UT Austin Scientists Develop Smallest Semiconductor Laser
CompoundSemi News Staff
July 30, 2012...At the University of Texas at Austin, physicists collaborated with researchers in Taiwan and China to develop what they claim is the world's smallest semiconductor laser. The development was reported in this week's Science. Applications such as ultrafast computer chips; highly sensitive biosensors for disease detection diagnosis and treatement, and next-generation communications require semiconductor lasers that are faster, smaller and use less energy.
The device is constructed of a gallium nitride nanorod that is partially filled with indium gallium nitride. When fired, the nanolaser emits a green light, but the laser is too small to be visible to the naked eye. The ultralow-threshold nanolaser has a single nanorod placed on a thin silver film (28 nm thick). The resonant electromagnetic field is concentrated at the 5-nm-thick silicon dioxide gap layer sandwiched by the semiconductor nanorod and the atomically smooth silver film. Physics graduate student Charlotte Sanders designed and built a special MBE system to to create the smooth silver thin film needed for the laser.
“We have developed a nanolaser device that operates well below the 3-D diffraction limit,” said Chih-Kang “Ken” Shih, professor of physics at The University of Texas at Austin. “We believe our research could have a large impact on nanoscale technologies.”
“Size mismatches between electronics and photonics have been a huge barrier to realize on-chip optical communications and computing systems,” said Shangjr Gwo, professor at National Tsing Hua University in Taiwan and a former doctoral student of Shih’s.
July 30, 2012...Plessey Semiconductors took delivery of an Aixtron CRIUS® II-XL reactor in a 7x6-inch wafer configuration. The reactor is the first part of a multi-million pound investment in an HB LED production line. The systems in the new production line will use Plessey's MaGIC™ (MAnufactured on Gan ICs) technology of GaN on 6-inch silicon substrates to produce the HB LEDs at Plessey's Plymouth facility.
Neil Harper, Plessey's HB LED Product Line Director, said, "We use a much thinner GaN layer at only 2.5µm compared to 6-8µm in other GaN on Si technologies. This means less deposition time so that we can do multiple production cycles in the reactor in 24 hours to achieve higher throughputs and lower costs."
The first samples producing the correct wavelength output from Plessey's 6-inch IC production line are being produced. Plessey notes that it is using standard, 6-inch silicon substrate, which offers significant cost reductions of about 80% compared to the current technologies using Silicon Carbide or Sapphire that are both expensive and hard to scale up. Plessey intends to move to 8-inch substrates for even greater cost savings. The company claims that typical MAGIC HB LEDs are yielding at 95% with over 14,000 1 sq. mm 1 Watt MAGIC HB LEDs per 6-inch wafer.
Barry Dennington, Plessey's COO, added, "MaGIC and EPIC are two unique, disruptive technologies that are instrumental in our plan to rapidly grow Plessey into a major electronics company producing smart lighting solutions."
Siemens Looks to Spin Off Osram Instead of IPO
July 30, 2012...Instead of a previously talked about IPO for Osram, Siemens announced plans to spin off the company. Osram ram includes: Osram Opto Semiconductors, the optoelectronics subsidiary of Osram, that manufactures optoelectronic semiconductor components including LEDs and high power laser diodes. The new plans came after the company missed analyst estimates of Euro 1.32 billion, due to an accounting charge relating to Osram. Also Siemens reported that new orders were down 23 percent in Q3 2012 compared to Q3 of last year.
Siemens CEO Peter Loscher said,
“Since market conditions continue to be volatile for an IPO, we have decided to pursue a spin-off to our shareholders as the most probable part of divestiture.”
Loscher noted that demand from China was slowing and said he doesn't expect recovery until 2013 or later.
“Our focus above all is on increasing our productivity and efficiency,” he added.
Toshiba to Start Mass Production of GaN on Silicon White LEDs
LIGHTimes News Staff
July 26, 2012...
Toshiba Corporation of Japan, announced plans to start mass production of white LEDs on a production line that the company will construct in the 200mm wafer facility at Kaga Toshiba Electronics Corporation, a production base for discrete products in northern Japan. Mass production will start from October 2012. Toshiba reportedly is applying GaN-on-Silicon technology to the development of white LEDs.
Since January 2012, Toshiba has collaborated with Bridgelux Inc. of Livermore, California USA, on GaN-on-Silicon technology for LEDs. Then, in May the companies announced a record 614mW output 1.1mm square blue-emitting LED operating at 350 mA and less than 3.1V. Content continues for LIGHTimes SecondPage members...
Azzurro Semiconductors Chooses Veeco MOCVD for GaN-on-Silicon Epiwafer Production
LIGHTimes News Staff
July 26, 2012...Veeco Instruments Inc. announced that Azzurro Semiconductors AG, a pioneer in GaN-on-Si technology headquartered in Dresden, Germany, has recently placed into production the TurboDisc® K465i™ gallium nitride (GaN) Metal Organic Chemical Vapor Deposition (MOCVD) System. The MOCVD system is used to make gallium nitride on silicon (GaN-on-Si) wafers for power semiconductors, LED wafers and LED template wafers.
The Veeco announcement comes just days after Azzuro announced that it had received $2.6 million Euro from the German state Saxony, for the development of GaN on 200mm silicon wafers. (Ref: Coverage). However, the Veeco system in the announcement is apparently for growing GaN on 150mm silicon wafers.
Dr. Markus Sickmöller, Vice President Operations of Azzurro, commented, “Our technology provides high-quality, crack-free GaN layers on 150mm standard silicon wafers with very good crystal quality and extremely low bow values. We chose Veeco’s K465i to produce these high efficiency GaN-on-Si stacks because of its performance, process stability, and high throughput. We believe our GaN-on-Si technology will trigger a new wave of highly efficient and cost-optimized components for the LED and power semiconductor industry.”
Jeff Hawthorne, Veeco’s Senior Vice President, MOCVD, commented, “The power electronics device market is forecasted by Yole Développement to be $25B by 2015, and GaN has the ability to make devices with higher efficiency and switching speed than traditional Si devices. Further, LEDs made with GaN-on-Si technology will enable lower LED costs. We anticipate that the demand for MOCVD tools for these GaN-on-Si applications will grow as technology moves from R&D into production. Veeco’s K465i systems offer key advantages for GaN-on-Si production, including low particle count and excellent yields, and we’re looking forward to continuing to support Azzurro’s growing GaN-on-Si business.”
Aixtron Introduces 5x200 mm GaN-on-Si Technology with the AIX G5 Reactor Platform
LIGHTimes News Staff
July 24, 2012...Aixtron SE of Aachen, Germany has introduced a 5x200 mm GaN-on-Si (Gallium Nitride on Silicon) technology package for its AIX G5 Planetary Reactor platform. Following a customer-focused development program, the technology was designed and created in Aixtron's R&D laboratory. The technology consisting of specially designed reactor hardware and process capabilities is now available as a part of the AIX G5 product family. Aixtron says that any existing G5 system can be upgraded to this latest version. Aixtron Details of G5+ have already been disclosed to some of Aixtron’s key customers.
The company says that some initial feedback has particularly noted the fully rotationally symmetrical uniformity pattern on all five 200 mm wafers, the use of standard thickness silicon substrates, and the controlled wafer bow behavior is exactly what they require for silicon-style manufacturing. The company says that AIX G5 reactor platform offers superior uniformity and yield, and extremely stable processes. According to Aixtron the AIX G5 has a higher device yield than any other MOCVD platform and has a capacity of 5x200 mm.
Dr. Rainer Beccard, Vice President Marketing at Aixtron stated, “The wafer size and material plays a crucial role when it comes to cost effective manufacturing processes, and thus the transition to 200 mm Standard Silicon wafers is a logical next step on the manufacturing roadmaps, as it offers unique economies of scale.”
”Being convinced that uniformity and yield are the key success criteria in 200 mm GaN-on-Si processes, Aixtron conducted a dedicated R&D program," added Dr. Frank Wischmeyer, VP and program manager of power electronics at Aixtron. “We started the development process by conducting an extensive simulation program, which enabled us to design fundamentally new hardware components that provide unique process performance in our 5x200 mm processes, while still being compatible with the well-proven AIX G5 reactor platform.”
“This uniformity pattern has been an inherent feature of Aixtron’s Planetary Reactor technology, which we can now successfully obtain on 200 mm GaN-on-Si-wafers," Dr. Wischmeyer stated.
UCSB and Mitsubishi Chemical Develop Violet Non-polar LDs Comparable to C-plane LDs
CompoundSemi News Staff
July 23, 2012...The University of California Santa Barbara (UCSB) and Mitsubishi Chemical Corp report having developed violet non-polar laser diodes that are comparable to C-plane laser diodes. Specifically they developed aluminum gallium nitride (AlGaN) cladding-free (ACF) m-plane, nitride semiconductor laser diodes (LDs). The researchers assert in an article in Applied Physics Letters that theses LDs have output powers comparable to single-stripe emitter c-plane devices.
Non-polar and semi-polar nitride crystal orientations can eliminate, reduce, or change the polarization electric fields arising in c-plane materials and active light emitting layers in nitride semiconductor devices. Cladding layers in laser diodes confine the electromagnetic field and create the conditions for stimulated emission. While AlGaN is often used for this, it has a large lattice mismatch with InGaN/GaN structures, and is difficult to grow with sufficient quality.
The researchers used separate-confinement heterostructures of InGaN or GaN and thicker InGaN quantum well regions to leverage their cladding-free technology that expanded the optical field distribution of the transverse mode. In theory such devices could be more resistant to sudden failure or gradual degradation, and allow higher-power operation.
MOCVD was used to grow a five-period mulitquantum well structure of InGaN with GaN barriers on a GaN substrate from Mitsubishi Chemical. Simulations suggest that the device should have a transverse confinement factor in the active region (Γact) of 0.128 at a lasing wavelength of 410nm (violet).
Genesis Photonics to Open Kunshan, China Facility This Fall
CompoundSemi News Staff
July 23, 2012...Genesis Photonics of Taiwan reportedly plans o invest US$35 million in 2012 in China to construct its plant in Kunshan, southern China. The plant is expected to be completed in the third quarter. Genesis Photonics produces blue, green, and white LEDs.
The plant is a joint venture between Genesis and the Kunshan municipal government, according to a Digitimes article. Construction on the plant began in September 2010 with about US$200 million investment. Genesis will provide 49% of the funding and the municipal government will provide 51%.
Company chairman David Chung said that Genesis will provide technologies and management, and the municipal government is offering government resources and assist in business development. The plant will procure 2-inch MOCVD equipment and expects to have a monthly capacity of 20,000-30,000 epitaxial wafers as commercial production begins.
According to the Genesis Photonics, in the future, the company will have three production facilities in Greater China: Tainan, Dongguan, and Kunshan. The Tainan plant will focus on R&D and sales to international customers.
Chung says he plans to have the two China-based plants be completely responsible for supplying LED chips to China before 2015.
Our news features are reported
by the CompoundSemi News staff writers.
For submissions or content suggestions, you can contact us using
editor -at - compoundsemi.com
For more information and to reserve promotion space contact
Info7 -at - compoundsemi.com
or call +1 (512) 257-9888
Commentary & Perspective...
GaN on Silicon Gets Real
Tom Griffiths - Publisher
July 26, 2012...Never ones to knock a new technology approach, we've been dutifully covering
the GaN on Silicon progress for a decade or so at this point. I can't remember
if it was at one of our 2000 or 2001 conferences (CS Outlook) that we listened
enthusiastically to the "near production availability" of GaN-on-Si
substrates from Motorola (before they had Freescaled). Cheaper, easier to handle,
processes move to standard silicon-type platforms, and since the wafers were
smaller than state-of-the-art, those production platforms were likely older
and already amortized making the cheaper even cheaper. Cool. Followed by, pretty
As time passed, it seemed we'd hear about more amazing progress in GaN-on-silicon
about every two years, with a blip from a university or other research house.
But each time it appeared to be a little over-hyped, as a decent lab result
failed to translate into a production-ready solution. However, that's how technical
innovation (as well as fame and fortune) often progresses. The "overnight"
successes are often decades in the making, as hard workers take two steps forward,
and somewhere between 1 and 3 steps back, but as long as that average is less
then two back, it creates progress. And progress builds on progress.
What it means for lighting... Before we dive down into some LED material/technology
kind of detail, it should be noted that GaN-on-silicon LEDs won't be likely
to supersede existing technologies. As far as the luminaire and lamp designers
would be concerned, what they will be looking at is a richer variety of price-performance
choices that could be enabled by lowered manufacturing costs for the GaN-on-Si
based solutions. That will first hinge on actual high volume production, which
may still be a while away. Toshiba, for example, announced this week that they
were going into mass production on GaN-on-Si for white LEDs, but somewhat confusingly
noted their progress in "milliwatts of output power" which is not
the typical way of quoting white LED performance (that's the single-color metric),
and that it would start in October on a new production line they are building
for it. While only a few months away, that's still a lot time for things to
happen, so wait and see is still the order of the day. When we do get GaN-on-silicon
LEDs in mass production, they also likely won't be running at the same peak
efficiencies as we see from the existing technologies, and efficiency is still
a big part of the game here, since more lumens per watt generally lets you pack
more lumens into a given amount of space. While we have plenty of space to produce
a streetlight that is plenty bright enough at 100 lumens per watt, things get
a bit more squeezy when we look at an A-lamp or worse, an MR. Lots of lumens
per square millimeter are required, and more lumens per watt usually correlates.
GaN-on-Si will be good for some things, but not likely for everything. "Traditional"
LED costs also continue to fall, and manufacturing efficiencies are increasing,
so I don't believe that one "winner" can be forecast, except for the
designers and buyers who will have more types of technology competing for their
business, and more competition is better when it comes to driving down costs.
The tech part... In the most basic sense, the problem everyone was trying
to solve is pretty simple. Gallium nitride, with its combinations and derivatives,
which support the LEDs in the bluish end of the spectrum, as well as high-frequency,
high power transmission types of technology as well as blue laser, is not "lattice
compatible" with silicon. The whole game of building a wafer that becomes
the foundation of a semiconductor device is that you want to blend the right
materials as evenly and defect-free as possible. Pure GaN would be a great starting
point for all of these devices, but "growing" a pure GaN substrate
is not easy and is devilishly expensive. Sometimes it's worth it, if you're
just trying to get a few thousand dollar transistors off of it, but in most
cases, you want more devices, and you want them cheaper, so GaN needs to be
"deposited" on top of something else. That something else (sapphire
most commonly, along with silicon carbide (SiC) and now silicon, has a different
crystal structure than the GaN, so if you just squirted some GaN layers onto
it, the GaN part would end up with a bunch of cracks or other features that
would lead to dead or prematurely dying devices. The solution is to add a "buffer-layer"
between the starting substrate and the GaN layers. Buffers can be either a mix
of materials or a series of layer steps that can act as a transition between
the substrate's crystal structure and the GaN's structure. A typical buffer
description could be something like is found in Taiwan National Cheng-Kung University's
2006 patent 7,014,710 that reads "...characterized in that a buffer layer
of SiCN [silicon carbon nitride] is grown to avoid lattice mismatch which appears
when Gallium Nitride is grown directly on silicon substrate". Note that
the buffer blends things that appear above and below it, silicon at one end
and a nitride at the other. While the buffer thing is "simple" as
a problem, it's not "easy" to solve, since you don't have the dynamics
of the crystal structure, but the fact that the melting point that allows you
to "spray" a composition onto one material may be a thermal stress
point for the other material, and once they are layered, they may heat or cool
at different rates, like that bi-metal strip that makes incandescent Christmas
lights blink... bending not good in this case.
So folks are figuring it out... Bridgelux was the first to set off the
latest round of "we're gonna make it work and change the face of the LED
industry" and went with a chips-all-in attitude when describing the future
of their LED technology. In a March
2011 story (scroll down a bit when you get there), they announced 135 lm/W
result for a 4700K GaN-on-silicon LED that was a result of "quietly dedicated
GaN-on-silicon team" that had been operating for the previous 5 years.
As we've seen, we never know til we see chips on the market, but "all of
sudden", we're hearing a lot more about GaN on silicon than ever before.
In a January 2012 news release, Osram Opto Semiconductors,
a company not known for brash "lab results" claims, announced that
their own GaN-on-Si program had chips in the pilot stage. With similar timing,
Cree, who bases their LEDs on silicon carbide rather than sapphire, began a
concerted effort to message on what they see as inherent advantages to that
approach, essentially, "GaN-on-SiC produces more efficient, more reliable
LEDs, and always will". It begs the question of "why now?" when
they've been in the sapphire vs. SiC trenches for a decade, and have simply
messaged on their results, which are consistently good. But if someone if they
were trying to counter an up and coming approach that is going to be making
a lot of noise in the near future, their "re-emphasized" message is
exactly the one I'd recommend. Validation that GaN-on-SiC might finally be a
In this week's news, GaN-on-Si was riding a wave as: A) Toshiba
announced they where kicking into full volume production as a result of
announced in May with Bridgelux for GaN-on-Si LED technology; B) Aixtron
announced the launch of their AIX G5+ 5x200 mm GaN-on-Si technology package
for its AIX G5 Planetary Reactor® platform; C) Azzuro
Semiconductors announced it had received 2.6 million Euro grant from the
government of Saxony (Germany) for a 200 mm LED and Power Semiconductor GaN-on-Si-wafer
development program which; D) which was followed by a mildly head-scratching
announcement the next day that it had sold them a TurboDisc® K465i™
MOCVD system to do 150mm GaN on Si work (so is that the precursor to their 200mm
work, or will we see Azzuro "spreading the love" with an upcoming
AIX G5+ 200mm purchase?). Sounds like we're getting close...
If you have news or
views to share about the compound semiconductor, LED or solid
state lighting industries
contact our Publisher, Tom Griffiths
His direct tel in Austin is +1-512-257-9888