come truer to some extent by the invention of the scanning tunneling microscope (STM). The STM was invented as a tool to observe atoms, but it is also useful to manipulate atoms. In fact, several amazing demonstrations have been made of the manipulation of an atom or several atoms by using the STM. However, in most cases, the mechanisms of the manipulation have not been clarified and the reproducibility of it is insufficient. Namely, many scientific and technical hurdles remain in order to master this new technology.

Our Atomcraft project has been organized to make systematic studies to overcome such scientific and technical hurdles and apply the results to the various fields mentioned above. For these purposes, our project has three research groups, i.e., the Basic Analysis, Structure Control, and Surface Measurement Groups.

We attach importance to the close cooperation between experimentalists and theorists, so that we have three theorists in the Basic Analysis Group. The theorists are making calculations of electronic structure, atomic structure, atomic motion, atom transfer by a field, etc. by using a supercomputer. An interesting fact has been found recently. It is usually believed that protrusions in an observed STM image correspond to individual atoms, but this is actually wrong. According to the theoretical calculations of the STM image of the Si(111)rt3xrt3-Ag surface, which agree well with a corresponding observed STM image, each protrusion in the STM image does not correspond to any atom but corresponds to the center of three Ag atoms. That is, observed STM images do not always represent the arrangement of atoms. The experimentalists in this group are constructing a novel apparatus to obtain information on the species and number of atoms transferred by a field.

In the Structure Control Group, they are studying various techniques to manipulate an atom or a group of atoms by using the STM. For this purpose, they have developed hardware and software that can control the motion of an STM tip, the mode of application of the voltage to the tip, etc. in a sophisticated manner. In order to directly measure the electric properties of micromaterials and micropatterns created by the technique, a novel apparatus, which consists of an STM and a low-temperature measurement chamber connected with a sample transfer rod, is now under construction. Such a measurement has already been done with another STM for a double tunneling junction with a liquid crystal molecule as the intermediate electrode, the two outer electrodes being an STM tip and a Pt substrate, and a series of singleelectron tunneling events have been observed at room temperature. In addition to these, a molecular beam epitaxy apparatus equipped with a novel ion scattering spectrometer has been constructed to control the composition of the growing outermost atomic layer at will.

As already mentioned, the key to manipulate an atom or a group of atoms by using the STM is to clarify the mechanisms of the manipulation. The Surface Measurement Group has been studying the mechanisms in cooperation with the Structure Control Group. For example, if we apply an appropriate positive or negative voltage to an STM tip (Ag, W, Au, or Pt) and scan the tip parallel to a Si(111) surface, we can create a desired nanometer etching pattern on the Si(111) surface. On the basis of detailed experiments done by changing various parameters widely, it has been found that the nanometer etching is caused by the field evaporation of surface Si atoms as positive or negative ions depending on the polarity of the voltage applied to the tip.

Atomic-Scale Observation of Material Structures: An Important Role of Theory

Satoshi Watanabe, Basic Analysis Group

Rapid development of experimental techniques in these days such as the invention of scanning tunneling microscopy (STM) has made it possible to observe material structures on the atomic scale. However, the interpretation of obtained experimental results, such as STM images, is not necessarily straightforward. First-principle theoretical calculations are often very helpful to derive a reliable conclusion from such experimental results. In this paper, we would like to demonstrate it by taking the structure analysis of the Si(111)rt3xrt3-Ag surface, which has been a pending problem in surface science for more than 20 years, as an example.

Aono, the director of this project, and coworkers (Ref 39) have recently proposed a new structural model, or the modified honeycomb-chainedtrimer model, for this surface. This model is consistent with most of the reported experimental results regarding atomic geometry, but it has not been determined if this model is consistent with those experimental results that are related to electronic properties. In particular, this model appears to be inconsistent with reported STM images (Ref 40) at first sight; bright spots in the STM images, which correspond to protrusions, are arranged in a honeycomb structure, while the Ag atoms forming the top layer of the model have no honeycomb arrangement at all. We have theoretically calculated (Ref 40 and 41) the electronic structure and the STM image of this model from first principles using the local density functional method. The calculated electronic structure agrees very well with reported experimental results on electronic

properties such as photoemission (Ref 42) and inverse photoemission (Ref 43,44) spectra. The calculated STM images also agree very well with the reported STM images (Ref 40); it has been found that each bright spot in the observed STM images represents neither Ag nor Si atom but corresponds to the center of three Ag atoms.

In this way, we have been able to understand the atomic and electronic structures of the Si(111)rt3xrt3-Ag surface very well by combining the experimental results with the theoretical calculations.

Atomic-Scale Control of Electron Movement

As one of the applications of the incremental charging of a fine metal particle, a single electron transistor was proposed by Likharev (Ref 47) in 1987. After that, many people have tried to realize such a transistor by using microrealize such a transistor by using microlithography techniques; the key to realize such a transistor is to make very small capacitors. Recently, a few groups (Ref 48-50) have succeeded in realizing such a transistor, but it works only at temperatures as low as 4 K. As mentioned above, the STS tip-moleculesubstrate double tunnel junction causes substrate double tunnel junction causes the incremental charging of the molecule. We added the third electrode to this double tunneling junction to realize a single electron transistor of a capacitive type that works even at room

Although many liquid crystal (LC) molecules have been imaged using scanning tunneling microscopy (STM), little has been done on their scanning tunneling spectroscopy (STS). On the other hand, many people have observed the incremental charging of a fine metal particle on a thin insulated layer formed on a metal substrate by using STS. However, the incremental charging, which is due to Coulomb blockade (Ref 45), has been observed only at temperatures as low as 4 K, since the charging energy associated with the incremental charging is smaller than the thermal fluctuation at room temperature. In the present study, we have measured the tunneling current via one of the LC molecules on a Pt(111) substrate by using STS and have succeeded in observing the incremental charging at room temperature (Ref 46). This is interpreted as follows. Since the size of the molecule is very small, the capacitance values between the STS tip and the molecule and that between the molecule and the substrate are so small that the charging energy is larger than the thermal fluctuation at room temperature.

of Pt-Pd deposited on a SiO, substrate act as the source and the drain and the STS tip acts as the gate in the usual FET. This single electron transistor was currently biased and the output voltage was measured. The output voltage versus input voltage characteristic agrees with theoretical simulations by Likharev (Ref 47), although the output voltage has an offset. We attribute this offset to a current through the surrounding media of the single electron transistor.

Atomic-Scale Control of Material Structures

a cluster of atoms on a sample surface using an STM tip. The most recent striking report was done by Eigler et al. (Ref53); they demonstrated that it was possible to exchange a single Xe atom between an STM tip and a Ni surface at will, if the tip and the surface were cooled to 4 K. However, in almost all the cases mentioned above, the mechanism of the manipulation has not been clarified.

We have found that if we place an STM tip close to a Si(111) surface and apply an appropriate positive or negative voltage to the tip, Si atoms are removed from the surface (Ref 54). That is, a hole is created on the surface. If we scan the tip parallel to the surface, a ditch is created on the surface. The diameter of the hole and the width of the ditch can be controlled by changing the magnitude and duration of the voltage (Ref 54). In this way, we can create a desired nanometer-scale etching pattern on the Si surface (Ref 54). As we can suppose easily, this technique will be of great importance in the near future in relation to the realization of novel nanometer-scale devices, huge memories, etc.

In order to clarify the mechanism of the removal of Si atoms from the Si surface, we have observed how the amount of removed Si atoms depends on various experimental parameters (the polarity, magnitude, duration of the voltage applied to the tip, the electron

The scanning tunneling microscope (STM), which was invented by Binnig and Rohrer (Ref 51), is useful not only to "observe" atoms but to "manipulate" atoms. The first demonstration of the latter was done by Becker et al. (Ref 52); they manipulated a single atom on a Ge surface by using an STM tip, although the reproducibility of the manipulation was not necessarily good. After that, many reports have appeared on the manipulation of a single atom or

the surface, and the material of the tip) (Ref 54 and 55). From detailed experiments we have found the following mechanism. If we apply an appropriate positive (negative) voltage to the STM tip, a strong positive (negative) field is created at the Si surface, and the strong field ionizes Si atoms at the surface into a negative (positive) ion and pulls them apart from the surface. That is, Si atoms at the surface evaporate as a negative or positive ion, depending on the polarity of the tip voltage. The field evaporation of negative ions observed

in the experiments is a new phenomenon in that there has been little study about it, although recent theoretical studies (Ref 56-58) indicate the possibility of the phenomenon.

developments in high-precision realtime computer image analysis, and applications of electron holography to various fields ranging from physics to biology.

A New Method for Real-Time Electron Holography

in work function) between the tip and the surface (Ref 55). This is of general importance because it is indicated that a strong field is automatically applied to the sample surface during STM imaging of the surface even if the tip voltage is small. The strong field possibly affects the electronic structure of the surface.

TONOMURA ELECTRONWAVEFRONT PROJECT: EXPLORATION OF MICROSCOPIC WORLD WITH ELECTRON WAVE

Project Director: Akira Tonomura

Although the wave nature of electrons once was evident only in the microscopic region, such as in atoms and molecules, interference phenom

ena have been observable on the macroscopic scale since the advent of a "coherent" field emission electron beam. Its use in combination with electron holography has opened up various new possibilities, since versatile optical techniques can be employed in the optical reconstruction stage of electron holography. An actual example is the optical and numerical compensation for the inevitable aberrations of an electron lens for higher resolution. In addition, the phase of an electron wave can now be employed to observe and measure microscopic fields and matter that have been inaccessible by conventional electron microscopy, in which only the intensity of the wave is observed. This study investigates the basic nature of a coherent electron beam,

A new method for real-time electron holography is proposed and experimentally confirmed to be effective (Ref 59). This method is based on fringe scanning interferometry developed in laser optics.

As the phase difference between an object wave and a reference wave is changed from 0 to 2 pi, a brightness at each pixel varies in a sinusoidal manner. When more than three brightness values between 0 to 2 pi are measured for each pixel, we can calculate a phase value of the sine curve, i.e., a phase value of the object wave passed through the pixel (Ref 60). To realize this method, we controlled an incident angle of the electron beam step by step by changing the excitation current of a beam-tilt coil. The increment of the current was selected so that a movement of biprism fringes was equal to 1 Nth (N= integer greater than 3) of the fringe spacing. In each step, an interference pattern viewed through a TV camera was digitized. From N interference patterns, a phase value of each pixel was calculated.

The time required to obtain phase distribution of the object wave is about 1 minute. This value is two orders of magnitude shorter than that of the conventional method in which a photographic process is employed. The time will be further shortened by a factor of one order of magnitude by improvement of the image processing method (Ref61). The accuracy and image quality obtained in this method have not yet been superior to those obtained in the conventional method. These problems may be caused by inaccuracy in

controlling the phase difference and by the definition of the image processed.

Phase Measurements at Atomic Dimensions

Kazoo Ishizuka, Image Analysis Group

Electron holograms have been mainly processed optically. In this case, the process is troublesome and requires a long processing time. Moreover, it is difficult to obtain quantitative results. To overcome these constraints, we developed a digital system to process electron holograms based on a personal computer. To increase processing power, we installed an array processor on our system. By using this system, we can measure a phase distribution at atomic dimension. We also proposed a new technique to process an electron hologram from a crystal specimen. With this technique, we can obtain phase information from the reconstructed wave, even when the interference fringe spacing is one-third of the usual requirement.

Direct Observation of Atomic Surface Potentials by Electron Holography

Takayoshi Tanji, Measurement and Observation Group

Electron holography allows the direct and clear observation of how each potential of Mg and O atoms located at an MgO crystal surface extends far into a vacuum when an electron beam is incident parallel to the surface in a certain direction, i.e., a profile mode (Ref 62-65).

In high resolution electron microscopy, images near the crystal edge are affected by strong Fresnel diffraction, especially in this profile mode. Clear surface images are for the first time. obtained by the electron phase distribution, which greatly reduces the difficulty in this mode (Ref 66).

KIMURA METAMELT PROJECT: QUEST FOR SOLUTION OF MELT MYSTERIES

Project Director: Shigeyuki Kimura

Most products around us are solids made from melts--glass, steel, aluminum, copper wire, plastics, semiconductors. All are made by cooling melts. Melts and the ways melts are cooled are critically important. Silicon semiconductors must be grown from melted silicon, but impurities enter while the silicon is molten. Molten glass must be cooled at just the right rate in order that the resulting glass has good properties. Melts, their microstructures, and their internal movements are critically important, yet little is known about them. Everyone knows that molasses is slow in January, but knowledge about why is superficial. Since the attractions between molecules in a melt are stronger than in air, we know that the melt must have structure. Yet, because the melt is fluid, it is hard to study. It does not stand still for pictures. Despite years of research, our knowledge of melts is only partial and mainly founded on mathematical models.

melt structure analysis has been developed using a technique called Rayleigh scattering. Recent research has attempted to combine computer modeling with radiographic observation of melts.

This project will focus on the changes that occur in melts over time, will analyze the changes in melt structure and behavior, and will explore new ways to grow crystals. We will use the melts of semiconductors and oxide materials and follow their changes with sophisticated methods including x rays. Other quick and precise methods will be developed to determine the causes of changes. These will include measurements of viscosity, surface tension, density, and heat conduction, as well as research on the measurements themselves. Direct observations of flow pattern in melts will be combined with simultaneous computer modeling.

Our increased understanding of the microstructure and ordering of melts is expected to lead to new materials and new processing technology.

NAGAYAMA PROTEIN ARRAY PROJECT: A TECHNOLOGY EMERGENT FROM BIOSYSTEMS

Recently, we have learned that melts change with time, even as they are kept Project Director: Kuniaki Nagayama at a constant temperature and in a constant environment. It is known that the crystalline state of an aged melt differs from that of a fresh melt. The viscosity of a melt depends not only on its temperature but also on how that temperature was reached. The slowness of molasses depends not only on how cold it is but also on how fast it was cooled and on how long it has been cold. It is assumed that this is due to differing structures, but our knowledge does not go beyond this assumption.

The scientific tools for studying melt structure are just now being established. There are new technologies using x rays and neutron beams. For melts that can be studied with light, a new method of

The Protein Array Project is attempting to establish a universal technology of fabricating 2D protein crystals with desired molecular alignment and crystal form of excellent quality (protein array). The basic strategy is to explore techniques that can implement the biological principle for making the macromolecule and proteins and then assembling them into intercellular devices, called organelles ("biological technology"). Component parts should then be automatically assembled to give final forms through mutual recognition (specific interaction manifested through structural information). The direction of such structuring into a 2D

manner can be achieved by mutating amino acids on the protein surface and reshuffling domains as well as introducing a 2D substratum, which is extremely well defined, using “human technology." This is a technology emergent from biosystems, arising from a combination between "biological technology" and "human technology."

The project is composed of three research groups: Array Design Group, Array Engineering Group, and Array Characterization Group. In addition to these domestic groups, a research group from the Laboratory of Thermodynamics and Physico-Chemical Hydrodynamics, the University of Sofia in Bulgaria, participates in the project. Each group is individually unique in its academic major. The point of this kind of organization in terms of promoting the new technology is extensive and there is active fusion of a variety of expertise. The Array Design Group aims to design a protein array based on analysis of interprotein interactions. The major techniques used are computational physics, computer graphics, and nuclear magnetic resonance (NMR) spectroscopy. The Array Engineering Group, which is composed of molecular biologists, protein biochemists, and organic chemists, seeks mass production of pure protein specimens underlying the protein array. The Array Characterization Group is required to fabricate an excellent protein array and to define its structure using modern and accurate morphological techniques like transmission electron microscopy (TEM), scanning tunneling microscopy, and scanning atomic force microscopy. It is also expected to develop an ideal substratum not only for array fabrication but also for transferring the array on it, which may possess potential as a prototype device utilizing a protein array. The Bulgarian team specializes in the process of lattice formation since the array formation is directed firstly by lattice formation as well as the following crystallization.

The Array Design Group has been engaged mostly in analysis of electrostatic interactions that likely contribute predominantly toward the crystallization process: computational simulation of the 2D crystallization process of the poker chip model and analysis of atomic interactions found in the proteinprotein interface of real 3D crystals of proteins. The simulation experiment of the poker chip model, in which edge both positive and negative charges are hexagonally distributed, succeeds in designing a uniformly aligned array, trimmer-unit array, etc. The analysis of the 3D crystals illustrates that electrostatic pairing, such as salt-bridge and hydrogen bonding, principally governs crystallization of proteins. The Array Engineering and Characterization Groups have shown that the quality of array depends significantly on the purity of the protein specimen using horse spleen ferritin. They also obtained a preliminary clue as to possibly controlling the crystal form by examining ferritin from different species, in which not many numbers of amino acids are different from each other. Using polystyrene lattices the Bulgarian team has developed a basic and very universal method that enables governing the process of lattice formation of submicron particles by adjusting water level of the particle suspension laid on a 2D substratum. These research products are to be combined and reorganized as set advanced research subjects in order to approach the process toward the final goal of the project.

SHINKAI CHEMIRECOGNICS PROJECT

Project Director: Seiji Shinkai

Introduction

Chemical reactions have conventionally been explained in terms of the "collision probability" among molecules governed by thermodynamics. In

contrast, reactions taking place in living organisms can be grasped as “inevitable or necessitated outcome" rather than "probability." These two positions are totally different: the former sees chemical reactions as phenomena explainable by molecular dynamics, while the latter reactions are characterized as proceeding by way of "recognition process."

A primary objective of our research in this project is to analyze the "recognition process" from a chemist's viewpoint and to construct this process at the molecular level. This can be accomplished only if we establish a powerful plished only if we establish a powerful methodology and the tools to implement it. Fortunately, now we have "calixarenes," which have increasingly been attracting the attention of many chemists as "the third inclusion compound" following cyclodextrin and crownethers. Evidence is mounting that as single compounds calixarenes can be designed as elements capable of recognizing atoms and molecules and also that as molecular assemblies they are capable of exhibiting very unique and interesting functions.

Thus, the project consists of three groups: (1) the Atomic Recognition Group, (2) the Molecular Recognition Group, and (3) the Intelligent Assembly Group. These groups will be working in cooperation, by using calixarenes and related compounds as the "tools," to answer the question, "What is recognition?"

Research Activities and Results So Far

One year has passed since the Shinkai Project was launched. As of September 1991, the project consists of 12 researchers. While the project's research activities are still generally at the beginning stage, we have obtained some interesting and novel results, as summarized below.

(1) Atomic Recognition Group: Various systems have been designed as elements for recognizing metal ions, in which a site for chemically binding a metal ion and a reporter site are introduced in a calixarene. The reporter site is for outputting the binding as a physical signal, such as fluorescence or luminescence, or an electrochemical signal. Some systems have excellent potential as elements capable of recognizing metal ions.

(2) Molecular Recognition Group: Studies are currently being conducted mainly with saccharides as guest molecules. Various phenyl borates are synthesized and used as the hosts for studying selective recognition of the guest saccharides.

Methods for the preparation of novel calixarenes are being established, in which the OH groups are selectively substituted with amino groups or deoxylated. The resulting calixarenes are promising elements for molecular recognition.

(3) Intelligent Assembly Group: Studies in progress are aimed at constructing higher structures by assembling the modified calixarenes to develop new functions that cannot be obtained by a single calixarene. The assembly methods being tried include gellation, polymerization, liquid crystallization, and micelle formation. It has been found that certain calixarenes and crownetherbearing compounds act as gellators of organic fluids, and studies are being done to elucidate the gellator mechanism in connection with the recognition of metal ions.