A publication of the Foresight Institute
Jeffrey Soreff's Technical Progress column is continued from the previous page.
One approach to nanotechnology is to construct systems with
scanning probe techniques. This has the advantage of providing
excellent control of the locations of the molecules being
assembled, but it has the disadvantage of being an inherently
serial process. It therefore requires techniques to investigate
very small numbers of molecules, ideally single molecules. The
following five items describe techniques applicable in this
The first paper in this section extends a spectroscopic technique to single molecules. Writing in [Science 275:1102-1105--MEDLINE Abstract], S. Nie and S.R. Emory describe a type of Raman spectroscopy that is sufficiently sensitive to detect individual molecules. In their experiments, rhodamine 6G was bound to silver particles with average diameters of 35 nm. Free rhodamine 6G absorbs light efficiently. Unfortunately, its visible absorption (and fluorescence) spectra are fairly broad, carrying little detailed information. In a normal dye's visible absorbtion spectrum the incoming light is absorbed by a transition from one electronic state to another and the effect of molecular vibrations on the energy difference between the states spreads out the absorbtion. In Raman spectroscopy, on the other hand, the energy of the visible photon is split into exciting a well-defined molecular vibration and a reradiated visible photon. This process yields much sharper spectra, with much more information about molecular features. The spectra in the paper show 6-7 distinct peaks from separate vibrations.
Normally, the "Raman process is an extremely inefficient process, and its cross sections (~10-30 cm2 per molecule) are about 14 orders of magnitude smaller than those of fluorescent dyes (~10-16 cm2 per molecule." Raman scattering is, however, enhanced at certain surfaces. These authors found that by binding a chromophore to the surface of a nanoparticle they could enhance it by factors "as high as 1014 to 1015," making it feasible to measure the Raman spectrum of individual molecules. In addition to the enhancement, another feature of the authors' system is that "photochemical decomposition or photobleaching is significantly reduced for single molecules adsorbed on metal nanoparticles because the metal surface rapidly quenches the excited electronic state and thus prevents excited state reactions." This is important since it increases the numbers of photons that can contribute information.
One odd feature of this paper is that the authors found that only certain Ag particles were optically "hot" (perhaps 1% to 0.1% of the total), and "only one out of 10,000 surface sites on a hot particle shows efficient enhancement." The authors suggest a number of ways to better control the fabrication of their nanoparticles, including some scanning probe methods.
This technique looks primarily attractive as a diagnostic technique for determining whether an attempt to produce a chemical change to a single molecule with an STM or AFM probe has been successful. Vibration spectroscopy of organic molecules has been a powerful analytical technique for decades. Making it feasible for individual molecules will add a valuable tool for the development of nanotechnology.
The second paper in this section extends what might be considered a hydrodynamic diagnostic to individual molecules. Writing in [Science 275:1106-1109 21Feb97--MEDLINE Abstract], X.-H. Xu and E.S. Yeung describe detecting the diffusion of individual fluorescent rhodamine-6G molecules. In one mode, they imaged the fluorescence in a 4-um thick layer of rhodamine solution confined between a cover slip and a prism. In another mode, they imaged it in a 0.15 um thick layer excited with the evanescent wave from a totally internally reflected laser. In both modes, they used sufficiently low concentrations of rhodamine (5 nM in the first case, 100 nM in the second) that molecules were isolated, even accounting for diffusion in a 100-200 msec exposure window. The authors measured the distribution of the fluorescence for each molecule and calculated diffusion coefficients for them. In addition to the rhodamine-6G experiment, the authors bound rhodamine to a 30-base strand of DNA. It had a reduced diffusion rate, as expected. This technique may be useful as a diagnostic for structures available in very small quantities, measuring the diffusion rate of a handful of structures built with scanning probe techniques, for example, perhaps to detect if an assembly step had succeeded.
The third item in this section describes the steady advances in application of electron imaging to determination of the structure of small numbers of particles. A trio of papers in Nature describe the determination of the hepatitus B virus by electron cryomicroscopy. B. B÷ttcher et. al. [Nature 386:88-91 6Mar97--MEDLINE Abstract] and J.F. Conway et. al. [ibid. 91-94--MEDLINE Abstract] wrote the detailed papers with D. J. De Rosier writing a commentary on them [ibid. 26-27]. This technique involves solving a structure by taking many electron micrographs (done at cryogenic temperatures, hence the name) rather than by growing a crystal and diffracting X-rays with it. From a biological point of view this technique is notable primarily because it permits the structural analysis of things such as membrane proteins and ribosomes which are difficult or impossible to crystallize. From a nanotechnological viewpoint, it is notable because it allows 3D structural analysis from a small number of particles. In B. B÷ttcher et. al.'s paper 6,400 images were used, and J.F. Conway et. al. used 600. If a structure is being built by a scanning probe technique, these numbers of copies are substantial but conceivable, while diffraction techniques requiring macroscopic quantities of structures are not. Note that electron micrographs contain information about the interior of objects, so, for example, they could provide valuable confirmation of a successful alignment of two interior subsystems in a structure, even if the structure's surface is being monitored with an AFM while it is being assembled. The B÷ttcher team achieved 0.74 nm resolution while the Conway team achieved 0.9 nm resolution. De Rosier writes that: "Structural analysis by electron cryomicroscopy of two-dimensional crystals now extends to essentially atomic resolution, of 3 to 4 ─," even with only thousands of units in the crystal.
The fourth item in this section describes imaging a single molecule's motion during the course of a reaction. In [Science 275:1882 28Mar97], E. Stokstad describes a film made by P. Hansma et. al. of an enzyme actually working its way down a DNA strand. The enzyme was an RNA polymerase. The scene was imaged with a tapping mode AFM. A difficult part of the experiment was to attach the DNA sufficiently firmly to the substrate that it would not be pushed away by the AFM or diffuse away but sufficiently loosely that the polymerase could still operate. Stokstad writes: "After several attempts, the team found that zinc ions added to the water would loosely attach the molecules to the sample dish." Hansma suggests that "an AFM movie might reveal subtle changes in the shape of the RNA polymerase as it passes over different letters of the genetic code on its way down a DNA strand." As far I am aware, this would be a novel sensing architecture, using a hybrid of a biochemically produced molecule acting as both an atomically precise sensor and a mechanical sample feed mechanism, together with a AFM used as a sensitive, but less precisely fabricated, readout mechanism.
The last item in this section describes a technique that can detect single charges, and which places an amplifier in submicron proximity to the system under study. Writing in [Science 276:579-582 25Apr97--MEDLINE Abstract], M.J. Yoo et. al. describe a high resolution electrometer that uses a single electron transistor (SET) as a sensing element on the very tip of a scanning probe. They detected charge distributions on a surface by scanning it with a tip terminated in a 100 nm patch of aluminum. Two tunnel junctions were made to this patch, and the current through it measured. The current varies approximately sinusoidally with the potential of the patch, including the potential induced from charges on the sample surface. It "passes through a full period each time the electric field lines terminating on the island induce a charge of exactly one additional electron." In this paper, this charge is primarily induced by the bias on the source and drain electrodes coupled to the island. The effect of the sample charges is to shift the phase of the variation. The phase shift can detect a signal of as little as 1% of an electron's charge. The SET is biased with ~1mV, switching currents of around ~1nA, so the signal power from the SET is ~1pW. The SET island capacitance was ~0.1fF, so the input signal energy was only around ~10-22 J and there is power gain for any scanning speed below 1010 pixels/sec. This is useful because one limitation in rapidly retrieving information stored in atomically precise structures is the available signal power from these structures. In both STMs and AFMs micron scale structures (FETs and cantilevers respectively) must be driven by atomic structures, while this tiny SET avoids this requirement.
The fabrication of the SET is surprisingly simple. "Fabrication of the SET involves the evaporation of three separate areas of a thin (10 to 20 nm) aluminum film onto a specially shaped glass fiber...The films for source and drain leads spread out from the edges of the tip and extend up the side of the fiber to electrical contacts...The three electrode shapes are defined by natural shadowing." I would have expected a much more complex process to be necessary to avoid shorting at such fine geometries. The success of this instrument suggests the possibility of many other uses for this technique. The authors have essentially gotten two voltage sources to within 100 nm of a sample on a single tip, while conventional STM tips only get one close to the sample. One might apply the same probe fabrication technique to dual tunnelling instruments, lateral conduction measurements on low conductivity samples, lateral electrostatic deflection of surface molecules or a variety of other fabrication or sensing experiments.
This electrometer requires operation at 2K in order for the coulomb blockade effect to make the SET operate. Scaling down the SET both improves resolution (currently set to 100 nm by the SET size) and increases operating temperature. The authors write: "Even room-temperature operation, although ambitious, is conceivable, requiring the development of molecular- or even atomic-sized SET tips."
An article by A. Hellemans in [Science 275:920 14Feb97] covers a new $3.7 million subproject in the EU's Esprit program. The project, "Fabrication and Architecture of Single-Electron Memories", intends to store bits with single electrons. Eight research labs are participating. The long term goal of the project is to build large scale memory chips, 1012 bit devices by 2015. The short term goal for "the initial 3-year contract is a 4 X 4 array of single-electron devices on a substrate of silicon." The researchers expect to place charges on conducting islets a few nanometers in size. While this project does not not directly attempt to control the placement of atoms to atomic precision, a successful single electron memory would allow routine contruction of precision electrostatic patterns that might then drive assembly of large patterns of charged structures. In addition, routine use of these memories would create incremental economic incentives for precision fabrication from the low nanometer scale down to atomically precise fabrication.
One approach to nanotechnology is to synthesize successively
larger, more complex structures by parallel techniques drawn from
chemistry and biochemistry. A key capability necessary for this
approach is the use of very selective catalysts to perform
reactions on selected sites on substrates while avoiding them on
chemically similar sites in other locations. For a number of
years, researchers have been using catalytic antibodies, abzymes,
as one of the approaches towards building selective catalysts.
The following two papers describe some recent advances in this
Writing in [Science 275:945-948 14Feb97--MEDLINE Abstract], K.D. Janda et. al. describe a novel selection mechanism for catalytic antibodies which captures the genes for the antibody as a result of the reaction itself rather as a result of binding to a transition state analog. In the particular reaction that they studied, a hydrolysis of a glycosidic bond, they synthesized a variation of the substrate which reacted with the abzyme to form a covalent bond connecting it to a solid support. More precisely, the substrate contained a phenol with an ortho difluoromethyl substituent. They write: "On enzymatic clevage, the difluoromethyl moiety generates a reactive quinone methide species at or near the active site [of the abzyme], thereby alkylating any nucleophile [in the enzyme, and capturing it]." The other end of the phenol was connected to a solid substrate through a disulphide bond. Thus, the catalysis itself bound the enzyme to the solid support. The authors also modified the abzymes themselves to carry the information needed to make additional copies of them. They used genetic engineering techniques to package the antibody genes into a phage, and to bind the phage to the antibody. After reaction with the substrate, phage, abzyme, and phenol are all left attached to an insoluble support, while non-catalytic members of the library were washed away. The authors then cleave the disulphide bond by reduction with DTT, and the freed phages are "amplified through infection of E. coli." This method can thus "detect catalysis in single phage particles", allowing for the direct screening of a very large number of potential abzymes in parallel. The authors write: "The advantage of purely chemical systems [such as this one] is one of generality in that many desired reactions do not yield intermediates or products that perturb cellular machinery, and thus biologically based selection systems cannot be used." On the other hand, this technique does require that the target reaction be modifiable into a form that captures the abzyme with the irreversible formation of a covalent bond, which may not be feasible for some reactions of interest.
J.-B. Charbonnier et. al., writing in [Science 275:1140-1142 21Feb97--MEDLINE Abstract] did an experiment that describes an improved procedure for finding catalytic antibodies, but may also probe the limits of this technology. The antibodies were generated in the usual way, by immunizing a mouse with a transition state analog for the reaction to be catalyzed. In this case, the reaction was an ester hydrolysis and the transition state analog was a phosphonate. The usual process is to screen "the immune response [library of antibodies] for binding to the hapten [transition state analog] and then testing the best scoring clones for catalytic activity." Instead, this group used a procedure called "catELISA, in which product-specific antibodies are used to detect the appearance of product after immobilised substrate is exposed to the supernatant of culture hybridoma cells." They were able to screen all 1570 clones derived from a mouse, of which 9 scored positive for catalysis, "a figure to be compared to 970 hapten-binding clones" which would otherwise have needed to be tested again for catalytic ability. The authors purified and crystallized the three most reactive antibodies complexed together with the transition state analog. They found a great deal of structural convergence of these abzymes, to the extent that "the conformations of the catalytic residues are similar." They write that: "The central question now is whether the mechanism we observe is a dead end or a point on the pathway which can be further refined." They suggest that further cycles of mutation and selection may improve it. It will be interesting to see if its catalytic efficiency can indeed be increased.
Jeffrey Soreff is a researcher at IBM with an interest in nanotechnology.
Newsweek continued its look toward the 21st Century by identifying nanotechnologist K. Eric Drexler as one of "100 people to watch as America prepares to pass through the gate to the next millennium." He was one of 15 selected by editors in the Science & Medicine category, and one of only five of those not associated with medical research. "Drexler studies the possibilities of molecule-size machines that might repair cells and build microscopic computers. He chairs Palo Alto's Foresight Institute," the magazine said. Newsweek selected its roster not by identifying "the great and powerful, or the beautiful and celebritous," but rather "personalities whose creativity or talent or brains or leadership will make a difference in the years ahead."
Scientific American continues its gradual retraction of its April 1996 ad hominem critique of molecular nanotechnology with a new posting on the magazine's electronic version on the World Wide Web. The article by contributing writer Alan Hall summarizes work by Al Globus and his colleagues at NASA. (Globus is cochair of the upcoming Fifth Foresight Conference on Molecular Nanotechnology ) "Globus and his colleagues at Ames's Numerical Aerospace Simulation Systems Division are among a growing number of investigators who now believe that atom-scale factories will one day produce new structural materials and advanced computer components, and may even act as tiny repairmen," Hall wrote. He reported on the NASA team's designs for molecular gears, their thoughts about "matter compilers" and the possibilities of "smart materials" that could heal themselves if torn or broken. "There is no question that real nanomachines are probably decades away. But more and more, research is demonstrating that such things are possiblepossibly sooner than most of us think," Hall concluded.
Shortly after the article appeared, NASA issued a substantial press release from its Washington headquarters describing the team's work.
Sky (the monthly in-flight publication of Delta Airlines with 500,000 copies in print and a total readership exceeding 1 million) devoted part of its March 1997 issue to "The Future Of The Future: Peering Into The Nanofuture." Author Robert Ebisch surveyed current micromachines created using lithographic techniques, and dismissed them as "crude monkey tricks compared with what is to come, if the core concepts of future nanotechnology are on target." He extensively quoted Ralph Merkle (computational nanotechnologist at Xerox Corp. and cochair of the upcoming Fifth Foresight Conference on Molecular Nanotechnology), cited Eric Drexler's Engines of Creation, and discussed resistance in the old line science community to its concepts. "A complete scandal," MIT professor Marvin Minsky is quoted as saying of the April 1996 Scientific American article on nanotechnology. Ebisch surveyed recently reported advances in the field, both in protein engineering and mechanical construction, and quoted Minsky that "it shows there's no technical reason why the stuff can't be done."
Popular Mechanics carried a brief item in the Tech Update column of its July 1997 issue about recent fullerene tube developments at Rice University and in Switzerland, and reported that "Nanotube cable...between 10 and 12 times stronger than steel" would make feasible a "space elevator" using a cable suspended from a geosynchronous satellite. "Even if a space elevator is never built, researchers say nanotubes will find their way into a variety of aerospace applications. They could also be used in bulletproof vests, sports equipment and automotive parts. Electrically conductive nanotubes could wire computer chips," the article said.
Knight-Ridder Newspapers science correspondent Robert Boyd authored a solid basic survey of nanotechnology carried by member newspapers in late March. The chain owns major dailies including the San Jose Mercury, Miami Herald, Detroit Free Press and Philadelphia Inquirer, and dozens of smaller papers. Boyd quoted Merkle, Nobel Laureate Richard Smalley of Rice University, Paul Green of Nanothinc, and Foresight executive director Chris Peterson. Nanotechnology is "a very broad field, and it is, in many ways, the ultimate playground" of science, Smalley is quoted as saying. The story also referred to research by Al Globus and his NASA colleagues. It echoed may of the themes in a February 20 story in the Inquirer by Reid Kanaley, which was discussed in MediaWatch.28.
The electronic computer is now 50 years old. Noting the anniversary both of its invention and of the formation of the Association for Computing Machinery, (ACM) Electronic Engineering Times used a review of the first half century to look forward to the next. At ACM's anniversary meeting in San Jose, plenary speakers discussed advancing the state of silicon chip architecture, the physical limits imposed on Moore's Law as circuit dimensions continue to shrink, and even the continued viability of the basic von Neumann computing architecture (in which a sequence of operations retrieves data from a central memory, processes it and then returns the result). Caltech electrical engineering professor Carver Mead discussed analog approaches to computing "inspired by biology's mode of information processing: neural networks. Departing once again from the conventional wisdom, Mead...now predicts that analog/neural systems represent the future of computing," the magazine wrote.
Associated Press Science Editor Matt Crenson wrote in February about a laboratory demonstration of the Casimir Effectvirtual photons that "spontaneously burst into existence like kernels of popping corn and then disappear almost instantly, (which) ought to push two narrowly separated metal plates together." Dutch physicist Hendrik Casimir postulated the effect in 1948, based on quantum electrodynamics, but nobody had set out to verify it until a University of Washington physicist did so last summer. Although too weak to be significant at macro scales, the Casimir force may need to be taken into account in nanomachines, Crenson wrote. "Right now the possibilities of nanotechnology are as endless as the imagination of the field's most enthusiastic proponents. But in the future, nanotechnology will rely on understanding the Casimir force and similar effects," he said.
From Foresight Update 29, originally published 30 June 1997.