& cplSiteName &

Opel: Cooking With GaAs

Light Reading
News Analysis
Light Reading
11/27/2002

Startup Opel Inc. is going planar -- with gallium arsenide.

Electronics have benefitted greatly from the "planar" process, which puts circuits down on a flat surface -- that's the concept that spawned the semiconductor industry. Optics are going planar as well, with several companies working on planar lasers -- Opel among them.

Some startups are looking at hybrid integration, building separate chips for the electronics and optical sides, then mounting them onto a single device. But the more glamorous challenge lies in monolithic integration, trying to craft both electronics and photonics from a single material, so that the entire device can be built at once (see Photonic Integrated Circuits). Opel's approach to monolithic integration of planar circuits is somewhat unusual. It plans on using gallium arsenide (GaAs) for both halves of the component, as opposed to the more popular combination of other materials such as indium phosphide (InP) or silicon germanium (SiGe).

Lasers and their associated electronics tend to be made of different materials, particularly when designed for high speeds. On the electronics side, it's tough to resist the low cost and low power of complementary metal-oxide semiconductor (CMOS) devices. But the optics, particularly at high speeds, need to be made of more exotic materials like InP or SiGe.

The Opel project is the brainchild of Geoff Taylor, whose research in GaAs spans 15 years, first with Bell Labs and, for the past eight years, with the University of Connecticut.

Taylor's innovations lie deep in the chemistry of semiconductor manufacturing. He had to focus on a process called epitaxy, where molecule-thin layers of chemicals are laid precisely onto a wafer, eventually building the structures that form transistors, mirrors, and other components. The problem is that the epitaxy for GaAs lasers is different from that of GaAs electronics, and it wasn't until a few years ago that Taylor perfected an epitaxy suitable for both sides.

"It's a matter of materials, physics, doping dosages, and other very alchemic arts," says Daniel Upp, Opel's chief executive and a founder of TranSwitch Corp. (Nasdaq: TXCC). Upp admits he's not up on the specifics of the epitaxy; he was brought in as the business guy, to help turn Taylor's research into useable products.

The resulting manufacturing process is called Planar Optoelectric Technology (POET) and will be the basis for integrated devices that could be ready in two years. For now, Opel's founders are happy just to have completed their first test devices.

"This has been largely just paperwork until the last couple of months," Upp says. The initial devices are quite basic, just a transistor and a laser. "But it does show the integration of the two in a monolithic fashion."

There's more to Opel than just manufacturing ideas, however. On the optical side, Taylor's research has concentrated on thyristors, which can be used to form lasers or photodetectors (though they do sound more like late-Jurassic carnivores). And the electronics can be made configurable to apply to the transmit or receive path. Taylor also made strides on the electronics side, creating a GaAs transistor "that allows us to build very high-speed logic, just like CMOS," Upp says.

But Upp is careful not to boast that Opel's technology can replace CMOS. Wise move, because GaAs has a checkered history. Years ago, it was considered the material electronics would have to use when CMOS ran out of speed. But CMOS's life span keeps getting extended through techniques such as fancy signal processing; CMOS chips have proven capable of handling 10-Gbit/s signals; and few doubt its ability to eventually hit 40 Gbit/s.

That's made GaAs the continual prince-in-waiting. The material has found uses in wireless and radio-frequency (RF) applications, but in wire-line applications it's seen as a place holder for the products that are one step ahead of CMOS's current ability.

"Historically, CMOS has decimated GaAs on the digital side. For guys like Vitesse Semiconductor Corp., their GaAs business just imploded," says Fred Zieber, analyst with Pathfinder Research.

Upp concedes that for some very high-speed applications, hybrid integration will be preferable to Opel's approach because of the need for CMOS on the electronics side.

Still, Upp sees advantages to Opel's approach, because of the precision assembly required in the hybrid approach. "Our technology has inherently a lower assembly cost, and the die size will be smaller in most applications," he says. "It's not that you want to slap down a lot of gallium arsenide for the speed. You can get some systems benefits from moderate amounts of [integrated] electronics."

For example, monolithic integration offers the promise of putting all the components of a 10-Gbit/s Ethernet transceiver into a single chip. Moreover, the GaAs logic circuits could handle the encoding and decoding requirements of the standard XAUI interface, obviating the need for a separate XAUI chip and making line-card design one step simpler.

A more innovative example might lie in parallel optics, where the thyristor's ability to transmit or receive could come into play. Upp envisions a chain of Opel devices that can swap from transmit to receive mode. "You can decide on the fly whether you want to use one as a laser and another as a photoreceiver. Now you can manufacture one component, and it can be all send, all receive, or a mix."

And like most startups, Opel has its eye on exotic, futuristic applications. Upp says his lab has produced analog-to-digital converters running in the "several GHz range," creating a freakishly fast spin on a very common type of analog-electronics device. Word Upp!

Most of Opel's funding has come from its founders, which include Upp, Taylor, COO Lee Pierhal, and Rohinton Dehmubed, vice president of optical devices. Taylor dug up extra money in the form of small-business government grants. With a staff of only 12 and operations still based at U. Conn, the startup is looking for outside funding to help commercialize what it's got.

— Craig Matsumoto, Senior Editor, Light Reading
www.lightreading.com

Movers and shakers from more than 100 companies will be speaking at LightSpeed Europe. Check it out at http://www.lightspeedeurope.com .



(2)  | 
Comment  | 
Print  | 
Related Stories
Newest First  |  Oldest First  |  Threaded View        ADD A COMMENT
beachboy
beachboy
12/4/2012 | 9:16:19 PM
re: Opel: Cooking With GaAs
Monolithic may not always produce the best solution. Look to the mature electronics industry where not all volume applications are totally integrated. It still makes sense to keep some functionality separate.
Look to low volume applications where very few functions are monolithically integrated.
In semiconductors, individual functional yields are well into the 90s when they are combined in one chip.
In optics if you combine a 50% yielding function with a 50% and a 40% and another 50% etc you soon end up with a monolithic yield approaching 0%, piles of scrap and massive capacity to produce the scrap.
next-gen-wisdom
next-gen-wisdom
12/4/2012 | 9:16:15 PM
re: Opel: Cooking With GaAs
yeah right.
Featured Video
Upcoming Live Events
December 3-5, 2019, Vienna, Austria
December 3, 2019, New York, New York
March 16-18, 2020, Embassy Suites, Denver, Colorado
May 18-20, 2020, Irving Convention Center, Dallas, TX
All Upcoming Live Events