The initial work on developing a hybrid silicon laser began in the
summer of 2004 supported by Jag Shah at DARPA (the Defense
Advance Research Projects Agency) and Intel. Hyundai Park
explored the wafer bonding of InP active structures to silicon
based on the bonding knowledge base that had been developed
by the Bowers’ group in the 1990s. The hybrid silicon laser was
actually the third type of hybrid device that we had created.
Previous devices led to the first long-wavelength VCSEL made
of InGa AsP active regions bonded to Ga As mirrors as well as
avalanche photodiodes made of InGa As bonded onto silicon.
Early on, our team found the thermal expansion mismatch
between InP and silicon to be the key hurdle for transferring
InP epitaxial films to silicon, since conventional direct wafer
bonding techniques were done at high temperatures (650° C).
Large cracks would form in the transferred materials, rendering
them useless for large devices like photonic integrated circuits.
After many experiments, the team achieved the high-film-qual-ity transfer using a low-temperature (300° C) plasma-activated
approach. This change together with a vertical outgassing
channel design developed by Di Liang eventually led to the
bonding of 150-mm-in-diameter InP and silicon wafers.
Alexander Fang, Hyundai Park and Richard Jones attempted
to make optically pumped lasers using these wafer bonded
III-V layers on SOI waveguides. Early attempts failed, however,
because it was difficult to create high-quality laser facets through
the bonded interface; cleaving techniques yielded stepped facets
due to the incorrect orientation of the silicon and the InP layers.
1,560 1,600 1,640
(Top) Photo of an array of three distributed feedback (DFB)
lasers and the power monitors. (Bottom) DFB spectra.
Stronger bonding and better polishing techniques eventually
led to optically pumped hybrid silicon lasers.
Now that the team had demonstrated optically pumped
lasers, we set our sights on the Holy Grail of silicon photonics: an electrically pumped laser on silicon. This structure
provided unique challenges that had not been seen in typical
in-plane semiconductor lasers, since current couldn’t be passed
through the bonded interface. Instead, current below the
quantum wells had to flow from the center of the waveguide
and out laterally to the contacts. In June 2006, the team successfully demonstrated an electrically pumped hybrid silicon
laser that lased continuous wave.
Since our team demonstrated an electrically pumped hybrid
laser, other researchers have validated the hybrid silicon design
as a promising device platform—and the pace of work in
this area has rapidly accelerated. Yariv’s group at Caltech has
demonstrated a variety of high confinement structures, for
example. Bowers’ group and Arai’s group at the Tokyo Institute of Technology demonstrated distributed feedback (DFB)
lasers. These DFB lasers were important because they removed
the need for polished facets for reflectors and enabled the single
wavelength operation capability that is needed for multiple
wavelength applications such as for DWDM systems.
Researchers at Ghent University demonstrated four microdisk
lasers for dense wavelength division multiplexed (DWDM)
Wavelength[nm] .571.5751.581.5851.591.5951.61.6051.61 – 30 – 35 – 40 – 45 – 50 – 55 –60 –65 –70
Power [dBm/0.1 nm]
(Top) Photo of array of four microdisk lasers. (Bottom) Spectra
from the array.