杏吧原创

Birmingham

WITH 92 natural elements to choose from, anybody trying to create a new
material should be spoilt for choice. Or so you might think. In fact, this menu
has proven rather restrictive, and scientists would love to be able to extend
it. So here鈥檚 a thought: the properties of a material don鈥檛 depend just on which
atoms it鈥檚 made from, but also on the way those atoms are arranged. Groups of
atoms can behave very differently from individual atoms of the same element. So
imagine we could create a new type of building block for matter out of different
sized groups of atoms. If these blocks existed, we might call them atomic
clusters鈥攁nd they do.

Researchers can now produce tiny particles containing a specified number of
atoms, usually of the same element. What鈥檚 so exciting is that the properties of
these clusters depend critically on the number of atoms they contain, so we can
tune the properties of our building blocks. Until recently, most of the research
has focused on clusters that are free to move in the gas phase. But for the past
two or three years, several groups have been depositing these clusters onto
surfaces in order to make thin films and new materials with unique properties.
The first results are full of promise.

My own eyes were opened to the world of atomic clusters while browsing in the
library at Cornell University in New York state, during a short visit in 1990. I
read that Walter Knight and his colleagues at the University of California,
Berkeley, had studied clusters of alkali metals, such as sodium and potassium,
and found that certain clusters containing 鈥渕agic numbers鈥� of atoms were
especially stable.

Adding atoms one by one to a cluster was like stepping along the periodic
table, with the stable, magic-size clusters following the sequence 2, 8, 20鈥�
What鈥檚 more, these numbers were the same as the magic numbers that describe
stable configurations of protons and neutrons inside the atomic nucleus (鈥淗eart
of the Atom鈥�, Inside Science, New 杏吧原创, 8 July 1995).

Hot metal

So here was something rather deep. The explanation was that electrons in the
clusters can move between atoms and occupy discrete energy levels. The magic
numbers correspond to the number of electrons needed to fill successive shells
of electrons, just as in a single atom.

It turns out that Knight鈥檚 magic numbers were not the first to be found in
clusters. In 1981, Klaus Sattler and colleagues at the University of Konstanz in
Germany found a set of magic numbers in the mass spectra of inert gas clusters.
In this case, the magic numbers (13, 55, 147 . . .) came not from filled
electron shells but from stable icosahedral shapes, each one made by adding a
shell of atoms around its predecessor.

These discoveries were made possible by the development of cluster beam
sources. One of the most versatile of these is the gas condensation cluster
source. The principle is simple: you heat a piece of metal to create a hot
vapour, which is then mixed with cold, inert gas so that the metal condenses to
form droplets鈥攐r clusters. The process is completely analogous to the way
water droplets form when hot steam hits a cold window. Another type of cluster
source depends on 鈥渟puttering鈥�, in which atomic-scale chunks of material are
broken off a target by bombarding it with an energetic beam of ions.

These sources produce clusters with a whole range of sizes. To separate out
clusters of a particular size, they are first ionised by passing them through a
plasma, to knock off an electron, and then deflected by an electric and a
magnetic field (see
Diagram). Clusters with the same charge but different masses
are deflected through different angles.

Once you can make a beam of clusters, it doesn鈥檛 take a huge leap of
imagination to envisage laying down a film of them on a surface 鈥攖he first
step in making new materials. Indeed, as people were discovering the magic
numbers in clusters, others were depositing clusters on surfaces to produce
鈥渘anostructured鈥� materials. Even though these researchers didn鈥檛 have precise
control over the size of their particles, they produced some remarkable
materials, such as ceramic materials that can be stretched like
plastic鈥攂ut that鈥檚 another story.

One of the first examples of depositing size-selected clusters on a surface
came in 1985 from Pierre Fayet and his colleagues at the Swiss Federal Institute
of Technology, Lausanne (EPFL). They explored the basis of black and white
photography, and specifically how an image forms in a silver bromide film. When
light hits a silver bromide crystal, it leaves behind a few atoms of metallic
silver. If these islands of silver are big enough, they provide points of attack
for chemicals in the developer, which reduce the whole crystal to silver.

Fayet and colleagues asked the question 鈥淗ow many atoms of silver do you need
to see a bright speck after the film is developed?鈥� They deposited silver
clusters of different sizes onto an ordinary photographic film, then developed
the film in the usual way. The answer to their question was striking: you need
four silver atoms to develop the film. Three isn鈥檛 enough. The size of the
cluster is crucial.

After I found out about clusters at Cornell, my research group鈥攖hen at
the University of Cambridge鈥攕tarted to build cluster beam sources with the
specific aim of depositing clusters on a surface. Although, we came from a
background in surface science, controlling these newfangled cluster sources took
some getting used to. The first device we got to work was a gas condensation
source with a few frills, with which we could make, among other things, silver
clusters containing between 20 and 300 atoms. These clusters measure around 1 to
2 nanometres across, but when we deposited them on graphite and looked at them
with a scanning electron microscope they were always about 10 nanometres
across.

So what was happening? We found that if you drop clusters onto a surface
gently, they can diffuse sideways and join up with other clusters to form bigger
particles. The uniform size of our particles seems to have arisen from
interactions between the particles and the graphite surface.

To stop the deposited clusters coalescing, we found we had to fire them with
more energy so that they implanted weakly where they landed. We had thought it
would be the other way round, but several people have since confirmed this
result. Going to the opposite extreme, if a cluster hits the surface too hard,
it is likely to break up on impact. In between, there seems to be a window of
energies where we can deposit an array of size-selected clusters.

Soft landing

Ideas about cluster deposition haven鈥檛 emerged just from experimental
studies. In 1993, Uzi Landmann at the Georgia Institute of Technology in Atlanta
used computer simulation to predict an elegant method for 鈥渟oft landing鈥� a
cluster on a surface. Landmann calculated that the energy of a cluster
approaching a surface at speed could be dissipated by laying down a thin film of
condensed inert gas, such as neon. The film would act as a braking layer,
bringing the clusters to a standstill before they hit the crystal
surface鈥攋ust as water stops you cracking your skull on the bottom when you
dive into a swimming pool. Within the past year, experiments by Wolfgang Harbich
and colleagues at the EPFL have verified Landmann鈥檚 predictions.

So, once we know how to deposit clusters, what can we make? And is it any
different from what we could make before? The answers are 鈥渓ots of things鈥� and
鈥測es, definitely鈥濃�攅ven though most of the materials produced to date use
only a coarse selection of cluster size.

One of the pioneers of making materials from clusters is Alain Perez at
Claude Bernard University in Lyon, France. Perez and his colleagues vaporise
their target with a laser before condensing out the clusters in an inert gas.
This, incidentally, is the kind of cluster source that Robert Curl, Harry Kroto
and Richard Smalley used in their discovery of buckyballs, for which they
received last year鈥檚 Nobel Prize for Chemistry.

Perez operates at one end of the energy spectrum for depositing clusters
鈥攈e lands clusters very gently onto surfaces to create thin films. He has
found that the physical properties of those films, such as the hardness,
electrical conductivity and optical absorption, depend on the size of the
clusters. Films made from clusters with 900 carbon atoms behave like graphite,
for example, while films made from 20-atom clusters behave like diamond.

The Lyon group has also made thin films from silicon clusters containing
around 50 atoms. When irradiated with a laser, these films emit a strong red
light which is not seen in either crystalline or amorphous forms of silicon.
鈥淭his is very important since the whole of microelectronics is currently based
on silicon,鈥� says Perez. 鈥淭he production of luminescent silicon allows silicon
to enter the field of optoelectronics too.鈥�

Researchers are already striving to create silicon devices that can process
not only electronic signals, but also signals carried by light. Most of this
work has focused on 鈥減orous silicon鈥�, which shows a photoluminescence similar to
that of Perez鈥檚 films. Porous silicon is riddled with microscopic holes, and is
made by electrochemical etching. In theory, cluster deposition should offer more
control over the structure and properties of the film.

A second intriguing result from Perez鈥檚 work is that the electrical
resistance of some of his materials changes dramatically when exposed to
magnetic fields, a phenomenon called giant magnetoresistance. GMR is attracting
intense interest from the computer industry, which sees it as a way to improve
magnetic recording of data. The films that exhibit GMR are produced by embedding
cobalt clusters in a silver matrix. Perez found that the strength of this
response depended on the density of clusters in the matrix, and hence the
spacing between them. 鈥淐lusters are nice nanomagnets which can lead to new
magnetic or magneto-optic systems,鈥� he says.

At the other end of the deposition energy spectrum stands Hellmut Haberland
at Freiburg University in Germany. Haberland has pioneered a technique he calls
energetic cluster impact. 鈥淚f you throw a snowball with supersonic velocity
against a wall, you get a deposit which adheres very well,鈥� he says. 鈥淚f you
throw lots of snowballs you will get a very compact snow deposit. We have just
scaled this idea to nanometre dimensions. Today鈥檚 conventional methods to
produce thin films have deficiencies which can, at least in principle, be
overcome by cluster deposition.鈥�

Stuck on Teflon

Haberland deposits large clusters, typically containing several thousand
atoms, to make films that are smooth and contain very few pinholes. They are
generally better in these respects than films made by conventional methods, such
as sputtering or evaporation. Also, they can often be made at room
temperature.

Unlike traditional methods, Haberland鈥檚 technique can produce metal films
鈥攐f molybdenum, for example鈥攖hat even stick to nonstick Teflon. He
can also direct energetic copper clusters to fill the micrometre-sized contact
holes that are created during the fabrication of integrated circuits. This is a
big problem in microelectronics as device dimensions get ever smaller, and the
key advantage of Haberland鈥檚 repair process is that it can be done at
temperatures of around 80 掳C. Higher temperatures can ruin the delicate silicon
structures. Haberland鈥檚 goal is to develop his laboratory method into an
industry standard.

Another application of clusters is being pioneered by my group in Birmingham
in collaboration with Tetsuo Tada and Toshihiko Kanayama at the Joint Research
Center for Atom Technology in Tsukuba, Japan. This sees clusters not as building
blocks but as nanometre-scale tools in the fabrication of semiconductor devices.
The idea is simple: we deposit an array of silver clusters of specific size and
energy onto the surface of a silicon wafer, and post the samples to Japan. In
Tsukuba, Tada and Kanayama expose the wafers to a hostile plasma that eats away
the wafer鈥� except where it is protected by the metal clusters. The result
is an array of nanometre-scale silicon pillars, looking like a set of ancient
Greek columns. So far the pillars have a diameter of about 10 nanometres and a
height of about 100 nanometres.

These structures could be of huge technological significance. Silicon鈥檚
optical properties change dramatically at the nanometre scale, and theory
predicts that the colour of light emitted by silicon structures less than 10
nanometres across should depend dramatically on their size. So an array of
silicon pillars that emit red, yellow and blue light could one day be used to
make a colour TV screen. The goal is to make pillars with very accurately
controlled diameters of, say, 3, 4 or 5 nanometres.

What does the future hold for cluster deposition and new materials? Arguably
the first challenge is to rework the kind of experiments on clusters that Perez
and others have carried out, but with tighter control over cluster size. Precise
size control is essential if we are going to fully exploit the tunable
properties of cluster materials.

Another promising line of research is to coat the clusters before depositing
them. This would contain the appetite of clusters made from reactive elements
such as iron, which would otherwise react or coalesce as soon as they landed on
a surface. There has already been progress in this direction by researchers such
as Jess Wilcoxon at Sandia National Laboratory in New Mexico, who has tried out
a variety of organic molecules as cluster coatings.

The ability to position clusters precisely on a surface, rather than
depositing them randomly, would also bring big benefits. Once this is possible,
Perez sees a future for cluster deposition in making quantum devices for use as
memory chips. One fascinating possibility here is to use atomic-scale features
on a surface to capture and organise clusters. When clusters land gently on a
surface and diffuse across it, they tend to find strong binding
sites鈥攗sually at steps or ridges on the surface. The clusters collect in
lines along these features, opening up the possibility of making nanometre-scale
wires and other components for tomorrow鈥檚 nanoscale electronic circuits.

Clusters may also have a part to play in biology and medicine. A European
Consortium on Nanomaterials, chaired by Heinrich Hofmann of the EPFL, has been
formed to promote the development of new products based on nanometre-scale
technology. Hofmann believes that small particles鈥攚hich have a large
surface to volume ratio鈥攃ould be used to deliver vaccines. By attaching a
vaccine molecule to the surface of a cluster a 鈥渓arge amount of vaccine can be
compacted into a small space鈥�, says Hofmann. 鈥淣ew oral vaccines can be designed
by adsorbing the vaccines on small particles and pressing them into
迟补产濒别迟蝉.鈥�

Hofmann sees the science of small particles as underpinning a range of new
materials. 鈥淭he major benefit of this new technology is the possibility to
design materials with unique properties which could solve some of the present
problems related to specific applications in optics, electronics, catalysis,
mechanical materials and so on,鈥� he says.

To solve these problems, we must be able to produce ever more intense beams
of size-selected clusters, and improve our understanding of the basic processes
going on in cluster deposition. Once we have this knowledge, cluster technology
promises tocreate a rich variety of valuable new materials.