperature with 15-30 nm gold nanoparticles deposited on silicon previously coated with a silane layer, Requicha's group used an AFM to build a simple 3-D pyramidal structure by pushing one gold "nanopart" on top of two others, then off again; the group also used the AFM to rotate and translate a dimer unit formed by two linked "nanoparts".²⁸⁰⁷

Three-dimensional nanoassembly might perhaps be more readily achieved if nanoparts could assist in their own assembly process. R. Merkle [personal communication, 1998] suggests that an SPM could be used to position DNA-tagged molecular building blocks. A building block tagged with a specific single-stranded DNA should preferentially adhere to a surface covered with the complementary DNA. If the building block has a second single-stranded DNA tag which is complementary to single-stranded DNA on the tip of an SPM, and if it is initially attached to the SPM tip by this tag, then it could be positioned by the SPM and attached to the surface. The relative strength of attachments could be adjusted by changing the number of base pairs in the complementary region between the two strands.¹⁰⁶⁶ A related suggestion, due to G. Fahy,³²² is that SPM tips for assembly could be designed with more than one binding site. The existence of at least two binding sites could allow precise orientation of a nanopart, permitting the nanopart to be presented to a desired target site (or another nanopart, possibly attached to another manipulator) at exactly the location and in exactly the orientation desired, assuming that each nanopart can bind to each tip in only one possible orientation. A convenient post-assembly nanopart release mechanism is also required.

In 1997, the Avouris group^2750,2771 at IBM Yorktown Heights demonstrated that the tip of an atomic force microscope (AFM) could be used to control the shape and position of individual multiwalled carbon nanotubes dispersed on a surface. Nanotubes could be bent, straightened, translated, rotated, and (under certain conditions) cut.

In 1998, Zyvex LLC (a developmental engineering company whose goal is to create a molecular assembler capable of manufacturing atomically precise structures²⁷⁹⁴) demonstrated the ability to manipulate carbon nanotubes in three dimensions inside a scanning electron microscope (SEM) having ~6 nm resolution at near-video scan rates. Zyvex's custom piezoelectric vacuum manipulator achieved positional resolutions comparable to SPMs along with the ability to manipulate objects along one rotational and three linear degrees of freedom with 0.1 nm spatial resolution.²⁸⁰⁸ This prototypical device could probe, select and handle limited classes of nanometer-scale objects such as carbon nanotubes under real-time SEM inspection. Carbon nanotubes were attached to commercial atomic force microscope (AFM) tips either by van der Waals forces alone or by "nanosoldering" by a concentrated electron beam from the SEM. Forces applied to the nanotubes could be measured from the cantilevers' deflections (spring constants 0.01-100 N/m; see also Neumister and Ducker²⁸⁰⁹). Sometimes nanotubes were transferred from tip to tip. The manipulator could function both as a research tool for investigating properties of carbon nanotubes and other nanoscale objects without surface restrictions, and as a rudimentary building device for larger nanotube assemblies. Zyvex believed that this capability to select and manipulate nanoscale components and to examine directly their suitability as construction materials during various phases of the construction process would play an important role in enabling the technology of assembling mechanical and electronic devices from prefabricated components. Most impressively, the Zyvex system allowed the simultaneous coordinated operation of three independently controlled AFM tips within the same workspace, at different orientations. (A four-tip system was expected to be operational by mid-1999 [Mark Dyer, personal communication, 1999].) Two-handed coordinated micromanipulation under the view of an SEM had previously been demonstrated by others.²⁸¹⁰

Tuzun and colleagues²⁸¹¹ calculated the requirements for coaxial docking of two nanotubes of different diameters to form a molecular bearing. They looked at bearings formed from nanotubes 11 rings long, with 10 carbons per ring in the shaft and either 30 or 34 carbons per ring in the sleeve. For the computer simulations, Tuzun placed the sleeve of the bearing along the z axis and gave the shaft a small initial velocity towards it. A perfectly aligned shaft falls straight into a potential well from the van der Waals attraction to the sleeve; displaced or misoriented shafts can bounce off the edge of the sleeve. The docking envelopes for the molecular dynamics calculations and for the rigid body calculations had essentially the same shape, but with slightly different sizes. For the atomistic calculation, a shaft aligned parallel to the sleeve (with a 30 carbon ring sleeve) can be displaced by 0.26 nm before it fails to dock, while in the rigid body calculation the displacement can only be 0.16 nm. The authors note that "if an end of the shaft points closely enough to the center of the sleeve, it will fall into the nonbonded potential energy well and the two nanotubes will dock." These calculations are important because they tell us what tolerances can be accepted during the manual assembly process. Note the authors: "A question just as important as how to design or operate nanomachines is how to assemble them".²⁸¹¹ Interestingly, one of Zyvex's experiments²⁸⁰⁸ produced a multiwalled carbon nanotube that thinned in three steps along its length, suggesting the most likely explanation that the inner shells had been pulled from the outer shells. This phenomenon has also been observed by others and is known as the "sword and sheath" failure²⁸¹² -- the exact inverse of the purposeful nanotube insertion operation investigated by Tuzun and colleagues.

In order to produce large numbers of nanoparts and nanoassemblies, massively parallel SPM arrays and microscale SPMs^2772-2774 would be most convenient. Force sensing devices such as piezoelectric,²⁸¹³ piezoresistive,²⁸¹⁴ and capacitive²⁸¹⁵ micro-cantilevers made it possible to construct microscale AFMs on chips without an external deflection sensor. (We exclude fixed-tip arrays²⁸¹⁶ in the following discussion.) In 1995, Itoh and colleagues²⁸¹⁷ at the University of Tokyo fabricated an experimental piezoelectric ZnO₂onSiO₂ microcantilever array of ten tips on a single silicon chip. Each cantilever tip lay ~70 microns from its neighbor, and measured 150 microns long, 50 microns wide and 3.5 microns thick, or ~26,000 micron³/device, and each of the devices could be operated independently in the z-axis (e.g., vertically) up to near their mechanical resonance frequencies of 145-147 KHz at an actuation sensitivity of ~20 nm/volt -- for instance, 0.3-nm resolution at 125 KHz.

Parallel probe scanning and lithography has been achieved by Quate's group at Stanford, which has progressed from simple piezoresistive microcantilever arrays with 5 tips spaced 100 microns apart and 0.04-nm resolution at 1 KHz but only one z-axis actuator for the whole array,²⁸¹⁸ to arrays with integrated sensors and actuators that allow parallel imaging and lithography with feedback and independent control of each of up to 16 tips, with scanning speeds up to 3 mm/sec using a piezoresistive sensor.^2819,2820 By 1998, Quate's group had demonstrated^{2821-2827,2829} arrays of 50-100 independently-controllable AFM probe tips mounted in 2-D patterns with 60 KHz resonances, including a 10 x 10 cantilevered tip array fabricated in closely spaced rows using throughwafer interconnects on a single chip.

MacDonald's group at the Cornell Nanofabrication Facility has pursued similar goals. In 1991, the team fabricated their first submicron stylus, driven in the xy plane using interdigitating MEMS comb drives,²⁸³⁰ including the first opposable tip pair (Fig. 2.27). By 1993, they had produced a 25-tip array on one xyz actuator,^2831,2832 and by 1995 a complete working microSTM (including xy comb drives) measuring 200 microns on an edge and a microAFM measuring 2 mm on an edge including a 1-mm long cantilever with a 20-nm diameter integrated tip on a 6-micron high by 1-micron diameter support shaft.^1749,1974 MacDonald's group demonstrated tip arrays with 5 micron spacings, exploiting the same process used to make the working microSTM.^2833,2834 With the same technology tips or small arrays of tips could be spaced 25-50 microns apart and integrated with individual z-axis microactuators, so that one xy-axis manipulator could support many tips with each tip having a separate z actuator. By 1997, the group²⁸³⁵ had built and tested an array of microSTMs on the surface of an ordinary silicon chip, with each tip on a cantilever 150 microns long with 3-D sensing and control. The largest prototype array has 144 probes,²⁸³⁶ arranged in a square consisting of 12 rows of 12 probes each, with individual probe needles about 200 microns apart. Further development was to focus on increasing the range of movement and on fitting more and smaller probes into the same space.

Last updated on 16 April 2004