Technical Insight
Printing provides wafer-scale integration of III-V lasers on silicon
Silicon photonics and high-density data storage could benefit from a printing process that enables GaAs lasers on new platforms
An international research team has used a wafer-scale printing process to transfer epitaxial GaAs coupons to a silicon wafer, before converting them into lasers.
The Fabry-Pérot ridge waveguide lasers that result produce 824 nm, continuous-wave lasing up to 100 °C and combine output powers in excess of 60 mW with modulation speeds of more than 3 GHz.
The researchers from Tyndall National Institute, Ireland, Semprius and Seagate, believe that their technology could aid the development of silicon photonics and drive advances in the hard-disk drive industry.
The latter requires light sources that deliver more than 10 mW of power to a waveguide with an integrated plasmonic near-field transducer, because this will allow data storage density to increase beyond 1 Tb inch-2.
Various approaches have already been used to form III-V lasers on silicon substrates, including direct epitaxial growth and a variety of bonding techniques.
However, according to corresponding author Brian Corbett from Tyndall Institute, the approach that he and his co-workers take is unique in several ways.
“All lasers are processed simultaneously on wafer with lithographic alignment to features on that wafer,” explains Corbett, who adds that this approach makes very efficient use of the III-V material and can accommodate differences in the dimensions of the III-V wafer and the host wafer.
What’s more, in principle this team’s technology can transfer multiple, different die onto a new substrate, enabling the formation of complex integrated circuits. And the technique is not limited to transfer to silicon wafers – it could be used for those made from AlTiC, a material widely used in the hard-disk drive industry.
“Direct integration of the laser with the read-write head, which is built on a AlTiC substrate, would enable the introduction of heat-assisted magnetic recording at an affordable cost,” writes the team in its Nature Photonics paper.
The team employed its elastomeric stamp to transfer an array of 100 µm by 400 µm GaAs coupons onto a silicon wafer, where they bond to the host via van de Waals forces.
“It was a real bonus to find such excellent bonding,” remarks Corbett. “Van der Waals forces work on the proximity of adjacent atoms, so it is essential to have a perfectly clean and flat interface.”
The transfer process uses material very efficiently. “Nearly 200,000 coupons could be harvested from a 100 mm diameter wafer,” says Corbett. “In principle, we only need a 10 micron ‘street’ around each individual device. Since this is shared with the adjacent device, with a 350 micron x 100 micron ‘coupon’, the wasted percentage is only 12 percent.”
After transferring its lasers onto silicon, engineers formed mirror facets by plasma etching to yield 370 µm-long cavities with ridge-waveguide widths of either 3 µm (single transverse mode) or 6 µm (multi-transverse mode). P-type ohmic contacts were added to the top of the ridge, and n-type contacts to etched recesses.
These lasers have a threshold of 17 mA, a slope efficiency from the facets of 0.8 W A-1, and a bandwidth in excess of 3 GHz. “The design we used was not optimised for high speed,” says Corbett, who explains that it would be easy to modify the architecture to increase the bandwidth.
He and his co-workers from Tyndall are now collaborating with Seagate in a Marie-Curie funded Industry Academia Partnerships and Pathways (IAPP) project called COMPASS. This aims to improve the reliability of the lasers and their performance that they deliver at elevated temperatures. “We also have proposals on integrating III-V gain elements with silicon photonics,” says Corbett.
An elastomeric stamp allows the transfer of an array of GaAs coupons onto silicon wafers, where they can be processed into lasers with output powers in excess of 60 mW.
The Fabry-Pérot ridge waveguide lasers that result produce 824 nm, continuous-wave lasing up to 100 °C and combine output powers in excess of 60 mW with modulation speeds of more than 3 GHz.
The researchers from Tyndall National Institute, Ireland, Semprius and Seagate, believe that their technology could aid the development of silicon photonics and drive advances in the hard-disk drive industry.
The latter requires light sources that deliver more than 10 mW of power to a waveguide with an integrated plasmonic near-field transducer, because this will allow data storage density to increase beyond 1 Tb inch-2.
Various approaches have already been used to form III-V lasers on silicon substrates, including direct epitaxial growth and a variety of bonding techniques.
However, according to corresponding author Brian Corbett from Tyndall Institute, the approach that he and his co-workers take is unique in several ways.
“All lasers are processed simultaneously on wafer with lithographic alignment to features on that wafer,” explains Corbett, who adds that this approach makes very efficient use of the III-V material and can accommodate differences in the dimensions of the III-V wafer and the host wafer.
What’s more, in principle this team’s technology can transfer multiple, different die onto a new substrate, enabling the formation of complex integrated circuits. And the technique is not limited to transfer to silicon wafers – it could be used for those made from AlTiC, a material widely used in the hard-disk drive industry.
“Direct integration of the laser with the read-write head, which is built on a AlTiC substrate, would enable the introduction of heat-assisted magnetic recording at an affordable cost,” writes the team in its Nature Photonics paper.
The team employed its elastomeric stamp to transfer an array of 100 µm by 400 µm GaAs coupons onto a silicon wafer, where they bond to the host via van de Waals forces.
“It was a real bonus to find such excellent bonding,” remarks Corbett. “Van der Waals forces work on the proximity of adjacent atoms, so it is essential to have a perfectly clean and flat interface.”
The transfer process uses material very efficiently. “Nearly 200,000 coupons could be harvested from a 100 mm diameter wafer,” says Corbett. “In principle, we only need a 10 micron ‘street’ around each individual device. Since this is shared with the adjacent device, with a 350 micron x 100 micron ‘coupon’, the wasted percentage is only 12 percent.”
After transferring its lasers onto silicon, engineers formed mirror facets by plasma etching to yield 370 µm-long cavities with ridge-waveguide widths of either 3 µm (single transverse mode) or 6 µm (multi-transverse mode). P-type ohmic contacts were added to the top of the ridge, and n-type contacts to etched recesses.
These lasers have a threshold of 17 mA, a slope efficiency from the facets of 0.8 W A-1, and a bandwidth in excess of 3 GHz. “The design we used was not optimised for high speed,” says Corbett, who explains that it would be easy to modify the architecture to increase the bandwidth.
He and his co-workers from Tyndall are now collaborating with Seagate in a Marie-Curie funded Industry Academia Partnerships and Pathways (IAPP) project called COMPASS. This aims to improve the reliability of the lasers and their performance that they deliver at elevated temperatures. “We also have proposals on integrating III-V gain elements with silicon photonics,” says Corbett.
An elastomeric stamp allows the transfer of an array of GaAs coupons onto silicon wafers, where they can be processed into lasers with output powers in excess of 60 mW.
