My Research: History, Current Interests and Aspirations

Introduction

I am interested in the atomic-scale mechanisms which control epitaxial growth at the atomic scale and in the formation of atomic-scale structures by self-assembly, and the characterisation of their properties. I have over a decade's experience in atomic-resolution imaging and characterisation of this and other semiconductor surfaces, specialising in elevated-temperature imaging required for the observation of dynamic diffusion, nucleation, and growth processes. In the last few years, I have developed, and applied for a patent for, a method called "Nanoline Templating" for self-assembly of epitaxial nanowires (1.5 nm x 300 nm) and arrays of metal nanoclusters (0.6 – 3 nm) on the technologically important Si(001) surface. I am currently working on the growth of metallic atomic chains, and their measurement by Spin-Resolved Scanning Tunnelling Microscopy. In the longer term, my aspiration is to look beyond the fabrication of isolated single-nm wires, to the ability to put together complete nanoelectronic devices, comprising a mixture of components, which may be "prefabricated" ex-situ, self-assembled in-situ, or defined by lithography.

Current Position

I moved from Japan back to Europe in 2007, to a new position at the University of Geneva, where I am working with Prof. Christoph Renner at the Dept. of Condensed Matter Physics. We are building a novel form of STM, known as "Mott-STM" to measure the magnetisation of samples at near-atomic-scale. The basic idea comes from a theoretical paper DOI: 10.1103/PhysRevLett.79.4593 by P. Bruno. While this promises to give the ability to measure the magnetization of structures with STM-type resolution, in practice, implementation of this speculative idea will take some time, and may not ultimately be successful.

Taking advantage of this novel STM, if it is successful, but not dependent upon it, I am continuing the experimental thread from my time as a research fellow in Nanomaterials at the International Centre for Young Scientists (ICYS), which is part of the National Institute for Materials Science in Japan. At ICYS, I developed the use of self-assembled Bi nanolines as nanoscale 1D templates for other materials, such as metal to make conductive wires, magnetic or optically-active nanoparticles, active molecules et al. I have been using a variable-temperature Scanning Tunnelling Microscope, which is capable of imaging from 300 K up to around 1000 K and of depositing material in the STM chamber. The abilities to vary the sample temperature and to deposit in situ are crucial to my studies, as it means that I am able to observe dynamic processes directly, rather than cool to room temperature, and try to work out what has been going on post-hoc. However, having reached the stage of reproducible fabrication of these structures, I need to move on to detailed characterization of their properties, and for this, operation below room temperature is necessary. We have just finished installing a new LT-STM system here in Geneva, which can operate down to liquid He temperatures (4.4 K). A growth chamber is attached for growth of the Bi nanolines or other nanostructures and deposition of small amounts in the STM chamber will also be possible. Thus detailed electrical characterization of the properties of the nanoclusters, and nanowires which I have grown will therefore be possible. Furthermore, we intend to implement the Mott-STM scheme outlined above into this system, so as to be able to measure the magnetic properties of many systems on the near-atomic-scale, including the nanostructures which I have been fabricating.

Major Research Achievements

1DV defect structure determination

Si(001) missing dimer defect The first topic that I worked on was the determination of the structure of the 1DV missing dimer defect on the Si(001) surface. While a relatively simple system, it was the prototype for a highly successful style of experimental-theoretical collaborative working, which has served me well in many more complex situations. The initial observation was that certain missing dimer defects had a brightening around them when observed at very low biases, close to the Fermi level. We modelled 3 different structures for the 1DV, looking at the electron density integrated up to different biases, as a simulation of the STM. Only one model matched the STM in that it showed a brightening effect at low biases, and thus we were able to conclude that we had found the right structure. We have subsequently used this same method with the Bismuth nanolines, and we were pleased to notice that it had been picked up by a larger study of defects on Si(001).

Measurement of Hydrogen diffusion

H/Si(001) The goal of my DPhil was to observe dynamic processes at elevated temperature in the STM with atomic resolution. Diffusion of atoms is an obvious example. An extremely important system to measure was that of atomic hydrogen on Si(001), as the diffusion rate of hydrogen has an effect on many CVD processes. Therefore I measured hydrogen diffusion on Si(001) by the very crude method of taking sequences of images, and counting how many atoms had moved, and how far. The easiest method was to find the temperature at which the atoms are moving faster than the tip scans, in which case they break up into streaks. I also looked at the saturated surface, where all the hydrogens have paired up on the Si dimers. I found that the gaps in the hydrogen layer changed position over time. This was a direct observation of vacancy diffusion mechanism which is common in materials - it is easier to think about the holes moving around than the atoms which are filling a hole and at the same time creating a new one. We determined a value for the diffusion barrier for single-atom motion, and for pairwise motion, finding a value of 1.65 eV for single-atom diffusion and ca. 1.8 eV for pairwise diffusion, which at the time were the only experimental numbers for these processes. More details about this can be found on my hydrogen pages.

Hot STM during disilane/germane growth

Si/Si(001)Moving on from the measurement of hydrogen diffusion, I used similar techniques to observe the nucleation and growth of new layers of silicon from disilane (Si2H6), and put together a pathway from the initial adsorption up to the nucleation of small islands of silicon (see my silicon pages). This work is described in the first of our two big Si papers (see my publications page). The next step was to grow whole layers of silicon, and investigate the dynamic processes involved. We observed that nucleation was almost entirely heterogeneous, for example, occurring only at antiphase boundaries between previous-layer islands. We also were able to observe differences between the morphology from solid-source growth and gas-source growth - the difference being due to the presence of hydrogen. This work is described in our second big Si paper.
Finally, we started work on investigating Ge/Si(001) growth, even taking a 4-hour movie of the growth of about 10 layers of Ge, which allowed us to watch as 3D features, such as pyramidal 'hut clusters' nucleated and formed. However, this work was continued by Ilan Goldfarb, as I had come to the end of my three years by this stage.

Linked RHEED oscillation decay rate and surface morphology

RHEED Oscillation DecayIn 1996, I was offered a position at UC, Santa Barbara in the Chemical Engineering Department. There I spent my time performing atomic-scale experiments on semiconductors, looking at the adsorption and surface reactions of gases. This work is done using a variety of techniques including Reflected High-Energy Electron Diffraction (RHEED), Low-Energy Electron Diffraction (LEED), Temperature-Programmed Desorption (TPD), Electron Energy Loss Spectroscopy (EELS) and scanning tunneling microscopy (STM).
I became involved in part of a DARPA project called the Virtual Integrated Prototyping (VIP) project which was a collaboration between the Maths Dept. at UCLA, HRL Labs in Malibu, Harvard and Imperial College. The aim of the project was to use a variety of experimental and modelling techniques to control the MBE growth of AlSb on InAs, with the aim of growing a Resonant Tunneling Device (RTD). RTD's can be used in communications, where they can switch on and off extremely rapidly, at terahertz frequencies, enabling data to be sent many times faster than at present (gigahertz frequencies). By controlling the growth, we hoped to make RTD's which have better performance. These devices are very sensitive to surface roughness on the atomic scale, and so we were trying to achieve minimum roughness in our growth, and to do so with much higher consistency than at present, so that the yield of usable devices is much higher. I proposed that we use the As flux as the control medium, as this could be changed rapidly via the valve on the As cracker. My major research achievement was to demonstrate that close control of the As flux in an MBE chamber was possible, and that small changes in the As flux affected InAs growth morphology considerably. I made a connection between the RHEED oscillation decay rate and the surface roughness (see my InAs page for more details). Moreover, by looking at trends in the decay rate with various growth parameters, I found that the trend was in a direction opposite to that which our initial models had expected, a result which was of crucial importance to the project.

Self-assembled Bismuth nanolines

A Bi nanoline While in Oxford, I was working with Dr. Miki, who was visiting from Japan, on a project concerning the use of surfactants to grow flat Ge/Si layers. He was using Bismuth as the surfactant material, and so we were looking at the growth of Bi on Si(001). Usually he would work all night, and one morning when I came in, I decided to have a look at the sample he had left annealing all night. To my surprise, long, straight features, much longer than the maximum picture size could be seen on an otherwise very flat surface. These were my first sight of the Bismuth nanolines. We abandoned the original project goal, and turned our attention to these fascinating features. Several papers resulted, including a model for the structure of the nanoline, which later turned out to be wrong. Later in 2000, I was invited by Dr. Miki to come to work with him in Japan. In March 2001, I was awarded an STA (now JST) Fellowship. My research formed part of a larger project in Japan, headed by the Hitachi Advanced Research Lab, which was trying to develop many aspects of nanoelectronics. We initially wanted to use Bi nanolines as nanowires, but as they have a band gap larger than that of the surrounding silicon, I instead proposed to make use of them as templates for the deposition of other metals,to produce nanowires of a variety of electrical properties. Again, computational modelling has proved invaluable. Having realised that our previous structure was not correct, I set about trying to determine the true structure of this feature. I therefore branched out into modelling, learning a tightbinding code called DensEl which allowed me to run atomistic simulations to test hundreds of candidate nanoline structures on my desktop. We determined the structure of the nanoline, and showed that the nanolines are resistant to attack by hydrogen and oxidation. These were important first steps towards my "nanoline templating" proposal. We published four papers on the subject of Bi nanolines. We have recently written a review article, discussing all the work which has been done on the Bi nanoline, by the various groups, trying to put together a coherent picture of what is known about these structures. The article also reviews the rare-earth silicide nanowire systems, which have attracted a lot of attention.

Orbital structure of metallofullerenes

A C60 molecule In April 2002, I returned home to England, and was offered a position at Oxford, this time working with the Quantum Information Processing(QIP) LINK project, which involved groups at Oxford, the Hitachi Cambridge Lab, and Queen Mary College London. The aim was to be able to choose a suitable "peapod" nanotube (that is a nanotube with fullerenes inside) and then make connections to the outside world, so as to use the peapod nanotube as the essential part of a quantum computer. As part of this project, I have been concentrating on low-temperature STM of fullerenes, particularly metallofulleres, such as Nd@C82 and Sc@C82. They have unpaired electron spins, which make them ideal candidates for qubits. We looked at how they form on the surface, and also looked in detail, with the aid of DFT modelling, at the shapes of the molecular orbitals. One outcome of these experiments was that we have been able to make matches between the STM and the DFT simulation of the molecular orbitals of Nd@C82, and thereby show that we had two isomers present in the sample, and that they could be distinguished. We hoped to do the same with Sc@C82 and possibly Er3N@C80, which are very interesting molecules, and potentially has 3 unpaired electrons inside each fullerene. Again it is thought that two isomers are present, which might explain the very strange electronic contrast that we observed.

Using Bi nanolines as templates for nanowires

In on haiku In November 2004, I moved back to Tsukuba, to work at a new Institute within the National Institute of Materials Science(NIMS), called the International Centre for Young Scientists, or ICYS. The main objective of the ICYS was to bring together young scientists from all over the world and provide them with space to pursue their own research goals. However, a secondary aim seemed to be to bring the cold wind of Western scientific ways blowing through the dusty corridors of NIMS! I worked with K. Miki and Dave Bowler again, which proved very fruitful.

I call my idea Nanoline Templating. This is essentially a method for massively parallel nanoscale materials synthesis without patterning or lithography. The basic idea is to use the Bi nanolines as templates for the formation of nanostructures of other materials. As they are very long, very straight, well-ordered, 1.54 nm wide, and do not touch each other, they are a good candidate for this. There are many more details about how the templating process on the nanoline template page.

I looked at metal deposition onto the template to produce structures with interesting magnetic and plasmonic properties. I tested many elements, including Gr. III metals such as Al and In, noble metals - Pd, Pt, Au, Ag, transition metals - Fe, Co, which may be magnetic, and I am tried rare-earth metals such as Er and Y. All work to some extent, but each has individual characteristics. The Gr.III metals are the only ones which make wires, so far, and I can make single-crystal wires which are around 1 nm wide, and well over 100 nm long. The other metals that I have tested so far form nanoclusters. I am hoping that Er and Y may actually replace the Bi and form conducting nanowires at high temperatures, but we shall see!

Looking beyond these simple cases, there are a large number of possible species to deposit onto the template, from molecules (such as conducting molecules, or fullerenes), preformed nanoparticles, layer-by-layer growth of compound nanostructures, e.g. using alternate exposures of Al and NH3 to form AlN nanowires, or mixtures of Fe and Pt to form magnetic particles with much stronger magnetisation, but this is just scratching the surface of the possibilities.

Electronic and Magnetic Properties of Metal Atomic Chains

Au on haiku Since June 2007, I have been working to help develop a novel form of spin-resolved STM, and to use it to study the electronic and magnetic properties of nanowires. A new UHV LT-STM system has just been installed, August 2008, which will allow us to begin this work. Our main goal is to fabricate single atomic chains of metal atoms on non-interacting substrates, particularly the Bi nanoline templates. This will involve deposition of metal below room temperature to block the formation of metal clusters as has been seen after deposition at room temperature. We shall investigate the properties of the atomic chains using spin-resolved STM. Depending upon the choice of metal, and upon the interaction with the substrate, the spins on the metal atoms may arrange ferromagnetically, or antiferromagnetically. In either case, their spin properties will be very interesting. It is hoped that these chains will be much closer in properties to the isolated atomic chains typically modelled, as the interaction with the relatively inert Bi nanolines, and the H-terminated Si(001) surface will be minimal. Use of the nanoline templates will allow for a large population of atomic chains of different lengths and configurations, which we can use to our advantage to study many different chains at once. Also the self-assembly process is inherently scalable to large numbers of metal chains, which would be of use in possible future devices such as magnetic storage, interconnects and QIP.