An early expectation
When graphene was isolated in 2004, electronics was one of the first uses people imagined. Electrons in good graphene move quickly with little scattering, the sheet is only one atom thick, and carbon is plentiful. Silicon transistors were shrinking toward limits set by physics and manufacturing, so a thinner, faster material looked attractive.
The very first devices from the Manchester group were already field-effect devices, meaning that an electric field applied through a nearby gate changed how much current the graphene carried. That is the basic principle of a transistor. So in a loose sense, the earliest graphene experiments were also the earliest graphene electronics. The harder question was whether a useful circuit could be built.
The central obstacle: no off switch
A transistor works as a switch. In digital logic, it must be able to turn almost completely off, so that ones and zeros are clearly different. Silicon manages this because it is a semiconductor with an energy band gap, a range of energies that electrons cannot occupy.
Ordinary graphene has no band gap. Applying a voltage changes the number of charge carriers, but current never fully shuts off. A graphene transistor can therefore amplify signals well but is a poor choice for digital logic, which is what a computer processor mostly does.
Researchers tried several ways around this, and each has trade-offs.
- Narrow ribbons. Cutting graphene into very thin strips can open a small gap. The edges must be extremely clean and precise, though, and rough edges spoil the effect.
- Two layers with an electric field. In bilayer graphene, an applied field can open a gap. The gaps reported have generally been small.
- Chemical modification. Adding other atoms such as hydrogen can change graphene’s behavior, but it also damages the very conduction that made it attractive.
None of these has yet produced a clear replacement for silicon logic.
High-frequency devices
Analog radio-frequency electronics do not need a perfect off state, so they became the natural early target. In 2010, researchers at IBM reported graphene transistors operating at a cutoff frequency of around 100 gigahertz, grown on a silicon carbide surface. Later reports from several groups pushed numbers higher in laboratory devices.
These are real results, and they showed graphene could handle fast signals. A cutoff frequency is not the same as practical performance, however. Gain, noise and consistency across thousands of devices all matter in a working circuit. Graphene transistors generally have not shown a clear advantage over established semiconductors in that wider comparison.
Making the material
For electronics, a flake from sticky tape is too small and too unpredictable. Several production routes were developed.
- Growth on silicon carbide. Heating silicon carbide drives off silicon and leaves carbon in a graphene-like layer. Walter de Heer and colleagues at Georgia Tech pursued this in the early 2000s. It yields wafer-sized material, but the substrate is expensive.
- Chemical vapor deposition on metal. Around 2009, groups including one led by Rodney Ruoff reported growing large-area graphene on copper foil from a carbon-containing gas. The film can then be moved to another surface. This remains the most common route to large sheets, although transferring and keeping the film clean introduces wrinkles, tears and residue.
- Liquid exfoliation and graphene oxide. Graphite can be broken up in liquid, or oxidized and then reduced, to make flakes and inks. These are cheap and scalable, but the result is far less perfect than a clean single sheet and is not suited to high-speed transistors.
In 2010 a team in South Korea and Japan reported a roll-to-roll process producing a 30-inch graphene film used in a prototype touch panel. This pointed toward transparent conductors as a more realistic first market than processors.
Where graphene is actually being applied
Because the logic-chip route proved difficult, attention shifted to uses that play to graphene’s strengths and tolerate its weaknesses.
- Transparent, flexible conductors. A thin sheet can carry current while letting most light pass. The challenge is competing on cost and resistance with established materials such as indium tin oxide.
- Sensors. Because every atom of graphene sits on the surface, small changes in its surroundings can alter its conductivity. That makes it a candidate for chemical and biological sensors, including ones built on biochips.
- Photodetectors and modulators. Graphene absorbs light across a wide range of wavelengths. Prototypes have been built for optical communication and fast light detection, relevant to optical fibre networks.
- Printed and composite electronics. Graphene inks are being tried for antennas, conductive coatings and similar jobs where precision is less critical.
Some of these have reached commercial products in small niches. None has produced the broad “graphene revolution” that early press coverage suggested.
Why silicon is still king
Silicon has more than half a century of manufacturing know-how behind it. Entire factories, design tools and supply chains are built around silicon chips. A new material has to be better by a wide margin, in a form that can be made reliably in huge quantities at low cost, to justify changing all of that. Graphene has strengths, but it has not cleared that bar.
Research continues in directions such as twisted layers, graphene combined with other thin crystals, and hybrid devices where graphene does a single job next to conventional silicon. The history so far is a good example of the gap between a spectacular laboratory material and a product: the physics can be remarkable, and getting it onto a factory floor still takes years of unglamorous work. Claims of near-term replacement of silicon should be treated with caution, and the underlying story of the material itself is told in the account of how graphene was first isolated.