Hybrid source promises plastic-fibre applications

17 Jun 2002

A high-brightness light-emitting diode that has a structure resembling a vertical-cavity surface-emitting laser is promising to enable plastic-optical-fibre communication. Nadya Anscombe finds that European companies are leading the world in manufacturing these devices.

From Opto & Laser Europe April 2001

If you were looking for a high-brightness semiconductor light-source you would normally only have two choices: a light-emitting diode or a laser. Now there is a device available whose structure lies somewhere between a light-emitting diode and a vertical-cavity surface-emitting laser.

A resonant-cavity light-emitting diode (RC-LED) possesses a structure that is more efficient than a conventional LED because it has, as its name suggests, a resonant cavity. The device produces directional emission, but it is not a laser.

Currently there are no companies producing high volumes of RC-LEDs. However, this is set to change in the next few months. Three European firms - IQE of the UK, Osram Opto Semiconductor of Germany and Mitel of Sweden - are ramping up their production of RC-LEDs this year, and FireComms - a new Irish start-up - makes its market debut this month. The main driving force for this sudden enthusiasm for RC-LEDs is their potential use in plastic-optical-fibre communications. However, Osram believes that there are many other applications that could exploit the advantages of RC-LEDs.

Osram's project manager, Ralph Wirth, says that the firm is planning to launch two devices this year. He said: "The most promising application of RC-LEDs is in data communication through plastic optical fibre. We believe that RC-LEDs could also be used in printing, optical sensing and as indicators."

Osram started working on RC-LEDs in an attempt to improve the brightness of its conventional LEDs. "We were developing two techniques, one that uses a thick window and the other a transparent substrate," said Wirth. "However, we couldn't pursue these ideas further owing to patents held by Agilent and LumiLEDs. This forced us to think around the problem, so we started to develop the RC-LED."

This work not only gave Osram a high-brightness LED, but also resulted in a shorter LED-structure growth time and a more cost-efficient manufacturing process.

Wirth and his colleagues have had considerable success in developing red and orange RC-LEDs. Using red RC-LEDs with epoxy encapsulation, wall-plug efficiencies of more than 10% were achieved.

Wirth's group has also demonstrated 500 Mbit/s transmission over 50 m of plastic optical fibre. He said: "To get a faster transmission rate a smaller device and a graded-index fibre would be necessary."

Markus Pessa, director of the Optoelectronics Research Centre at the University of Tampere, Finland, agrees: "We have shown an error-free 622 Mbit/s data-transmission rate over a 1 m, 0.5 numerical aperture step-index plastic optical fibre. It is the fibre that sets the limit of the system bandwidth at distances of more than a few metres. Achieving a high transmission rate over long distances requires a graded-index small-numerical-aperture fibre. To my knowledge this is not yet available."

Pessa is currently working with several industrial partners - including fellow Finnish company Nordic Epitaxy. He hopes to develop a diffractive-optics technology that would allow lensless coupling of RC-LEDs to a plastic fibre with a coupling efficiency that is comparable to that attainable by lensed coupling. Pessa expects the first results to be available in the next few months. Plastic optical fibre is not the only application for RC-LEDs, says Pessa. Other potential uses are miniature monochromatic displays; low-cost light sources for imaging technologies and measurement systems; and optical printing.

However, Mitel Semiconductor in Sweden is concentrating solely on the lucrative plastic-optical-fibre applications. Bengt Ehrensparre, product line manager for optical communications told OLE:"Our RC-LED is the only LED that we have in the visible range and we have worked hard to optimize it for plastic-optical-fibre applications.

"The market for RC-LEDs in this area is growing rapidly because there are currently no VCSELs in the required wavelength range with the desired performance characteristics and temperature sensitivity."

One of Europe's main suppliers of epitaxy wafers for RC-LEDs is IQE in the UK. Geoff Duggan, product group manager, told OLE:" We cannot meet demand at the moment, but we plan to expand our capacity."

One European group that has anticipated this demand is at the National Materials Research Centre (NMRC) in Cork, Ireland. While most companies are still developing red RC-LEDs, the NMRC is in a pan-European project to develop next-generation technology for green RC-LEDs.

John Lambkin, coordinator of the AGETHA (Amber Green Emitters Targeting High-Temperature Applications) project told OLE:"Plastic optical fibre has a low attenuation at 510 nm, which means that the transmission length can be doubled. The RC-LEDs that we are developing for the green region also perform better at higher temperatures than red RC-LEDs."

The pan-European project, which has just finished its first year, is looking at developing GaN-based RC-LEDs, using quantum dots rather than quantum wells. It builds on the NMRC's expertise in developing red RC-LEDs, which will be commercialized by new start-up FireComms.

There is scepticism about how long RC-LEDs will keep their hold on the market when VCSELs, with their better performance characteristics, are available. Surely if an end-user had a choice between a VCSEL and an RC-LED, they would pick the VCSEL?

Not so, says Pessa: "If a VCSEL was available at 650 nm, no-one would use it for plastic-optical-fibre links because it would be sensitive to temperature variations.

"Also, there is no good reason to use a faster light source than the RC-LED, because the plastic optical fibre sets a limit to the system's bandwidth. The broad bandwidth provided by the VCSEL could not be fully exploited. For these two reasons the RC-LED for plastic-optical-fibre-based systems would not be a short-term solution. I regard it as an optimal solution." Pessa believes that Europe is one step ahead of Asia and the US when it comes to developing RC-LEDs. "Europe has many firms and research groups in this field, but there is nothing comparable in Japan or the US."

However, he warns: "Europe must work hard to keep its lead, or countries such as Taiwan, which has expertise in LED production, will take this lead away from us."

The structure of a typical RC-LED is similar to that of a red VCSEL except that there are fewer periods in the top Bragg mirror of an RC-LED. Typically, this contains between four and eight periods. In such a microcavity, spontaneous emission is no longer isotropic. Owing to Fabry-Perot resonances, certain wavelengths are only allowed in specific directions.

The thickness of an RC-LED can be significantly less than a VCSEL. Its total incorporated strain is low - because no lattice mismatched layers are involved - which simplifies device processing.

While the internal efficiency of a conventional LED can be high (in the red wavelength regime values close to 100% can be reached) its extraction efficiency is poor. This is due to the high index-of-refraction contrast between the semiconductor and the surrounding epoxy.

RC-LEDs are simple, cost-effective to make and have reasonably high efficiencies. Using an optical resonator does not circumvent the problem of total internal reflection, but it does help to direct the spontaneous emission of the quantum wells into the escape cone. An RC-LED that is optimized for power extraction exhibits the different wavelengths of the quantum-well-emission spectrum at different angles in the emission pattern. Long wavelengths are found in the forward direction and shorter wavelengths are emitted at oblique angles. RC-LEDs can therefore be tuned to be less temperature sensitive at specific wavelengths.

…over conventional LEDs: more efficient, brighter, greater spectral purity, better beam directionality and higher speed of modulation.

...over VCSELs: no threshold, simpler structure, more cost-effective manufacturing, better temperature sensitivity and available in the visible.