by Dr. Frances Ross, materials scientist at
IBM Research
Dr. Frances Ross at the IBM Thomas J Watson
Research Center. Photo credit: Chris Ramirez
for The New York Times
“Whiskers” have been recorded in natural ores
since the 1500s. These crystal formations indeed look like a cat’s whiskers, but
function as wires. And since Bell Labs in the 1960s began to grow whiskers deliberately
out of many different substances, materials scientists have been molding them
into everything from photovoltaic devices, transistors, sensors, solid state
lighting and batteries. At IBM Research, with colleagues from academia and the Brookhaven
National Laboratory, we have developed a way to make these structures grow in
an electron microscope. By recording the atoms self-assembling
to form the whiskers – renamed in modern fashion as nanowires – we hope to understand how they grow and how to tune the
growth conditions to build nano-devices. The term nanowire describes any crystal that
is thin in diameter, down to tens of nanometers, but with a length maybe
hundreds or thousands of times greater. The material that the nanowire is made
of, and its exact dimensions, determine its performance. For example, how solar
energy is absorbed, in the case of photovoltaics. Or how lithium moves along a nanowire,
in the case of rechargeable batteries. And how magnetic fields arrange
themselves along its length, in the case of IBM’s racetrack
memory. The big bang in these tiny structures comes from the fact that at
nanoscale dimensions, even conventional materials can behave in remarkable and
unexpected ways.
One exciting opportunity for nanowires is as
building blocks for transistors in microelectronics. While getting to the 7
nanometer node is heroic, the lithographic process conventionally used to
define the circuits is ultimately limited. Using self-assembled nanowires as
the cores of the transistors can provide a pathway to smaller circuits, and
perhaps even new designs that need less power to perform computations.
Our experiments started with the goal of understanding
how the atoms spontaneously arrange themselves into the simplest sorts of nanowires,
those that are made up of only one material. We chose our favorite element,
silicon, the semiconductor that forms the basis of microelectronics. The nanowires
grow with the help of a catalyst. So, we start with a flat wafer of silicon, add
a couple of layers of gold atoms and heat until the gold and silicon react chemically
and form tiny liquid droplets. The reaction takes place at 370oC,
and results in something that looks a lot like a pane of glass covered by raindrops
– at the nano-scale.
A silicon nanowire at 5nm and 20nm.
The trick is then to flow disilane gas, which
contains silicon atoms, across the hot, droplet-covered silicon surface. When a
molecule touches the surface of the silicon wafer, it bounces off and nothing
changes. But when a molecule hits the surface of the gold droplet, it sticks
and breaks apart, and its silicon atoms enter the droplet. As more silicon
enters a droplet, the excess settles out as solid silicon beneath the droplet.
Think of gradually adding spoonfuls of sugar to your coffee: excess sugar eventually
settles at the bottom of the cup. In the case of the nanowires, the settling does
not form disordered sludge, but instead a perfect layer of silicon atoms. The
droplet stays at the top, and the nanowire keeps growing as long as we keep the
gas flowing. The photo shows a nanowire during growth: the liquid droplet,
labeled Au-Si, sits on top of the nanowire.
Watching this process in our electron
microscope, and later in a microscope at Brookhaven National Laboratory with a
high-speed (400 frames–per-second) movie camera, was an amazing experience. The
video below shows the nanowire and the droplet, where the rows of dots indicate
positions of silicon atoms in the nanowire. Layers of silicon atoms add onto
the nanowire beneath the liquid droplet.
Watch how a silicon nanowire self-assembles under a gold-silicon droplet, and grows around a nanocrystal of nickel disilicide
Droplet size determined the nanowire’s
diameter. Temperature determined how long, and how quickly, the nanowire grew. We
also tested other important semiconductors – germanium, gallium arsenide and
gallium phosphide – and showed that their nanowires grew in a similar way. This
finding helps us to understand how to build more complicated structures. So, we
watched what happened as we switched from one gas to another, and learned the
rules that govern which semiconductors could be stacked layer by layer, within a
single nanowire.
Because each layer has different electronic behavior, these
complicated nanowires could form versatile nano-devices. And it’s possible to grow even more
complicated nanowires.
Our most recent experiments show what happens
when we add metals, such as nickel or cobalt, to the catalytic droplet. Instead
of settling out as a flat layer, the metal forms a single nanocrystal that
floats in the droplet. Eventually this nanocrystal settles out, and we can grow
the nanowire around it. This sequence of events was completely unexpected to us
– imagine if you add that extra spoonful of sugar to your coffee and then find
a perfect cube of sugar floating in the liquid! Electron microscope videos help
us understand why the atoms assemble in this way, and how to control the
resulting nanowire to optimize the shape, size and position of the embedded
nanocrystals.
Self-assembling nanowires creates new
concepts for nanomaterials, helping us build structures that cannot be
fabricated using conventional techniques. The self-assembly process is still challenging
because every nanowire comes out slightly different – perhaps imperfect for use
in a device. We are not yet precise enough to exactly replicate the process across
an entire wafer, much less build reliable nano-devices. However, each step
forward, such as the nanowires with embedded nanocrystals, suggests new opportunities
for electronics and other applications.
Self-assembly will succeed when we can
harness the spontaneous behavior of atoms. That’s an exciting prospect. We need
to change our way of thinking to avoid the need for perfection, but still
control how the self-assembled structures behave under certain conditions –
allowing us to create wholly unique devices and capabilities.
