MicroNanoSystems

Flat out for the future

One of the biggest surprises in carbon chemistry came in the late 1980s, when new carbon-only structures were isolated, firstly in football-like structures (fullerenes or “buckyballs”) and then in tubes with diameters on the nanoscale. In the past three years, relatively flat sheets of two-dimensional crystals of carbon, “graphene”, have been isolated and studied. Dr. Mike Cooke, Technology Journalist reports on some of the fascinating new properties of the new material and its prospects for true nanoscale applications.

Carbon is not the most reactive chemical in the world – drop it in water and it doesn’t explode like sodium or lithium. And yet, rather than being flashy, carbon has a subtlety that allows it to form, with other elements, a variety of molecules and polymers that could theoretically go on forever. Life is based on this infinite variety.

The physical picture of carbon is of a massive nucleus of positive charge six surrounded by six electrons of very small mass. Two of the six electrons are tightly bound and the four less tightly bound “valence” electrons can interact with the valence electrons of other atoms forming chemical bonds. In diamond, each carbon atom is bound to four other carbon atoms with equal angles between the bonds. The silicon in semiconductor wafers also have four valence electrons (with ten tightly bound) in a diamond structure. However, pure carbon is not most commonly found in the diamond form, but rather in a form called graphite that consists of layers of carbon sheets (called “graphene”) in a hexagonal “chicken wire” structure (Figure 1). Within the sheets each carbon atom has a strong bond with three others, while the bonding between the sheets is not so strong. This structure is used in the incorrectly named “lead pencil,” where the carbon sheets are drawn off the tip of the pencil to leave a visible line on paper.

The “sp2” bonding of graphite (diamond is sp3) is also responsible for carbon nanotubes, where the graphene sheets are rolled up into tubes. Fullerenes also depend on the graphite bond, but instead of forming just hexagons, adding some rings of five carbon atoms allows the sheet to curve into spherical structures.

About three years ago, scientists from the University of Manchester (UK) and the Institute for Microelectronics Technology in Chernogolovka (Russia) managed to isolate and study sheets of carbon consisting of single or a few layers of graphene [1]. The University of Manchester has since become one of the leading sites for graphene research. Graphene has been produced on silicon carbide, non-crystal substrates, in liquid suspension and as suspended membranes. After the discovery of graphene, other “two-dimensional” (2D) materials have also been found constructed of boron nitride and of a half-layer of the high temperature superconductor, bismuth strontium calcium copper oxide (BSCCO).

Since stable graphene material has only existed in researchable form for about three years, it is not surprising that potential applications are at this stage speculative. Indeed, University of Manchester-based researchers Professor Andre Geim and Dr Kostya Novoselov are rather modest when they write [2]: “There exists a popular opinion that graphene should be considered simply as unfolded carbon nanotubes and, therefore, can compete with them in the myriad of applications already suggested. Partisans of this view often claim that graphene will make nanotubes obsolete, allowing all the promised applications to reach an industrial stage because, unlike nanotubes, graphene can (probably) be produced in large quantities with fully reproducible properties. This view is both unfair and inaccurate.” They add:
“Depending on the particular problem in hand, graphene’s prospects can be sometimes superior, sometimes inferior, and most often completely different from those of carbon nanotubes or, for the sake of argument, of graphite.”

Production Processes
The most common “production” technique is micromechanical cleavage of graphite. This was the method used to first isolate graphene [1]. This method can produce graphene crystallites up to 100µm in diameter. Essentially one variant is just a more sophisticated process of drawing with a pencil and looking among the traces for graphene pieces. Another way is to repeatedly use adhesive tape on the graphite. Having got the flakes the researcher (or assistant) has to look among the pieces for the 1–10 layer crystallites. “The problem is that graphene crystallites left on a substrate are extremely rare and hidden in a ‘haystack’ of thousands of thick (graphite) flakes,” write Geim and Novoselov. The few micron-sized crystallites are spread over square centimetres (a 10µm square is a millionth of a square centimetre). Low throughput and the lack of signatures preclude the use of high-resolution scanning probe or electron microscopy.

Fortunately, the graphene flakes can be seen with an optical microscope if placed on top of a 300nm layer of silicon dioxide on silicon. Interestingly, the thickness of the SiO2 is quite critical – the graphene disappears on a 315nm oxide layer, only a 5% change of thickness. Once the graphene flakes have been found, their thickness can be determined using Raman spectroscopy to probe the vibration spectrum of the flake.

Such painstaking techniques are for the determined scientist who expects a big payoff in terms of finding new physics. Industrialists will want repeatable, mass production. In the 1990s, small islands (a couple of nanometres across) single and few atoms thick graphite layers were grown on metal substrates (platinum and titanium carbide) using chemical vapour deposition. However, the main alternative to the “exfoliation” of the previous paragraph, using techniques dating to the 1970s, is thermal decomposition of hexagonal (6H) silicon carbide crystals. However, mass production of controllable graphene layers still seems some way off.

Electronics
The paper [1] that launched much of the recent work found that the graphene films showed metallic behaviour. Although single-layer graphene is metallic (actually it has a zero energy gap, so is, rather, “semimetallic”, also two-layer graphene is well approximated with a zero-gap; beyond two layers the bands overlap, producing ever more metallic behaviour), an electric field can produce semiconducting properties. Confinement effects such as quantum wells and superlattices can also affect the electronic properties in a semiconducting direction.

Two-dimensional behaviour only persists up to about ten layers after which bulk, 3D, graphite properties dominate. The graphene electron and hole concentrations are commonly of the order 1013/cm2. Room temperature mobility (measuring the ease of moving the charged carriers with an electric field) was found to be of the order of 10,000cm2/Vs, a value that has increased with time to more than 30,000cm2/Vs [3]. The corresponding value for silicon is 1500cm2/Vs for electrons and 500cm2/Vs for holes. Since graphene’s mobility is thought to be limited by impurity scattering, further improvements can be expected as material quality increases, perhaps even reaching 100,000cm2/Vs.

Ballistic transport, where the carriers don’t scatter for some distance, has been seen up to 0.3µm, at least an order of magnitude better than silicon. The achievement of a terahertz frequency ballistic transistor would be “an important milestone for graphene-based electronics” according to Geim and Novoselov. A further important factor is that graphene’s mobility does not seem to be much affected by carrier concentration (the other main factor of conductivity), unlike in semiconductors where the mobility drops sharply at ultrahigh doping (reducing mean free paths).

Another aspect that excites physicists is that electron motions in graphene are better described using the relativistic Dirac quantum theory (albeit with an effective “speed of light” of ~106m/s rather than the true value for vacuum of 3x108m/s) as opposed to the traditional nonrelativistic Schrodinger equation that is almost universally adequate for the rest of condensed matter physics (at least as a first approximation).

Transistors
The first graphene-based transistor was at the same time as the material’s discovery [1], and other groups have since reproduced the result (e.g. Georgia Institute of Technology growing the graphene on silicon carbide substrates, [4]). But these graphene transistors were very ‘leaky’. If such high leakage rates continued, this would limit possible applications and rule out use in computer chips and other electronic circuits needing a high density of transistors.

Another approach to electronics is to use graphene as a conducting sheet on which components – nanoribbon conducting channels and interconnects, quantum dots (Figure 2), semi-transparent conduction barriers, etc. – are constructed. The conductance can be controlled by gates (back or side electrodes, also constructed from graphene). Coulomb blockade effects, showing the effects of single charge carriers, have been demonstrated in a relatively large 0.25µm diameter quantum dot (QD) at low temperature. A 10nm-scale device operating at room temperature shows complete pinch-off of current (Figure 3). One transistor structure built in Manchester was only one atom thick and less than 50 atoms wide.

A slightly puzzling aspect to graphene is that stable 2D crystal lattices are theoretically “impossible”. Such big names (in condensed matter physics, anyway) as Landau and Peierls gave calculations more than 70 years ago showing that a 2D crystal lattice structure would melt at any finite temperature. Experiments showing that below a certain thickness films on a lattice matched substrate decompose or segregate into islands would appear to confirm the theory.

So what about objects like graphene? Until recently, it could be argued that the graphene produced was part of some other object – embedded in a bulk material or supported by a substrate. However, scientists from Manchester, and the Max Planck Institute for Solid State Research (Stuttgart, Germany) and the University of Nijmegen (The Netherlands), have produced graphene sheets freely suspended on a microfabricated scaffold (Figure 4) [5]. Transmission electron microscopy reveal that the sheets are not strictly flat – the surface normal varies by several degrees and the out-of-plane deformation reaches 1nm. It is believed that this deformation in the third dimension allows “twodimensional” sheets such as graphene to be stable against the pure 2D theory.

Graphene researchers believe that roomtemperature performance is possible with graphene-based single electron transistors (SETs), unlike those constructed from other materials (caused in part by poor material stability at room temperature). Research has shown that graphene remains highly stable and conductive even when it is cut into strips only a few nanometres wide. All other known materials – including silicon – oxidise, decompose and become unstable at sizes ten times larger. The poor stability of these materials has been seen as a fundamental barrier to electronic device scaling into the true nanometre scale (< 10nm) – and this could halt development of microelectronics beyond 2020.

“Our approach is not scalable but it is good enough for research and proof-of-concept devices,” admits Geim and he does not expect that graphene-based circuits will come of age before about 2025. Until then, silicon technology should remain dominant. But he believes graphene is probably the only viable approach after the silicon era comes to an end: “This material combines many enticing features from other technologies that have been considered as alternatives to the silicon-based technology. Graphene combines most exciting features from carbon-nanotube, singleelectron and molecular electronics, all in one.”

Ribbon Cutting on a New Era
For room temperature electronic applications, 10nm wide graphene ribbons provide the needed confinement for engineering semiconducting properties. “We have made ribbons only a few nanometres wide and cannot rule out the possibility of confining graphene even further – down to maybe a single ring of carbon atoms,” says Geim. Dr Leonid Ponomarenko, another of the leading researchers involved in the work at Manchester, adds: “At the present time no technology can cut individual elements with nanometre precision. We have to rely on chance by narrowing our ribbons to a few nanometres in width. Some of them were too wide and did not work properly whereas others were over-cut and broken.”

Repeatable ribbons with crystallographic edges (zig-zag or “armchair”, see Figure 5) would need precise etching. Rough edged ribbons suffer significant conductance variations and additional scattering, leading to higher resistance, reduced speed and energy losses. Fortunately, anisotropic etching, depending on the different chemical reactivity of zig-zag and armchair edges, seems possible.

Nanomechanics
These researchers see nearer-term prospects for graphene coming from outside of electronics. One of the first expected areas is composite materials, particularly for conducting plastics. Mechanical strength is a problem here, better met by the properties of nanotubes (depending on volume production at low cost).

Graphene powder could also be useful in creating better batteries (where graphite and, for increased surface area, carbon nanofibres are already used). Some early plasma display systems used graphite flakes as emitters, and graphene could resurrect this structure.

One application giving graphene a definite edge over nanotubes is in gas sensors. Magnetic properties also put graphene in the frame for quantum computing efforts, along with spin valve and superconducting transistor research work.

Hydrogen storage, where nanotubes have been looked at as possible media, may also offer opportunities for graphene. Another direction could be to use the hexagonal carbon rings in the “chicken-wire” as a sieve for atoms or small molecules. If the production technology really advanced, one could foresee graphene’s use as an ultrathin, transparent robust substrate for a whole range of technologies such as nanomechanical devices. A first substrate application could be for high resolution electron microscopy, maybe allowing imaging of individual molecules with unprecedented accuracy. July 2007 www.micronanosystems.info

References
[1] Novoselov et al, Science, vol.306, p.666, 22 October 2004.
[2] Geim and Novoselov, Nature Materials, p.183, March 2007.
[3] Berger et al, physica status solidi (a), vol.204, p.1746, 2007.
[4] Berger et al, J. Phys. Chem. B, vol.108, p.19912, 2004.
[5] Meyer et al, Nature, vol.446, p.60, 1March 2007.