Silicon lasers are coming, and there will be many exciting applications beyond what is now being pursued, said Bahram Jalali, a University of California-Los Angeles electrical engineering professor, at an optoelectronics plenary session at Photonics West 2006.
Jalali told a standing-room-only crowd that silicon is an attractive prospect for lasers, and might even be the best material from which to make lasers, because of its high thermal conductivity and high optical damage threshold. Its major drawback: Silicon can't amplify light. Adding erbium could help it overcome that drawback, but silicon isn't a good host for the element, he said.
Jalali discussed some recent studies in which light was amplified using silicon, including work at Brown University a few months ago to fabricate devices resembling photonic crystals. Lasing was observed, with a total gain of 15 dB. Another way to amplify light with silicon is to use a Raman-based approach, Jalali said. Advantages of that method is that no unusual doping or cooling is required, and it is tunable through the pump laser. The limitation is that it requires optical pumping and can't compete with III-V diode lasers. Raman silicon lasers can also lase at two different wavelengths, something not observed in other lasing techniques, Jalali said.
By using silicon as a Raman medium, researchers have found devices to have large Raman gains, high damage thresholds and good thermal conductivity. Problems with two-photon absorption also vanish above the 2.25 um wavelength, Jalali said. For these reasons, silicon is an ideal Raman medium at the mid-wave infrared for applications in fiber Raman lasers that are being developed to replace Erbium:YAG lasers, in lasers used for skin resurfacing, and in dentistry, to whiten teeth and to eliminate the need for mechanical tooth drilling. The lasers also have potential applications in biochemical sensing in the defense industry, he said.
Looking further into the future, Jalali said, the technology could potentially be used to make photonic analog-to-digital converters, incorporating high-speed electrical signals into viable applications by digitizing and analyzing analog signals. Another goal for the future: terasample-per-second digital oscilloscope-on-a-chip technology.
Mario Paniccia, director of Intel Corp.'s Photonics Technology Lab, also spoke about silicon photonics at the session and described the latest developments in Intel's research on the subject. Paniccia said photonics applications are commonly thought of for communications and electronic interconnections on PCs, but new areas of research include using photonics for wireless radio frequency distribution, environmental monitoring and even some biomedical applications.
Paniccia said Intel has a strong optical business and sees optics as the key to its future; included in that future is the opportunity to develop silicon photonics. He said there is the potential to integrate multiple optical devices with silicon photonics, and that micromachining could provide smart packaging. It might also be possible to converge computing and communications, he said; but to benefit from this, optical wafers must be able to be fabricated in a CMOS fab.
Silicon's benefits include its transparency in the communications band, its compatibility with CMOS and its ability to be produced inexpensively, Paniccia said. But it has drawbacks: Silicon has no electro-optic effect, no detection in the 1.3- to 1.6-um communications band and lacks efficient light emission, he said.
For those reasons, silicon traditionally was not considered a good optical material; but advances in the last few years have made fundamental leaps in silicon device performance, he said. A major goal now is to get to the point of integrating photonics and electronics processed together onto a single wafer. CMOS integration challenges include the fact that many optical devices are much taller than transistors, making integration on the same chip challenging, and high temperatures can't be used in the manufacturing process in later steps without causing damage.
Paniccia said researchers into silicon photonics should concentrate on improving device performance above 10 GhZ, on developing efficient light emitters — especially lasers — and Raman scattering devices such as amps, lasers and converters. Monolithic optical isolators for laser applications are also highly desired, as are new ways to use silicon for coupling and packaging. Researchers should also work on structures to enable wafer-level testing and optoelectronic integration, Paniccia said.
They should also keep in mind that new device concepts, especially emitters, will need to take CMOS manufacturability into consideration and consider ideas that are CMOS fab-friendly and that enable new functions on silicon.
Paniccia said long-term convergence applications will be in silicon, silicon photonic device performance is advancing at an accelerated pace, and that the next challenge will be with integration. If successful, volume economics could allow optics to impact many areas, from communications to biomedicine, Paniccia said.
Reprinted with permission from www.Photonics.com. Copyright 2006 Laurin Publishing. All Rights Reserved.