Technical Insight
III-Nitrides: Advances and controversy
Breakthroughs in power electronics, advances in ultraviolet emitters and new arguments associated with the debate on droop featured at the latest international nitrides meeting
On the last day of the conference Claude Weisbuch, a professor at both the Department of Materials at the University of California, Santa Barbara and CNRS-à‰cole Polytechnique in France, made a case for Auger recombination as the cause of droop in nitride-based LEDs.
DUNCAN ALLSOPP FROM THE UNIVERSITY OF BATH REPORTS
THERE'S NO LET UP in the pace of development of III-Nitride technology, judging from the research reported at the recent International Workshop on Nitride semiconductors (IWN-2014). And for those that gathered in Wroclaw, Poland, a vibrant and welcoming venue for this year's Workshop, there was the opportunity to revel in an ideal environment for reporting exciting advances in GaN power electronics and solid-state lighting.
While the application of III-Nitrides to solid-state lighting will have caught the eye of many round the world, often quite literally, it is quite possible that an even greater impact on humanity will be the deployment of this wide bandgap material in power electronics. When this will happen is hard to say, but it could occur sooner than we think, thanks to advances reported at the recent IWN meeting. At this gathering Daisuke Ueda of Kyoto University of Technology detailed his team's advances in the monolithic integration of GaN gateinjection transistors. These devices could underpin the next generation of efficient consumer products, like air conditioning units, with the chances of commercial success aided by Ueda's collaboration with Panasonic Corporation.
This was not the only advance in GaN electronics to catch the eye. Alexandros Georgakias of the Foundation for Research and Technology-Hellas (FORTH), Greece, reported plasma assisted MBE growth of GaN on substrates made from polycrystalline diamond. After crystalline diamond, this substrate has the highest thermal conductivity of materials known to mankind, making it an ideal heatextracting substrate that could enable even more efficient GaN power transistors. Georgakias's talk also outlined the fabrication of a GaN/ AlGaN HEMT on this heat-spreading platform. The breakthrough in this work is the absence of a crystallographic relationship between III-Nitrides and the polycrystalline substrate.
Another driver for deployment of GaN power electronics is the need for highefficiency, high-speed power converters for wind turbines and photovoltaic systems. Srabanti Chowdhury of Arizona State University in the US pointed this out in her presentation, where she championed vertical geometry GaN switches that are grown on bulk GaN substrates. This transistor architecture boosts the breakdown electric field to a value that is more than twice that realised in lateral HEMTs.
Scrutinising reliability
As GaN matures, there has been increasing effort devoted to factors that either limit the performance and reliability of devices, or impair the yield of processes used in their manufacture. Like forensic science, this is painstaking and at times frustrating detective work, where the outcome is often a "˜smoking gun', rather than a clean, categorical relationship between cause and effect.
Several talks in this area were given at IWN-2014, including invited papers from the University of Padova group and from Martin Kuball of Bristol University, UK. Both presentations reviewed how impurities and dislocations in the buffer layer below the channel of a GaN-based HEMT can impact this transistor's transient and high-voltage performance. In these devices, carbon and iron impurities appear to be a doubleedged sword, delivering the benefit of inhibiting parasitic electrical current flow, but paying the price of contributing to unwanted transient effects in GaN based HEMTs. Such a situation is by no means uncommon for semiconductor technologies, as solving one problem often introduces another, leaving the engineer or researcher to evaluate the best compromise for the device application.
While attending IWN, many delegates catch up with colleagues and discuss advances in III-Nitrides. Here Bernard Gill (left) from Montpellier 2 University, France, is in conversation with Bruno Daudin (centre) from CEA-Grenoble, France, and Yasushi Nanishi (right) from Ritsumeikan University, Japan.
Reliability is also a big issue for LED manufacturers, with dislocations fingered as the origin of many ills. Eliminating them is a major driver behind the worldwide research effort at developing nano-LED arrays and nanowire lasers "“ forming these involves the growth of dislocation-free GaN nanowires by either MBE or MOCVD.
Many presentations at the meeting detailed advances in this area, including three invited talks: a plenary presentation by Pallab Bhattacharya from the University of Michigan, who reviewed his group's pioneering work on light emitters formed from III-Nitride nanowires and quantum dots; a paper by Miguel Sanchez-Garcia, Universidad Politécnica de Madrid, on impressive developments in the growth of nanowire light emitters by MBE; and a presentation by Nathan Gardner, who reviewed the progress of Glo AB towards commercial exploitation of InGaN/GaN core-shell array LEDs. Gardner's paper described core-shell devices with peak emission wavelengths ranging from 415 nm to 605 nm, peak external quantum efficiencies exceeding 35 percent, and electrical efficiencies at low currents that can be as high as 95 percent.
Fundamental studies of the growth and properties of III-Nitride nanowire structures were also discussed in Wroclaw. Shunsuke Ishizawa from the Kishino group at Sophia University, Japan, claimed that a photonic band structure had been observed in regular arrays of GaN/InGaN nanowire LEDs. This confirms a result recently reported in the literature, and one that, as Ishizawa demonstrated, has the potential to enable flexible shaping of the light beam emitted from nanoLED array devices.
Meanwhile, a French-Swiss collaboration between CEA-CNRS, Grenoble, the Ecole Polytechnique Fédérale de Lausanne and the University of Rouen, produced a GaN/AlInN core-shell structure that efficiently emitted ultraviolet light.
This paper was one of several reports at IWN describing advances in UV LED and laser technology. Progress in this field includes improvements to AlGaN crystal quality on sapphire and in LED structures. These advances have propelled external quantum efficiencies to over 3 percent for UV-C (280 nm "“ 100 nm) and over 6 percent for UV-B LEDs (280 nm "“ 100 nm), according to Cyril Pernot of Nikkiso Giken Company, Nagoya, Japan.
Challenges to realising efficient, electrically pumped UV-lasers emitting below 280 nm were outlined by Tim Wernicke of the Technical University of Berlin, while Thomas Wunderer from PARC, a Xerox company, reviewed strategies for making compact UV laser sources.
Another area of UV optoelectronics, solar blind avalanche photodiodes, was the subject of a couple of talks, including one by Russell Dupuis' group at the Georgia Institute of technology. Dupuis reported an avalanche gain of greater than 7000 at a reverse bias of 102.5 V in devices fabricated from MOCVD-grown GaN on free-standing native substrates.
Close to the main entrance of Centennial Hall's is the "˜Pergola' - a colossal, semi-circular, ivycovered colonnade winding around one of the city's most magnificent and popular attractions - the multimedia fountain. Unveiled in 2009, 20 years after the first free elections in post-war Poland, WrocÅ‚aw's fountain projects water up to 40 m high through an array of 300 different nozzles, which can rotate, gyrate, pulse and even create a giant screen of water. Animated projections can be displayed on this screen while music orchestrates the show through the park's speakers.
Better foundations
Given the location of this international workshop, it would remiss not to report on the progress announced on freestanding GaN substrates. Two entire sessions of oral presentations were devoted to the growth of III-Nitride substrates and related issues. On top of this there were a series of excellent posters on these topics, plus a wide range of papers on optical and electronic devices fabricated on such substrates.
It is well known that GaN crystals of exceptional quality can be formed via the ammonothermal method. This approach is being pursued by Ammono S.A. of Poland, Soraa of California, and Mitsubishi Chemical Corporation of Japan, and at IWN 14 all three companies reported the production of GaN crystals with excellent structural quality − high resolution X-ray diffraction curves exhibit very sharp peaks, and the material has a very low dislocation density of less than 104 cm-2. It is these low dislocation densities that are exciting the interest of commercial chipmakers seeking an edge on the performance of the top range LEDs and transistors.
One downside of the ammonothermal method is its slow growth rate. To try and replicate the quality of these crystals while speeding up the process, Michal Bockowski and colleagues at the Institute of High Pressure Physics in Warsaw, Poland, are investigating the use of ammonothermal substrates as very effective seed crystals for the production of substrates grown by HVPE. This form of epitaxy can lead to high growth rates without recourse to the combination of high pressure and high temperature.
The Workshop also provided a stage for reporting the demonstration of a III-Nitride quantum cascade laser emitting at 5.4 THz. Speaking about this effort, Eva Monroy of the Université GrenobleAlpes, France, reviewed the latest results on intersubband transitions in nitrides. Of particular note was the enhancement of nonlinear phenomena at 1.55 μm, obtained when GaN/AlN quantum wells were replaced with quantum dots. This result could have commercial implications, because it will be of interest to those developing optical switches and electro-optic modulators for optical fibre telecommunications networks.
As well as attending the stimulating sessions of oral presentations outlined above, delegates benefited from two lively, packed poster sessions. If anything the poster sessions were overwhelming both in the number and quality of the posters, making it impossible to view everything of interest. Maybe this is something that the organisers of IWN-2016, to be held in Florida, should bear in mind in their planning.
The International Workshop on Nitride Semiconductors was held in Wrocław Congress Centre, which forms part of Centennial Hall complex.
Debating droop
Finally, these days, what international gathering of researchers in III-Nitride science and technology can pass without a little contention over the origin of efficiency droop in LEDs? IWN-2014 did not disappoint in this respect, with the very first session on III-Nitride Optical Devices accommodating two invited talks offering differing views on the origin of this malady.
First up was an invited paper by Bastian Galler from Osram, Germany. He described a model that shows that an Auger process involving two electrons and one hole (eeh) is far stronger than that based on a one-electronand-two-holes (ehh) Auger process. In a later paper Galler's colleague, Anna Nirschl, reported the impact of this work on droop. She described an experiment that is thought to visualise photoluminescence arising from hot charge carriers occupying a quantum well of wider band gap than the photon energy used to excite to luminescence in narrower-band-gap quantum wells. From this experiment the authors were able to attribute this high-energy emission to Auger processes.
Next up was Francesco Bertazzi of the Politecnico di Torino, Italy. He described new calculations of the Auger coefficients in the eeh and ehh processes. His group also considered the theory of hot electron transport in the structure used by the team from the University of California, Santa Barbara, and CNRS-Ecole Polytechnique in France, in their renowned experiment that enabled the detection of electron emission from an InGaN/GaN LED. Bertazzi concluded that the hot electron emission detected by the UCSB and CNRS-Ecole Polytechnique team cannot be unequivocally correlated to Auger-enhanced leakage. Countering this on the last day of the Workshop Claude Weisbuch, in a tour-de-force plenary talk, eloquently defended the interpretation of the UCSB-CNRS experiment as Auger recombination being the cause of efficiency droop in LEDs.
This is an area where understanding is undergoing rapid development, so it is not that surprising that disagreement is rife. It is also worth noting that the cause of hot electron effects can be difficult to isolate, especially in complex heterostructures where very high electric fields exist, as is the case in GaN LEDs, making the quest for the origin of efficiency droop a particular challenge.
When asked, Anna Nirschl stated that the Osram Auger visualisation experiment could only account for a fraction of the leakage current observed in many LEDs under high drive currents. For Augersceptics and uncommitted observers alike this admission, combined with the theoretical work of the Politecnico di Torino group, means that we can expect further entertainment from a debate that seems to generate more heat than light, much as Auger recombination does.
