26 Sep 2003
Tami Freeman takes a look at how long-wavelength VCSELs are being used to enhance transceivers.
From FibreSystems Europe September 2003
With the optical comms market flatlining through 2003, now looks like a good time to go shopping for next-generation photonic technologies at bargain prices. At least that’s the view being taken by Optical Communication Products (OCP), a US optical subsystems maker which has been busy buying up intellectual property and transforming it into commercial products.
OCP designs and manufactures a range of active modules – high-speed transponders, transceivers and discrete transmitters and receivers – with applications in dense wavelength-division multiplexing, Synchronous Digital Hierarchy/Synchronous Optical Network, Gigabit Ethernet and Fibre Channel systems. One of the firm’s original investors is Furukawa Electric of Japan, which currently holds a 60% stake in the group.
Founded in the early 1990s, the company raised $120 m (7106 m) in an IPO in 2000, and has since used the cash to expand into Europe and to invest in acquisitions. “We have acquired companies that have the technologies we need for our products, rather than buying businesses that already have revenues and success,” said David Jenkins, managing director for OCP (Europe). “When the market gets back on an even keel you’ll start to see the investments we have made really drive us forward. [In contrast] our competitors are cutting back on their R&D and we think that’s quite short-sighted.”
A prime example of this strategy is the firm’s purchase of US start-up Cielo Communications last October. Cielo developed long-wavelength vertical-cavity surface-emitting lasers (VCSELs) that emit at around 1300 nm. Since the acquisition, researchers at OCP’s European R&D centre have been working with the Cielo group in Colorado to integrate the VCSEL technology into a range of commercial transceiver modules. The products are in the final stages of development and are slated for initial release later this year.
“We see a number of applications for the VCSEL-based modules,” said Jenkins. “They will be used for STM-1, STM-4, Gigabit Ethernet and STM-16 singlemode applications. And because of the very low power requirements there are some interesting uses within fibre-to-the-home architectures.” Another opportunity is the construction of transceiver arrays with multiple transmitters and receivers – an area where low power consumption is particularly crucial.
VCSELs are attractive because they combine the surface emission and low production costs of LEDs with the speed and power of a laser. The secret to their success is a miniature, highly efficient vertical cavity that sandwiches an active light-generating area between two highly reflective Bragg mirrors made from stacked layers of epitaxially grown semiconductor.
“People have been trying to make long-wavelength VCSELs work for a long time,” said Jenkins. “The problem was that they didn’t work well at high temperatures and were difficult to fabricate.” He explained that the wafer-bonding technique used to make early generations of long-wavelength VCSELs was unreliable and did not always produce a perfectly matched interface between the different material layers.
The OCP design uses an active region based on indium gallium arsenide nitride (InGaAsN) quantum wells. Incorporating nitrogen shifts the material’s bandgap to a longer wavelength to produce emission around 1300 nm. The fabrication issues are resolved by using monolithic growth, in which the entire VCSEL structure is grown in one step on a GaAs substrate in a molecular beam epitaxy machine.
OCP’s VCSELs offer high-performance operation at temperatures of up to 90 °C with a peak output power (at room temperature) of 1.3 mW. The spectrum is similar to that of a distributed-feedback (DFB) laser, with a narrow linewidth and effectively singlemode operation. The lasers’ drive current is just a few milliamps and is hardly affected by temperature, unlike the case for DFB and Fabry–Perot lasers. (see “Compare and contrast” table)
VCSELs are also generally easier and cheaper to fabricate than DFB and Fabry–Perot lasers. As the lasers are top emitters, manufacturers can carry out on-wafer testing before the wafers are diced into individual chips. This means that it is possible to check key parameters (such as threshold current and output power) of tens of thousands of devices before packaging, offering the potential for low-cost, high-volume production.
Going forward, Jenkins is keen to pursue additional investment opportunities. “Our plan is to buy important technologies and invest in them for a year to 18 months to get them to commercial products in the market-place. Our financial situation – with $140 m in the bank and being profitable for the last 10 years – positions us to do this.”
Earlier this year, for example, OCP purchased US-based VCSEL manufacturer Gore Photonics. Gore specialized in 850 nm VCSELs designed for use in applications such as backplane interconnect equipment, 300 m ribbon-cable connections and cross-connect switches. OCP is already sampling VCSEL arrays based on this product.
Tami Freeman is technology editor of FibreSystems Europe magazine.