http://www.newscientist.com/article.ns?id=dn7016
Engineers have passed a major milestone in their search for an elusive prize - a practical laser made of the silicon that is the heart of semiconductor chips.
Success would allow the integration of powerful electronic processors with lasers, so chips could distribute their clock signals optically - avoiding electronic bottlenecks as operating speeds increase.
Hundreds of millions of semiconductor lasers are sold each year, but at present they are all made from compounds such as gallium arsenide. These lasers emit light when positive and negative current carriers recombine at the junction between two layers of slightly different composition.
Recombination releases its energy efficiently as light in gallium arsenide and similar compounds - but not in silicon. Researchers have coaxed nanostructured, porous silicon to emit light, but after more than a decade of work the emission remains weak and poorly understood.
Electronic circuits can be made from gallium arsenide and integrated with existing semiconductor lasers. But developers want practical silicon lasers they can integrate with existing silicon technology, which would be much cheaper and produce significantly more powerful chips.
Bright pulses
In 2004, Bahram Jalali at the University of California at Los Angeles, US, took a new approach to silicon lasers. Instead of passing a current through the semiconductor, he illuminated it with bright pulses of laser light. The silicon atoms converted some of the light to a slightly longer wavelength through a process called stimulated Raman scattering, and this enabled him to make a laser.
Now, two new experiments have resulted in devices that "are already practical", Jalali told New Scientist. Both rely on guiding the laser light along the length of a channel on the chip with an electrical bias applied across it.
When the bias pulls current carriers out of the channel, it amplifies light more efficiently. In the online journal Optics Express, Jalali reports switching the laser beam off and on by changing the electrical bias across the channel, an essential feature for any practical laser.
Meanwhile, Haisheng Rong of the Intel Corporation has made the first silicon Raman laser to emit a steady beam - another practical necessity - by applying a steady bias across a longer channel, which he reports in this week's Nature. Jalali calls that "a major milestone". Intel adds that their design can easily be modified to switched the laser off and on.
Raman lasers have been demonstrated in other materials, but Rong's is the first one that can be modulated electronically - crucial if the beam is to carry signals. In addition to sending clock pulses that synchronise chip operation, the modulated beam could make high-speed connections between chips.
While it may take time to develop the large market needed to pay for integrated silicon opto-electronic circuits, Jalali believes silicon Raman lasers could also fill an important gap in the mid-infrared wavelengths, where no practical lasers currently exist.
Journal references: Nature (vol 433, p 725), Optics Express (vol 13, p 796)
Engineers have passed a major milestone in their search for an elusive prize - a practical laser made of the silicon that is the heart of semiconductor chips.
Success would allow the integration of powerful electronic processors with lasers, so chips could distribute their clock signals optically - avoiding electronic bottlenecks as operating speeds increase.
Hundreds of millions of semiconductor lasers are sold each year, but at present they are all made from compounds such as gallium arsenide. These lasers emit light when positive and negative current carriers recombine at the junction between two layers of slightly different composition.
Recombination releases its energy efficiently as light in gallium arsenide and similar compounds - but not in silicon. Researchers have coaxed nanostructured, porous silicon to emit light, but after more than a decade of work the emission remains weak and poorly understood.
Electronic circuits can be made from gallium arsenide and integrated with existing semiconductor lasers. But developers want practical silicon lasers they can integrate with existing silicon technology, which would be much cheaper and produce significantly more powerful chips.
Bright pulses
In 2004, Bahram Jalali at the University of California at Los Angeles, US, took a new approach to silicon lasers. Instead of passing a current through the semiconductor, he illuminated it with bright pulses of laser light. The silicon atoms converted some of the light to a slightly longer wavelength through a process called stimulated Raman scattering, and this enabled him to make a laser.
Now, two new experiments have resulted in devices that "are already practical", Jalali told New Scientist. Both rely on guiding the laser light along the length of a channel on the chip with an electrical bias applied across it.
When the bias pulls current carriers out of the channel, it amplifies light more efficiently. In the online journal Optics Express, Jalali reports switching the laser beam off and on by changing the electrical bias across the channel, an essential feature for any practical laser.
Meanwhile, Haisheng Rong of the Intel Corporation has made the first silicon Raman laser to emit a steady beam - another practical necessity - by applying a steady bias across a longer channel, which he reports in this week's Nature. Jalali calls that "a major milestone". Intel adds that their design can easily be modified to switched the laser off and on.
Raman lasers have been demonstrated in other materials, but Rong's is the first one that can be modulated electronically - crucial if the beam is to carry signals. In addition to sending clock pulses that synchronise chip operation, the modulated beam could make high-speed connections between chips.
While it may take time to develop the large market needed to pay for integrated silicon opto-electronic circuits, Jalali believes silicon Raman lasers could also fill an important gap in the mid-infrared wavelengths, where no practical lasers currently exist.
Journal references: Nature (vol 433, p 725), Optics Express (vol 13, p 796)
