= V4 Outline MultiLine NoSorting TabWidth=30 H="1, Nanowires/tubes" Talk 1 Charles Lieber, Harvard Univ Nanowires and nanotubes: building blocks for nanotechnologies Understandability: 4/5 Motivation: given the appropriate building blocks, and hierarchical assembly, new tools and devices can be created. Carbon nanotubes and semiconductor nanowires, assembled with functionalized scanning probe tips, are promising candidates. NANOTUBE-TIPPED SPMS: In the past decade, scanning probe microscopes (SPMs, including AFT CFM EFM MFM STM) have become the workhorses of molecular imaging, but their use as manipulators is limited because the size/shape of the scanning tip is not reproduceable. Some groups have tried attaching bundles of SWNTs to the tips, but Lieber's group prefers to grow a single tube from the tip, choosing an appropriate catalyst and CVD flow rate to control its diameter; and this process *is* reproduceable. They can further tailor/functionalize the tip, attaching organic groups using well-understood synthetic chemistry. NANOTUBE CROSSBAR ARRAY: Lieber's group envisions a high-density memory device based on a criss-crossed "suspended crossbar array" of nanotubes: an upper weft, a lower woof, with switches at the intersections. These crossings are bistable; at a large spacing, the nanotubes are in a mechanically stable "elastic" state. Move them a little closer, and Clunk! they flex together into a Van der Waals-dominated state. The predicted operating properties of the crossbar array are very good. One trillion switches (a terabit of memory) fit in a square centimeter, operate at 200GHz at room temperature, and dissipate heat easily. As a computer, the array performs 1e11 OPS. As memory, they're bistable at 10-plus times the thermal limit, and nonvolatile (unlike DRAM). The states are easily distinguishable, due to the switch between ohmic and tunnelling behavior. They can operate at 3|5 volts, the standard 0|1 levels of today's CMOS. To do: develop scalable assembly strategies. They plan to use a microfluidic device: a first laminar flow deposits/aligns the nanotubes of the weft, then a second perpendicular flow aligns the woof. NANOWIRES: Nanoparticles prefer to grow in "zero" dimensions, ie as spherically-symmetric crystals. To grow useful materials, such as one-dimensional (ie linear) nanowires, that symmetry must be broken. One technique uses catalysts, supersaturation and precipitation. Silicon nanowires can be doped with boron to make p-type materials, or phosphorus for n-type. Ditto for InP (indium phosphide); moreover its conductivity can be varied by 1e5, and lightly doped it acts as a FET. Doped nanowires can be assembled on the lab benchtop; no billion-dollar clean-room factory is required. A forward-biased InP nanoscale junction acts as a point-source LED, with emission frequency related to the nanowire's diameter. (Image: glowing nanowire suspended between four CMOS terminals.) H="2, Moving contact" Talk 2 Jacqueline Krim, North Carolina State Univ Viewing a moving contact: an STM-QCM study of sliding friction in adsorbed molecules Understandability: 2/5 Motivation: to use friction to control the motion of atoms on surfaces. Krim has studied the atomic origins of friction for 15 years. A venerable experimental device is the quartz crystal microbalance, a piezoelectric slab exquisitely sensitive to the number of atoms deposited on its surface, and the layers in which they self-assemble; it responds by altering its vibration. At the nanoscale, the processes of diffusion (thermal kinetics) and friction meet; they're no longer mere perturbations. There are marketable MEMS devices today with cantilivers (eg accelerometers in auto airbag triggers), but none with sliding/rotating parts, due to the peculiarities of wear and sticky-friction, or "stiction". "MEMS tribology" is the study of friction and lubrication at this scale. On a surface (interface) motion is confined, to two dimensions, and sometimes to certain directions within that (due to the surface lattice). There are five mechanisms to move items across the surface, to induce their assembly: self-assembly (homogenous and unspecific), shaking to encourage diffusion, tilting the surface, creating a wind across it, or dragging individual items with an SPM tip. Krim's group has studied: sliptime versus surface corrugation. Xenon on Ag(111) surfaces. Frictional losses are due to phonons (about 10% on a perfectly clean metal surface) and electron dragging. At the superconducting transition, where electron coupling changes, so does friction. Quantum mechanical non-continuum issues. When an object falls through a fluid, it eventually achieves terminal velocity, when its weight is balanced by drag (ie fluid friction). Krim's group has found an analogous effect with atom clusters sliding over a surface in a vacuum. Gradually tilt a surface to the vertical, and the clusters attain a speed of 2 nm/s. H="3, Atomic sliding" Talk 3 Liangchi Zhang, Sydney Univ The size effect on the friction in atomic scale sliding Understandability: 3/5 Motivation: from the POV of a mechanical engineer, what are the atomic-scale transition mechanisms between frictional modes (analogous to sliding and rolling at the macroscale)? Is the transition dependent on material properties? Simulations: of a hydrogen-terminated diamond object (an "asperity") moving against a copper surface (the "substrate") in a vacuum. The asperity has a radius R, moves with velocity V, and digs into the substrate to an "indentation depth" d. The asperity in a so-called two-body (sliding) interaction will (given decreasing depth) successively cut, plough, adhere, or exert no wear on the substrate. In three-body (rolling) interaction, it will plough, adhere, condense, or not wear. Conclusions: At a very small contact length (small R), frictional stress is at the theoretical shear strength of the substrate, and concurrent stick-slip movement occurs. (Pic: a friction-time graph, showing a sawtooth with jagged upward slopes.) Above a certain critical length (large R), there's a transition, and no dislocation occurs. (Pic: a low-amplitude jagged flat line.) Friction varies with R and d, and the transition is material-dependent. H="4, Synthons" Talk 4 James T. Spencer, Syracuse Univ On the way to molecular nanosystems: the design and synthesis of molecular synthons for formation of nanostructural architectures See: Poster 1 Understandability: 4/5 Motivation: to use solution chemistry to synthesize units ("synthons") that will self-assemble into larger structures (rods, rings, gears, helices) amenable to positional mechanosynthesis. Polyhedral borane clusters are a good candidate, as are bora- and aza-adamantane. A successful synthon will meet several criteria: unidirectional synthesis, high yield, and rigidity (ie limited degrees of freedom) with controllable chemical/electrical properties. There are well-known synthetic pathways to produce polyhedral borane clusters. They have a rigid framework, are chemically and thermally stable, and their properties are controllable by substitution (as with conventional organic chemistry). One possible synthon consists of two icosahedra, joined by a 5-ring which sprouts a pennant, a multi-ringed functional group. The principle of "pi-stacking" (in which the pennants electron orbitals reinforce, favoring synthesis) can be used to produce rods, rings with the pennants facing in or out, and single- and double-helixes. Adamantane is essentially a very small, pyramidal fragment of diamond. If boron or nitrogen is substituted for an apical (apex/vertex) carbon atom, it is redesignated bora- or aza-adamantane. (Only the former has yet been synthesized; Spencer's group is investigating the latter). Such substitutions provide flexibility and control in designs. The adamantanes can be used as pendentives (curved corners) to join together multiple open-ended buckytubes, creating nanoscale plumbing junctions (3-, 4- or 6-way). H="5, NIST nanofab" Talk 5 Robert J. Celotta, NIST (National Institute of Standards and Technology), Electron Physics Group Nanofabrication for nanoscale science Understandability: 4/5 Motivation: to fabricate perfect standard nanostructures against which to test measurement tools for the support of US industry; to create a systematically varied series of structures. These structures may best be made with bottom-up techniques. MAGNETIC THIN FILMS How do magnetic thin films couple (ie how do their fields relate) when separated by different-size nanometer-scale nonmagnetic spacers? Relevance: magnetic hard drives. Celotta's group took an iron whisker (with an atomically flat surface) 500nm long, deposited a wedge-shaped layer of chromium atop it, then added a second whisker. The chromium wedge consisted of stairsteps, each one atomic layer (0.2nm) thick; these steps permitted measurement at a regular series of spacings. The findings: the horizontal magnetic field reversed with each step, except for each twentieth step; at the edges it was vertical. Surprisingly, on the first step the field was confused; later analysis indicated the Fe/Cr boundary was contaminated/alloyed by atoms that had migrated in each direction. Relying on thermodynamics to create the steps had compromised them. ATOM OPTICS The team created a regular pattern of atoms using so-called "atom optics". An MBE source of atoms was collimated with lasers, then focused with laser standing waves. Positional accuracy was limited by diffusion after the atoms landed on the surface. The collimator consisted of a laser beam transverse to the atom beam, tuned to just outside the absorbtion frequency of the atoms. Atoms with horizontal velocity perceived a doppler shift in the light, interacted, and were pushed back towards the center of the beam. The laser standing wave acted as a lens, its electric field inducing dipole moments in the atoms, causing them to pile into ridges. A pair of intersecting standing waves created peaks. To do: adapt the process to lower temperatures, to reduce diffusion. AUTOMATED ATOM ASSEMBLY As envisioned, "A3" will rearrange scattered atoms on a flat surface into a desired two-dimensional final product. Atoms are first evaporated onto the (very cold) surface, and their locations are recorded by an STM. The planner software then calculates a sequence of moves to drag the atoms together, one-by-one (using an "expanding convex hull" to avoid leaving unused feedstock within the product). The atoms are dragged along the lattice (ie through the valleys), lest they stick or fly loose altogether. (Animation: hypnotic series of zigzag moves, output by the current version of the planner.) To do: program the planner to correct for errors (ie feedstock that drifts, or atoms that *do* stick or fall loose). FURTHER TOPICS OF INTEREST The Kondo Effect. Confinement in two dimensions. For spintronics (systems that use up/down electron spin, not the presence/absence of charge), spin-dependent tunneling and superconducting-ferromagnetic tunneling. Hi-res tunneling spectroscopy. Tricks to permit three-dimensional assembly (many inspired by Egyptian pyramid-building techniques): screw dislocations in the substrate, sacrificial stairsteps. H="6, Quantum 1/f" Talk 6 Peter H. Handel, Univ Missouri-St.Louis Quantum 1/f proximity effect in nanotechnology Understandability: 1/5 Many of the presenters had difficult accents, many viewfoils were apparently recycled from discipline-specific conferences, and a few topics lacked any obvious connection to molecular nanotechnology (in the engineering/manufacturing sense). Handel's presentation suffered from all of these, and consisted mostly of a long mathematical derivation. Some choice statements: "There are many device applications; I'd like to jump over those ... Our whole existence in time and space is due to this: noise currents with 1/f amplitude ... fundamental fluctuation of all process rates and all cross-sections the way I define them." The upshot seems to be: the quantum 1/f effect does not linearly scale down to the nanoscale, and is not as great an impediment as you'd expect. This has implications for the permissible spacing of quantum wires. H="7, NIST M3" Talk 7 Kevin W. Lyons, NIST An open architecture for virtual reality in nano-scale manipulation measurement and manufacturing (M3) Understandability: 4/5 Motivation: to create a scientific/engineering software foundation to support measurement/standardization for the US manufacturers. Prior efforts have created software models to describe "conventional scale assembly", models which perform realtime calculations of rigid body dynamics, gravity, and simple friction. Nanoscale assembly is different: gravity has been replaced by Van der Waals, electrostatic, electrodynamic, and surface tension forces. Synthetic visualizations are needed, plus hints to users to disregard inappropriate macroscale "common sense". Simulating the full gamut of nanoscale forces is too time-consuming for realtime interaction. Instead, M3 aims to extract key information from larger/slower models; the user can explore that subset as a supplement, enhancing visual/tactile understanding. NIST's "model-centric" architecture must coordinate/synchronize multiple I/O devices, eg realtime control of optical tweezers and feedback through haptic (tactile) controls. Their efforts aren't so much concerned with the functionality of the application, as with ensuring the model can adequately represent the key aspects of a nanoscale devices and the processes that measure/manufacture it. This promotes the interoperability of other emerging applications. At this time, they've defined the initial architecture, coding framework and collaborative framework (in UML). They're working to eliminate failure modes. They're waiting for a later generation of haptic devices that can represent torques. H="8, Directed assembly" Talk 8 Paul Weiss, Pennsylvania State Univ, Dept of Chemistry Placement, control and isolation of molecules via directed assembly Understandability: Motivation: create nanoelectronic devices with SAMs containing multiple types of thiol chain. To organize them, define surface interfaces, including defect types and density, to control mixing vs. separation of monolayers. Currently popular in research, organic chains with a sulfur foot can organize themselves on a Au(111) surface (substrate) as a SAM (self-assembled monolayer), like pencils jostling in a box, or the lipids in a cell membrane. Two different varieties of chain can phase-separate (segregate themselves) by appropriate choice of terminal (opposite the sulfur) and internal groups. When tugged with an SPM, what actually moves is an adsorbate-substrate complex, since the S-Au bond is stronger than Au-Au. An altered chain has altered electron transport properties, and thus looks different to an STM -- and to a nanoelectronic circuit. The designer can vary the thiol head, the functional groups on the chain, the chain's length, and the terminal group. A device needs multiple types of molecule in specific locations, not a homogenous mixture. The fabrication process has three steps: 1. Heat the SAM in a bath of the original thiol, healing film defects, growing the domains (clusters of rods), and removing substrate defects. 2. Heat the SAM in neat solvent, desorbing the thiolate. 3. Immerse in the second thiol solution, grafting new molecules onto existing domains. Otherwise miscible molecules are now segregated! To build a molecular switch, you need to insert a fully conjugated molecule with an inbuilt dipole into a SAM alkanethiol film. Question: how many of these molecules are needed for switching behavior, or for persistence (as a memory element)? To do: use control of SAMs with electron-beam lithography to produce smaller on-chip features. First, etch two parent electrodes on a surface at the minimum distance achievable, creating two small, square hills. Second, deposit multiple thiol/carboxylate layers, turning the hills into round mountains bounding a very specifically-sized valley. Third, add a metal coat to the hills. Finally, deposit a third electorde in the valley. To do: create an artificial molecular motor based on tetrahedral carbon. The rotor is shaped like a caltrop (one of those nasty spiked anti-horse weapons), with three feet and a vertical shaft, and is placed between several electrostatic terminals built with standard STL. H="11, Bio nanodevices" Talk 11 Klaus Schulten, Univ of Illinois Urbana-Champlain, Beckman Inst Theory and modeling of biological nanodevices Understandability: See: www.ks.uiuc.edu/Research/namd/ NAMD is a molecular dynamics simulation package, for fast calculation of full electron states and constant pressure ensembles of 1e6-plus atoms. It's scalable to thousands of CPUs, and runs on the IBM Blue Gene, Origin, the Pittsburg Supercomputing Center's teraflop, SP3, and T3E clusters. It uses Tcl for scripting, and its data file is compatible with CHARMM and X-PLOR. NAMD2 uses NpT simulation. (The group is currently working at GFlops, but will soon have access to a TFlops machine, hence 100 times the power. Blue Gene's architecture would provide another 100-fold improvement, and they're close to an agreement with IBM for NAMD to be a primary protein-folding package.) VMD is a package for interactive molecular dynamics and visualization. It runs under both MSWindows and Unix. It also uses Tcl, and supports stereo 3D displays and haptic feedback. (Pic: VMD sends commands to NAMD, which responds with coordinates. Both communicate with BioCORE.) Studied: the passive elasticity of muscle, as each domain (alpha-helix) stretches and as hydrogen bonds snap one-by-one. (The group doesn't actually know the timescale for H-bond *re*connection, since the simulation covers only a few nanoseconds and the full folding process is a thousand times longer.) Studied: The retinal system. Every atom (protein, lipid, water) is now known, so proton-pumping can be described starting from the properties of the bulk phase. Studied: the purple (photosynthetic) membrane of halobacteria, which contains a protein that assembles in triads to form hexagonal proton pumps. Studied: the full photosynthetic unit of purple bacteria, as found in _Rs. molischianum_ and _Rb. sphaeroides_. A computationally-derived search model was used to interpret crystallographic data. LH-II consists of 24 chlorophylls and 8 carotenoids supported by multiple peptide helices. NAMD can compare the experimental and predicted times of electron transfer (the various stages take 40fs-35ps). For nanotechnologists: these complex psuedomechanical systems all operate at physiological temperatures despite vibration. (Pic: cross-section of cytoplasm-periplasm membrane, with many photon-capturing LH-II feeding to one LH-I with an RC (reaction center) core. Nearby, bc1 passes protons to ATPase. From both the RC and bc1 hang a cytochrome C2.) H="12, BARC" Talk 12 Richard Colton, US Naval Research Laboratory BARC: A Magnetoresistive Biosensor Understandability: 2/5 Motivation: to rapidly detect the signature proteins of bacteria and viri, for medical and biowarfare applications. A multi-analyte pathogen sensor, specific, fast, field-usable, as a subsystem for other devices. 30 minutes preparation plus 15-180 minutes to detect. Sensing by force discrimination. Basic design uses an AFM to detect interaction between a single ligand and receptor. Replace the single AFM tip with a parallel device: a suspension of functionalized paramagnetic beads (~1-um diam, ~0.5-pN force). Have developed from 2-um to 700-nm, heading smaller. After force discrimination, there are 3 detection methods: an array of piezoresistive cantilevers (FABs), optical microscopy (FDB), or a magnetoresistive bead array counter (BARC). Thiolated PEG 5000 (polyethylene glycol) with a nonfouling coating. 8-analyte DNA hybridization assay, using 5-um sensors. 68 contacts, servicing 66 GMR sensors plus 2 grounds. 5-mask fab. Mounted beneath a diffuser flow cell. Custom ----, ~100G rms at 200Hz. The signal is compared against a protected reference sensor, in a Wheatstone configuration with lock-in detection. The device has no front-end as yet, but there are many ways to quickly collect airborne pathogens and then extract their DNA. To do: build a model with 64 sensor regions (8 times as many), each 2-um (small), with a total active area of ~5e4-um2 (10 times as much). Build a multi-analyte assay. H="13, Stereocilium myosin" Talk 13 Peter G. Gillespie, Oregon Health Science Univ Manipulating a molecular motor by changing substrate inhibitor sensitivity Understandability: 4/5 Lining the snail-shaped COCHLEA of the inner ear is a strip of nerve tissue called the ORGAN OF CORTI, composed of so-called HAIR CELLS; these are the actual transducers of hearing. Each cell sprouts hundreds of STEREOCILIA, lined up in neat rows of increasing height. Each cilium is tied to the next-higher by a narrow fiber, the TIP LINK. When a cilium bends in response to a passing pressure wave in the fluid-filled cochlea, the tip link tugs open a MEMBRANE CHANNEL on the next-taller cilium, initiating a neural impulse. This channel is linked to an ADAPTATION MOTOR on an actin fiber, which can tow it through the cell membrane to reduce the tip link tension after a sustained stimulus; this implements HABITUATION. The system can detect a wave of only 0.2nm against a background of 2nm brownian motion. Question: what is this motor? It's probably a form of MYOSIN, since that's the only motor molecule known to be associated with actin. Problem: the myosin family is huge, over a hundred types in 17 families; eg mice have 35-plus myosin genes. It's known that deleting the genes for myosin-VI, -VII or -XV causes deafness in both mice and humans. The target: myosin-I-beta, over 100 molecules of which occur in each stereocilium. The standard knockout technique (whereby the gene's expression is inhibited organism-wide) could have unexpected and confounding effects. Instead, they'll use a newer technique: inhibit its activity on a short timescale, by engineering a mutation into the gene, delivering the gene to the hair cells, waiting for it be expressed, then delivering the inhibitor. They alter the 61st residue (amino acid) from tryptophan to glycine (Y61G), increasing the size of the ATP binding site. It still uses (hydrolyzes) ATP normally, but now an N^6-modified ADP inhibits it (the myosin sticks to the actin). The result: the hair cell no longer habituates! Implications: a general tool for investigating the role of particular myosin isozymes, and a generalizable technique to determine the role of a known protein in a well-characterized biochemical system. Engineering: In nanoengineering, a system could be built with multiple isozymes, each response to a different inhibitor. H="14, Molecular shuttles" Talk 14 Henry Hess, Univ of Washington Molecular shuttles: building a monorail on the nanoscale Understandability: 4/5 Kinesin is an ATP-powered cellular motor molecule that travels along tubulin microtubules at 800nm/s. Its behavior is typically studied via an INVERTED MOTILITY ASSAY, in which the motors are adhered to a glass plate, and the tubules move. To guide them, PTFE (teflon) is rubbed against the glass to make parallel grooves. (Animation: time-elapse footage of tubules crawling to and fro, with a few rogues caught on surface defects, running in circles.) To build a monorail, cargo must be attached to the tubules. Hess's group uses the common biotin-streptavidin technique. The tubules are BIOTINYLATED, and magnetic beads are coated with STREPTAVIDIN. Motion is controlled by changing the ATP concentration (denoted [ATP]): increasing it with CAGED ATP, and decreasing it with HEXOKINASE. To do: place an artificial photosynthetic system on a liposome/vesicle, to provide a local supply of ATP, controllable by an external light source. Problem: ATPase happens to prefer a high [ADP] while kinesin doesn't, but nature's own solution can be applied: ADENYLATE KINESE catalyzes ADP+ADP->ATP+AMP, to balance [ADP] and [ATP] as they're used by synthases and kinesin, respectively. Question: can specific paths be carved with STL? Only certain types of path, because the tubules are stiff. H="15, BioNanotechnology" Talk 15 William A. Goddard III, Caltech Beckman Inst BioNanotechnology: de novo Simulation and Design www.wag.caltech.edu See: Talk ---- Understandability: Motivation: to simulate useful devices made with new techniques in organic chemistry. Optimum development requires cooperation between experiment and theory. At NYU, N.Seeman's group developed a switch, based on the propensity of a DNA double helix to switch between its B and Z forms in the presence of MgCl2 salt. Also a Holliday junction array. At NYU, S.Wilson's group used mice to create antibodies to fullerenes: C60 C70 C84 and nanotubes. These could be used for filtration/separation, and the simulations explored how their specificity could be altered by modifying the residues forming the binding site. Exploring biosynthetic strategies for nanoscale assemblies: fasteners, containers, sensors, power supplies, and templates that form regular arrays. Some of these would use non-natural/novel amino acid analogs. Necessary to design new RNA synthetases, the class of enzymes that attach amino acids to the appropriate tRNAs, sending them off to the ribosome. Sensors, using a 7TMR-GPCR (7-transmembrane, G-protein-coupled receptor). To do: create a transducer, in which the binding event generates a detectable signal. Possibly a bistable system, so that a change in the helices produces an electric charge detectable by chemFET. Use the MembStruk modeling protocol to predict structure, and perform coarse-grain optimization with MPSim. H="16, Voltage junctions" Talk 16 Mark Ratner, Northwestern Univ Voltage issues in molecular wire junctions - control and mechanism Understandability: 2/5 Motivation: to study the simplest molecular device, the interconnect/wire. There are 5 transport mechanisms, including coherent resonant tunnelling, coherent non-resonant tunnelling. Behavior is complicated: the equation for I(V) (current as a function of voltage) couples the behavior of both electrodes and the molecule between. At 10-MOhm and 1-V, ~0.1-uA flow through a single short molecule. Current drops exponentially with length. (Known materials with mega-ohm resistances are usually called "insulators", but given 1e12 electrons transported across a single molecule with a cross-section of 1-nm2, that's actually 1e6 greater current density than a macroscopic wire.) For an effective junction versus a controllable junction, use vibronic gating. For a cool junction, use resonant transfer or voltage-drop Poisson engineering. H="17, Computer components" Talk 17 James Tour, Rice Univ Constructing a computer from molecular components www.jmtour.com Understandability: 3/5 Motivation: to substitute a small number of molecules for each silicon transistor; to increase packing density, bypass the limits of layered SiO2, and reduce fab cost. Chains Synthesis of the chains is performed on a polymer support, and is exponential: two rings join into a 2-mer, two of those become a 4-mer, then 8- and 16-mers. Every component needs "alligator clip" end groups for interconnection. Nanopore. While conventional silicon DRAM holds a charge for only 1e-5 to 1e-4 seconds (ie it's "volatile"), the nitroaniline DRAM cell holds it for 600s, and nitrobenzene for 900s. Simulated: the Nanocell, a randomly-networked collection of nanoparts trained to perform a logic function (OR, AND, a full half-adder, &c). It's a 1-um2 cell with I/O pins on all four sides, filled with metallic nanoparts joined by functional groups. The network is self-assembled, but not for any specific purpose; it must be trained, using a genetic algorithm whose fitness function is the desired truth table. The final result resembles a percolation network, and its massive connectivity provides fault tolerance; knock out a few connections, and charge still flows through the remainder. Projected: If a processor consists of 1e6 cells and each takes 1s to train, the array will require 12 days (better than a 2-month IC fab pipeline). But that time can be reduced with exponential training: the first cell trains the next, &c. To do: Find a better alligator group that isn't so dominated by parasitics. Finish preparing a 4-year timeline to submit to DARPA. Expect to build the first chips in 1 year. H="18, Circuit analysis" Talk 18 James Ellenborgen, MITRE Nanosystems Group Molecular-scale circuit analysis and novel approaches to nanomanipulation www.mitre.org/technology/nanotech Understandability: 3/5 Motivation: To rapidly simulate molecular circuits by including only a few quantum effects. Due to quantum effects, Kirchoff's Laws must be revamped; nanocircuits may perform better than microcircuits, with sharper switching behavior. Resistances are added as follows: Macroscale: 1/R_tot = 1/R1 + 1/R2. Nanoscale: 1/R_tot = 1/R1 + 1/R2 + 2/sqrt(R1*R2). Examples: Use MolSPICE to analyze a 100-nm^2 adder circuit built of demonstrated molecules. Design an XOR gate with diodes, rectifiers and molecular RTDs (resonant tunnelling diodes). "Millions of magic nanofingers" .... PGM (patterned granular motion) is a phenomenon whereby standing waves (stationary peaks) can be created in a mass of millimeter-scale granules, given appropriate agitation. A MITRE intern has demonstrated the same can be done with buckyballs, and has shown how to create any desired pattern. Status: experimental apparatus has been built for manipulating nonstick-coated particles of size 5-10-um (1e-1 precedent). To do: develop a detailed understanding of forces on agitated nanoparts, and scale down to 1e-3 precedent. H="19, Lattice for MEMS" Talk 19 Richard Superfine, Univ of North Carolina The atoms matter: lattice effects for NEMS Understandability: 3/5 See: http://www.physics.unc.edu/~rsuper/research/ See: http://www.cs.unc.edu/Research/nano/ Topic 1: The group is experimenting with VR/AFM nanomanipulation, with head tracking, stereovision, and haptic feedback from the AFM tip. The 1e6 magnification feeds a 3D rendering. The system switches between two modes, touch/modify and oscillate/image. Topic 2: The parts list for NEMS (nano electro/mechanical systems) devices includes gears, bearings, couplings and hinges; these must variously roll and flex. When a rigid body (such as a nanotube) moves across a surface (lattice, substrate), there's competition between phi/theta (rolling/spinning) and x/y motion. When registered with the lattice, the tubes roll; when not, they slide, which represents a *lower* energy state. Superfine's group has discovered that nanotubes won't rotate freely on the substrate (like the spinner in a board game); they lock into specific angular positions. The nanotubes aren't simply sticking to the lattice (like a pencil trapped in bubblewrap), since different tubes exhibit different angles (phases); eg one will lock to noon-2:00-4:00, and a second to 1:00-3:00-5:00. They maintain their preferred phase even after collisions. The effect seems to depend on both the helical angle of the tube's lattice and the particular substrate: 60° increments on Si(111), 90° on Si(100). Nanotubes can be grown with CVD. The relative band structures of graphite versus carbon nanotubes. Tube-tube junctions are controllable/reproduceable, and could permit NEMS sensor. These effects could be used to fabricate a voltage-controlled oscillator, a mixer, or a linear encoder. H="20, Nanotube electronics" Talk 20 Michael Fuhrer, Univ of Maryland, Dept Phys Nanotube nanoelectronics: new devices and imaging electronic transport Understandability: 1/5 See: http://www.physics.berkeley.edu/research/mceuen/ A graphene sheet has a 2-atom unit cell. It is not an insulator, because its bands (sp2 and sp3) cross. Fuhrer's group has designed a 3-terminal rectifier with 2400 atoms in the active region. H="21, Quantum wells/barriers" Talk 21 Panos Datskos, ORNL (Oak Ridge National Labs) Semiconductor devices based on micromechanical quantum well and quantum barrier structures Understandability: 1/5 See: http://www.ornl.gov/ORNLReview/rev32_3/brave.htm Motivation: a noncooled tunable detection of IR photons using a novel fabrication technique, using suspended micromechanical quantum wells. A bimaterial layer (GaAs, Si, Ge, InSb) will shrink/expand/bend in response to incident photons. Such a device can be fabricated with diamond cutters or ion-beam milling. H="22, Switching/wiring" Talk 22 James Heath, UCLA (Univ of California, Los Angeles) Molecular electronic switching and wiring structures Understandability: Power cost of information tfr, P = nk_BT(d/c)(nu)^2 for boolean non-reversible logic. Num parallel ops, Boltzman, temp, transmission distance, c, operating freq. So, maximize the parallel ops, minimize distance. >13% all US pwr to computing! Cycle betw open/closed. HPL Teramac. Download logic onto crossbar struc. CMOS sw vs. moly sw. Dia logic on a crystalline ary. DL onto a struc the equiv of an adder, by putting hi/lo V on lines. 0.15um FET. 3 gates. At least one wire must be individually addressable, so you can't tile/crystal in 2D. Crossbars (1nm) *are*. Al crossboars, rotaxane. 3-input Or gate, noise. Oxidize away in-C. Need: analytical voltage. Interlocked catenane rings, with long-chain phospholipids. Avoid measuring capacitor-hysteresis. Bistable pseudorotaxane sw. All expt at 300k/ambient. Want to avoid STL; try nwire. Nwires in soln, fluid. SWNTs aren't soluble. Functionalize, but kills e- props. Wrap in lipid. +25mV bias, SWNT photovoltaic. -50mV, is photorectifier. Bundle of wrapped tubes, or wrapped bundle? The latter. H="23, Time/space shapes" Talk 23 Mostafa A. El-Sayed, Georgia Tech ... confined in time and space of different shapes "An inventor...a person who makes use of a property to better our lives..." Confinement fx on elecc props of matter. New props of nparts. Sz and shape fx on metallic npart props - Pt for catalysis, Au for luminescence and laser photothermal (the elecc surf plasmon oscillation). Particles shape-dependent. Why nm, not pm or um? The atom, but also mechly - prop char'd by specc len scale for elecc motion, detd by forces present. For semicon, Bohr radius (e-/hole sepn of excitons). For conducting met, mean free path of e-. So new props are expected if any matl phys reduced below "natural" len scale. Qtm sz confinement fx on lowest transition E (delta E) - band gap. Eg CdSe partc, 2.3nm/470nm but 5.5nm/620nm - emsn color changes! Synthetic methods in solution. Metals: reduction of ions, to atoms, to clusters, to npart. Semicon: precip ions, to molec, to clusters, to npart. Eg, medieval stained glass windows, red, used Au - unknowingly npart. Pt 1:1 11nn cybesm ir 6nn 1:5 polymer tetrahedra. Reduction of Fe(111) to Fe(11). Reaction rate differs for cubes and tetra. Temp fx. Au nanorod, ~20nm wide and 100nm long. Synth electrochem, soln, 42degC. Surf plasmon absorbn, is collective excitation of the *free* e- of a met cluster, leads to coherent oscillation. Size dependent. E- repulsion dephases sys, falls to noncoh rapidly. Mie theory for absorbtion of sph met ptc. Gans mdoif for rods. How long to melt Au nrod to sph? Pulse w/ laser, an dhow long till absbn band vanishes? Hi-res TEM study of shape tfmn. Size dep of fluor spectrum and yield of Au nrods. Ensn... Surf plasmon absorbtion sens to adsorbed species. 60fJ/1 rod, melt in ~30ps, >1.7ps for healing lattice by hot e-, and <100ps cool by phonon-pfonon relaaxation of solvent. Fluor yield enhancement xe6 is due to coupling with intense surf plasmon absorbtion. H="24, Growth kinetics" Talk 24 Peter Searson, Johns Hopkins Univ Growth kinetics of metal oxide nanoparticles Understandability: 1/5 See: http://www.jhu.edu/~matsci/people/faculty/searson/PCSgrouphome.html Nanoparticles are simply very small clusters of atoms, but they display properties (optical, electronic, magnetic, chemical) not predictable from the bulk phase of that material. They can be synthesized in the vapor phase (molecular beams, flames), solution phase (electrochemical reduction, chemical precipitation). Their properties are dependent on their size, which is dependent on nucleation and growth, which are known only empirically. Nucleation is fast. Particles can be grown by aggregation (random orientations, epitaxial) or coarsening (Ostwald ripening), in which large particles grow at the expense of small. The nucleation rate is dependent on the dissociation constant of the metal salt. The growth kinetics follow the LSW rate law. Gives example of synthesis of zinc oxide from zinc acetate. Quantum confinement. Absorbance of zinc oxide colloids. Can quench growh by injecting a suitable capping ligand, such as octanethiol. Topotactical attachment. H="25, Nonvolatile static charges" Talk 25 Edwin C. Kan, Cornell Univ, Dept of Electrical and Computer Engineering Sub-millimeter autonomous microsystems realized by mobile nonvolatile static charges (in nanocrystals) Si CMOS heading twd 20nm. New apps, prods, matls and mfg? Biomed implants, usensor, RF ID - but not computing. CMOS and help of self-asm ncrystals. Auton usys: computing and ctrl - plus sense perturbation and make judgement. Actuate perturb to env as resp. Comm for collective/coordinated. Pwer gen/conv. Insects. Another way? Inject static chg in EEPROM (nonvol, >3y at STP). Amt of static chg, polarity (e/hole) tunable. Alter v/y w/ ncrystal - stc chg won't rearrange (img charge), but is not tunable. Want sys <0.5mm. Integrate CMOS and MEMS. Self-starting vibrational generator (no batt). Motion det and actuation with det'd harmonic ctrl. Near-field static-chg transmitter (like radio static). Reduce mech fric and wear, with estc floating. Since can't move in ncrys, further reducn. No self-rep, so mfg temp can differ from operational. Immune to temp, radn, chem and bio hazards. Q: How much of this has been fabbed? H="26, Nanotubes for thermal" Talk 26 Brian Mayeaux, NASA Johnson Space Center Application of carbon nanotube composites for thermal management (Canceled) H="27, Nanotube membranes" Talk 27 Susan Sinnott, Univ of Florida Computational studies of carbon nanotube-based membranes and new materials Understandability: H="28, Curvature reactivity" Talk 28 Deepak Srivastava, MRJ at NASA Ames Curvature dependent reactivity of fullerenes and nanotubes Understandability: 2/5 See: http://www.nas.nasa.gov/~deepak/home.html See: http://amesnews.arc.nasa.gov/releases/1997/97_92AR.html Motivation: curved surfaces offer a wide range of templates for materials, devices and applications; they can be functionalized on either side. Chemical reactivity (when attempting to functionalize) of the interior and exterior (concave and convex) of a buckyball differ, and the difference does not equal the strain energy difference. The interior is less reactive (due to the lesser electron density), and is subject to more effects. Depends on the pyramidalization angle or pi orbitals, somewhere between sp2 and sp3. Mechanical kink-catalyzed chemistry. Contribution of binding energy. Small errors due to global relaxation. The group studied how strain energy contributes to reactivity, by tugging on a single carbon: in graphite, buckytubes, buckyballs, and by hydrogenating C36 and C60. A hydrogen atom within a C20D20 buckyball (carbon-deuterium) has no metastable site, and the diffusion barrier is 1.17-eV. An application in solid-state quantum computing: a bucky-onion (a nested buckyball) could be used to encapsulate a phosphate-31 dopant, then located within a silicon substrate. First, the dopant can be placed precisely; and second, it won't move subsequently. H="29, Diamond nanotube clusters" Talk 29 Olga Shenderova, North Carolina State Univ Carbon based nanostructures: diamond clusters structured with nanotubes H="30, DNA nanotech" Talk 30 Nadrian Seeman, New York Univ DNA nanotechnology DNA is linear antiparallel, right-handed helix Can design DNA strands to self-assemble into branched, not linear Fundamental ... the sticky end, then ligation. Predictable affinity (easy) but also predictable structure (often overlooked) So branches plus sticky ends allow crosses and lattices Objective: to generate desired moleculares, macroscropic ary, both periodic and aperiodic. As scaffold for crystals, for nanoelectronics. Why DNA? Predicatble intermoly (unlike ab-antigen), convenient automated chemistry, convenient modifyizg enzyems (ligases and restriction, exonucelases, topoisomerases), externally readable code, locally stiff polymer Problem: stiff helices but floppy junctions. Polyhedra. Rqmts for lattice design comp: predictalbe intersection, rigid. Branched junction, then stole double crossover. DAE, DAE+J, DAO. Don't actually bother ligating; the hydrogen bonds are adequate. 2D ary predictable, AB*, ABCD*, AB^CD* with restric for ^, or B0 and anneal stable. Parallelogram arrays, using Holiday and Bowie junctions Triple crossover arrays - periodic assembly that can extend into 3D, algorithmic design, Wang tiles for computing -- a 4-arm junction is equiv, according to the complementarity on the arms. For a species based on Wang tiles, see _Exodus_ (Greg Egan, fic, 199----) H="31, George Skidmore" Talk 31 George Skidmore, Zyvex Exponential assembly www.zyvex.com Understandability: Exponential assembly (large number of parts, parallel assembly) is distinct from self-replication. Zyvex start with a MEMS embodiment. With parts of different sizes, convergent assembly. Shared translation, shared control instructions for rotation. Each station needs 2 rot dof over 90deg and a gripper. Rich components, simple assembly. No risk of uncontrolled growth. Produce an array of unassembled stations. Assemble the first station. The first station assembles the second, on the opposite plane. #1 and #2 construct two more. MEMS STL - huge industry, mature, can produce a large variety of complex parts, but limited in materials, not true 3D (2D layered) and assembly is primitive. Self-rep, see Jacobson-1958, Penrose-1959. Electrothermal tweezers - pick, move and place a polysilicon plate. 62-um, oops surface tension. Expansion is scalable downward, as a top-down path to MNT. Open feedback, y. Deal with failure? Two parallel horizontal stages, each with a wafer covered with unassembled stations. Consist of two parts: a rotational base and a vertical flange with rotating shoulder, arm, gripper. Lift from surface, rotate to vertical, slot into the waist. Power connection made. [Movie] [Movie] Don't seem to be yet available on the Zyvex website. H="32, Systems issues" Talk 32 Ralph Merkle, Zyvex Systems issues in the development of nanotechnology www.zyvex.com/nano/selfRep.html The goal of MNT (molecular nanotechnology) is to fabricate most structures consistent with physical law, to do so precisely (every atom in its place), and inexpensively ($0.10-$0.50/kg). POSITIONAL ASSEMBLY addresses the precision requirement, and SELF-REPLICATION the cost. Both are applicable at many scales. Many architectures have been proposed for self-replication: Von Neumann, bacterial, Drexler's original robot arm, simplified Merkle-Drexler hydrocarbon Stewart platform, exponential assembly. The Von Neumann design is a manufacturing element, consists of a computer and a constructor, together comprising the manufacturing element. It follows instructions to make a new manufacturing element, then copies instructions into it. Does *not* contain instructions to build the instructions; that way lies infinite recursion. You can simplify the manufacturing element by enriching the environment. External tape. A bacterium follows a similar pattern. The ribosomes self-assemble, but each then performs positional assembly in a liquid environment with dissolved feedstock (amino acids), and the chromosome is duplicated. Drexler's first design was similar to a bacterium, a device with a flexible cylindrical wall and a set of assembly arms, driven by computer, with fuel/feedstock uptake. They'd build the new device within the cylinder, maintaining a eutectic environment. Floats in liquid environment with feedstock. Simplify: a macroscopic computer broadcasts to microscopic constructors. Are now smaller/simpler, and can be easily redirected to manufacture product; no need to propagate a new instruction tape. Also inherently safe, since lack instructions to replicate selves. HC assembler: broadcast (no onboard instructions). Limited range of output to stiff hydrocarbons. Acoustic sensor. Liquid env, feedstock three itmes: HC, Ne, vitamin (plus the solvent). Hydrogen abstraction tool for building diamond. (At Philcon, argument. In the variant of a device which is constructed in a flask by broadcast, but then passed an instruction tape for operation in the field, objection: the "copy" operation could be inadvertently activated early, recording the assembly instructions. Easy fix: make it a tripod: assembler, instructions, feedstock supplier. It might escape into the wild with instructions, but they'd refer to parts provided on a very specific conveyor belt that won't be found.) Exponential assembly. Common XYZ is a "broadcast" of positions. No onboard computer. MEMS 2-dof must be onboard, since rotation (unlike translation) is hard to share. Operate in air. Blind pick-and-place by dead reckoning. The term "self-replicaton" carries assumptions and connotations, mostly from biology, that are grossly inappropriate to MNT. Popular misconception. Living systems are indeed complex, adaptable, carry onboard instructions, self sufficient. Misconception leads to fear, leads to blocking research, prevents a deeper understanding of systems with valid concerns. *Attempting* to ban research is itself dangerous; you can't gain beneifts, nor detect dangers. Need more exploratory replicators systmes, public education. Self-assem, self-rep, ext rep H="33, Probing properties" Talk 33 Phillip E. Russell, North Carolina State Univ Probing the properties of nanostructures and nanomaterials Why probe? How? Eg. Physical spm, EM, charged particles (ion, e, clusters). Moving atoms - the 1991 articles were read by kids who are now entering college and asking questions like "why study conventional chemistry? I want to start off with nanotech." Why probe? To measure props, obsv response, modify. Can do: size sub-nanometer, force nN-pN, current << nA. atomic-scale chemistry. AFMs image, but: must interact with matter to image it. Not optical, but map related to tip/sample interaction. Manip-behavior. Nano vs macro, eg VDW forces >> gravity. Non-imag: measure mechanical properties, eg tug on proteins. Nanoparticles, sub-micrometer grain sizes. SAMs (as sensors/lubrication/adhesive) to enhance macroscale friction/wear/hardness. IFM - interfacial force microscope. Force vs. displacement. Nanomechanics of indentation. In=situ AFM, high desnity of nucleation sites. 2D movies are easy to make, but 3D are not. Can we believe the images? Tips not reproduceable, produce artifacts. So use multiple tools when possible. AFM min/img - faster? better tips? Driven by IC industry's desire to do litho. Ion beams Ions are far more massive than electrons, hence can transfer momentum to the sample. Ion beams can remove material from a surface, implant beneath, deposit layers atop, and induce chemical reactions. FIB (focused ion beam) STM tips. Can carve diamond tips into nanometer-scale tools The surface first swells due to implantation and lattice damage; then material is removed. Changes in "sputter rate" in the first few nm can be see with optical interferometry on an AFM. Electron beams Electrons are low mass and transfer little momentum, but can break chemical bonds, thereby generating detectable signals. The bias on an STM tip allows electrons to tunnel. But also FESEM and FESTEM (field emission scanning (transmission) electron microscopes) for 30 y, but are finally (again due to the IC inudstry) stable/reliable tools. Can induce tip growth with an e-beam. STEM combines SEM and TEM. The HD-2000 can achieve 0.235-nm resolution across a sample 4x0.5x0.6-mm. Contamination is lower than with a TEM, because of the samll sample chamber volume, easier to evacuate, no photographic film, no moving parts or lubricants. Atom type can be detected, because heavy ones scatter electrons more. Can examine quantum wells and lattice features. STEM can induce X-ray signatures. Can perform 3D spectral imaging: at each spot on the sample surface, takes a spectrograph. A FESTEM can image the walls of a nanotube. Most of the tools used by pure science are being improved by the needs of industry, mostly semiconductor. Certain items (eg moly pusher) we need. Upcomoing improvements: mini-SEMs, faster AFMs, nanomechanical testers. Need more 3D data. For R&D and QA (quality assurance) AFMs have the lowest cost, electron microscopes can scan deep into the sample, and FIB/SIMS can not only image, but also cut and deposit.