"The response bandwidth in conventional photodetectors made from silicon, germanium or compound semiconductors is limited to photons with an energy above the electronic bandgap of these materials," explains team member Xuetao Gan. "Graphene has no such cut-off energy and can, in principle, detect photons with any wavelength. For instance, a photon with a wavelength of two microns produces a measurable photocurrent in our device but the same photon would fly straight though a germanium photodetector without being absorbed."

Graphene is a sheet of carbon atoms arranged in a honeycomb-like lattice just one atom thick. The material could find use in a number of technological applications and even replace silicon as the electronic industry's material of choice in the future thanks to its unique properties, such as extremely high electrical conductivity.

Graphene also shows great promise for photonics applications because it has an ideal "internal quantum efficiency" – almost every photon absorbed by the material generates an electron-hole pair that could, in principle, be converted into electric current. Thanks to its "Dirac" electrons, which whizz through graphene at extremely high speeds, it can also absorb light of any colour.

However, there is a snag in that graphene's "external quantum efficiency" is low – it absorbs less than 3% of the light falling on it. And to add to the problem, useful electrical current can only be extracted from graphene-based devices that have electrical contacts with an optimized "asymmetry".

100% of photons are absorbed

The MIT-IBM-Columbia team, led by Dirk Englund, has now overcome these problems by evanescently coupling optical waveguides in silicon membranes to bilayer graphene sheets. In this set up, photons interact with graphene over distances as long as tens of microns, which means that nearly 100% of these photons are absorbed. "The waveguide-integrated configuration thus allows the graphene detector to harvest photons and effectively convert them into electrical signals," Gan told nanotechweb.org. "The detector also has a photoresponsivity of 100 mAW–1, a value that approaches that of a germanium photodetector integrated in a silicon photonics circuit."

Graphene photodetectors were first made in 2009 using graphene field-effect transistors in a so-called normal-incidence configuration. However the maximum photoresponsitivities of these devices (of 6.1 mA/W) were limited by the atomically thin graphene layer that absorbed only a small fraction of incident light. Researchers increased the amount of light absorbed by integrating the graphene FETs into micro- and nanocavities and plasmon resonators, but such resonant devices only detected light in a narrow band of the electromagnetic spectrum. Hybrid graphene-quantum dot systems fared a little better in that they showed much-improved responsitivity, but, unfortunately, this came at the cost of device speed.

"Our device, on the other hand, shows high responsivity, high speed (exceeding 20 GHz), and broad spectral bandwidth," said Gan.

Another good property of the new photodetector is that it works without an external energy source. And, graphene may soon become cheaper than germanium to produce in mass quantities and is easier to incorporate into a silicon chip. Indeed, the MIT team has managed to fabricate its waveguide-integrated photodetector using only two lithography steps, which is far fewer than required for making chip-integrated germanium detectors.

High internal quantum efficiency for collecting photocarriers

The researchers made their photodetector by first transferring graphene layers onto a planar silicon waveguide. Next, they deposited two metal electrodes at the opposite sides of the waveguide. The two electrodes were placed parallel to the waveguide at distances of 100 nm and 3.5 µm.

The potential difference of the graphene near the electrode closer to the waveguide allows photocarriers generated around the waveguide to be separated efficiently. The fact that the other electrode is placed asymmetrically with respect to the first one ensures that carriers are accelerated in one direction only and can so be collected more easily. "Our optimized asymmetric metal electrode design therefore provides a high internal quantum efficiency for collecting photocarriers," explains Gan.

"Our graphene photodetector is integrated on a silicon chip and the processes to fabricate it are compatible with complementary-metal-oxide-silicon (CMOS) fabrication techniques. The device might therefore be integrated into computer chips or mobile devices to make optical interconnects. Integrating graphene into devices this way would increase their signal processing speed, lower the amount of power they consume and make them cheaper in the long term."

The researchers say that they now plan to improve the design of their device and increase its photoresponse further so that it can realistically compete with state-of-the-art germanium detectors.

The present work is detailed in Nature Photonics doi:10.1038/nphoton.2013.253.

Further reading

Quantum dot blend gives wide-bandwidth FET-based photodetector (Jul 2012)
Ultrabroadband photodetector made using graphene ink (Nov 2010)
Graphene photodetector is a first (Apr 2010)
Graphene-QD photodetector breaks new record (May 2012)
QD photodetector speeds up (Nov 2012)