Can Translume Direct-Write Success?
The company hasn't yet said what products it intends to manufacture, but it's likely that these will include passive components like splitters, attenuators, and arrayed waveguide gratings (AWGs). In that respect it's following in the footsteps of other integrated optics vendors like Bookham Technology PLC (Nasdaq: BKHM; London: BHM), Kymata Ltd., Lightwave Microsystems Corp., and NTT Electronics Corp. (NEL) (to name but a few). Translume was founded just a few weeks ago, so it's very early days. So far it has a bunch of ideas, four employees, and the support of Ardesta LLC, an investor cum incubator. The technology it's planning to use was developed at Clark MXR Inc. and the University of Michigan. Here's how direct-writing works. If an ultra-fast, high-power pulse from a laser is tightly focused below the surface of a glass block or wafer, then the glass melts. Each pulse melts a sphere of about 10 microns in diameter inside the glass.
The laser beam is scanned over the surface of the glass, tracing out each component, one at a time. It leaves a channel of molten material behind, which cools to form a waveguide.
The glass at the edge of the channel cools first to form low-density glass with a low refractive index. This process increases the pressure on the molten glass in the center of the channel, which then cools to form high-density, high-refractive-index glass.
Phillipe Bado, Translume's chief science officer, claims that laser-writing of optical components has three big advantages. One, it's cheap: "It's a simple technology that doesn't require a large clean-room production facility, so the cost of production should be significantly lower."
Two, the yield is high. "We're shooting for yields above 90 percent," he says. Although most component manufacturers don't like to admit it, yields in the low tens of percent are common for devices like AWGs, he adds. A high yield helps cut costs, too.
The third advantage is unique. The laser-writing process creates waveguides that have a circular cross-section rather than the usual rectangular one. As a result, it's possible to couple light from the waveguide into an optical fiber more easily.
Critics of the process say that direct-write methods, which "draw" components one at a time, are slow compared to parallel techniques, like photolithography. But Bado says this is a fallacy. "We believe we can get throughputs as high as anyone can get using parallel fabrication methods." That's mainly because optical circuits typically measure millimeters or even centimeters across, so not many of them fit on a wafer. "In optics, the advantages of writing in parallel are simply not as great as they are in microelectronics," Bado contends.
— Pauline Rigby, Senior Editor, Light Reading, http://www.lightreading.com