Slashing Temperatures For Nitride Growth
Deposition of nitride epilayer stacks by MOCVD requires high temperatures and plenty of ammonia. But these downsides can be sidestepped with an alternative growth process called migration enhanced afterglow,which has been developed by Canadian start-up Meaglow. Richard Stevenson reports.
Sales of MOCVD tools have reached staggering levels. According to IMS Research, around 800 of these reactors were shipped in 2010, and this year the figure is forecast to be 4 percent higher. Many of these tools are heading to China, where they will be used to deposit epitaxial stacks of InGaN and GaN layers, which will form the key ingredient in billions and billions of LEDs.
From a commercial perspective, it is obvious that deposition of nitride layers – partly by chipmakers in China, but predominantly by LED makers in other parts of the world – is a great success. However, that does not mean that there is no room for improvement. The MOCVD growth technique has several downsides, including high temperatures needed for deposition of the nitride layers – typically around 1000 °C. This is a big issue: GaNbased LEDs are grown on silicon, sapphire or SiC, and high growth temperatures exacerbate epiwafer bow that is caused by differences in thermal expansion coefficients between the epilayers and substrate.
This is a particular challenge for leading chipmakers that are migrating to larger diameter substrates to make LED lighting more affordable, because as the epiwafer get bigger, distortion is more pronounced. Complex buffer layers incorporating strain management can combat bowing, but a more attractive solution is to simply grow the epilayers at lower temperatures.
Slashing growth temperature has other benefits too: A greater range of substrates is then available, including ZnO, which is temperature sensitive and closely latticematched to GaN; and it is also possible to grow indiumrich InGaN layers at lower growth temperatures. The latter advantage will help expand the spectral range of green LEDs, and also aid the development of other classes of device, such as higher mobility field effect transistors and solar cells covering the entire spectral range.
An alternative, well established deposition technique with the potential to grow nitride films at lower temperatures is MBE. If this is to follow in the footsteps of MOCVD, growth should predominantly be N-face GaN. However, when MBE is used to form layers of such films at low temperatures, they tend to have rough surfaces. That’s because columnar poly-crystals form, which have pyramidal tops. Switching to the Ga-face – which is renowned for yielding better quality material that is suitable for the development of most GaN-based devices including LEDs and laser diodes – is problematic, because it is harder to deposit this class of material at low temperatures. It can be done, but MBE growth of Ga-face material has only been widely successful on MOCVD grown GaN templates, or on AlN buffer layers grown at higher temperatures. Direct growth of Ga-face material on nitrided sapphire is largely unheard of.
Fortunately, another growth option is now available – migration enhanced afterglow. This borrows some insights from MBE, but it is afundamentally different technology, combining far higher pressures of close to a Torr with a CVD-based plasma technique. Trailblazing this novel growth process is a spin-off of Lakehead University in Northwestern Ontario, Canada, called Meaglow. This start-up that was formed in late 2009 is now starting to commercialise its growth technique via a two-pronged approach: It is producing growth tools with low capital cost; and it has plans to offer epiwafer services for the growth of InN films later this year.
The stainless steel UHV growth chamber (left) and computerized electronic control system (right) of the prototype Meaglow reactor housed at Lakehead University
The current driving force behind the Thunder Bay startup is Chief Scientist Scott Butcher, a veteran of InN film growth with a strong academic and industrial background: “We’ve been growing for 6 months with the prototype system, and large leaps are being made forward in these early days. However, we’re still nowhere near reaching the limits of what it can do."
Butcher is willing to offer some insights into the pioneering deposition process. He explains that the high pressures associated with migration enhanced afterglow cause the high-energy plasma species, which are also present in MBE, to be largely converted into active species with lower energies. “High energy bombardment encourages N-face growth," adds Butcher. “By avoiding such conditions, using predominantly lower energy species, we have been able to grow Ga-face material at 630 °C directly on nitride sapphire."
Examples of success to date include GaN and InN films with a very low surface roughness. Atomic force microscopy scans on a 200 nm-thick GaN film revealed a root-mean-square (RMS) surface roughness as low as 0.24 nm. “We’ve also seen atomic terracing for InN grown at 470 °C. This has a 0.10 nm RMS surface roughness." Crystal quality of both these films is very good, according to X-ray diffraction measurements. For the GaN film, engineers at Meaglow have recorded (0002) ω-2θ XRD reflections with a full-width halfmaximum as low as 223 arcsec, and corresponding values for the InN layer of 290 arcsec.
Above left: Migration enhanced overflow can form Ga-face GaN films with a thickness of 200 nm at 630 ºC. Atomic force microscopy reveal that the root-mean-square surface roughness of this film is 0.23 nm. Molecular terraces can be distinguished in the image. Above right: Atomic force microscopy reveals that the InN surface has a root-mean-square roughness of 0.1 nm and features molecular terraces
Standing out from the crowd
Butcher’s interest in low temperature nitride growth canbe traced back to his days as a PhD student in the early1990s when he worked in the group of the late TrevorTansley from Sydney’s Macquarie University, Australia.“When [Shuji] Nakamura first demonstrated his blueGaInN/GaN LEDs, huge resources from Japan, Americaand Europe were diverted into MOCVD growth of GaN,"reminisces Butcher. “Trevor had the foresight not to try and compete with the larger groups overseas. Our group, who had been active in nitrides a good ten years before then, took a different route, concentrating on the low temperature growth of nitrides."
Working in partnership with colleagues Bing Zhou and Xin Li, Butcher constructed a low-temperature film growth system using laser-induced CVD (LICVD). This included a remote plasma microwave source that Butcher developed. “However, later on I dropped the laser system out of the development as the uniformity using LICVD was too problematic."
Butcher’s career briefly headed into new directions: He worked for Pacific Solar (now CSG Solar AG) from 1995 until 1997, while he finished his PhD; and for two years after that he was employed by the Australian Nuclear Science and Technology Organisation. However, he was still an Honorary Research Associate at Macquarie University, and in 1999 he was able to return to full-time research at this institution, after his and Tansley’s work caught the attention of Colin Wood from the US Office of Naval Research. Wood was able to fund Butcher’s research.
During the middle of the last decade, Butcher started to think how it would be possible to exploit the low temperature deposition technology developed at Macquarie University. This culminated in the launch of the spin-off BluGlass in 2005, a company that set itself the ambitious task of building optoelectronic devices not only on sapphire, but also on glass, an incredibly cheap substrate. According to recent announcements by Bluglass on the Australian Stock Exchange, its material’s crystal quality still needs improvement.
As Bluglass made progress, the health of Butcher, the firm’s Chief Technology Officer, went into decline – he was diagnosed with cancer. Although he received successful treatment in 2007 and 2008, he needed a year out to recover. And as he regained his strength, he started to mull over what he should do next. “I found that I wanted to get back to science and tackle some of the fundamental problems of low-temperature growth by developing a new technique that is a generation or two beyond what I was doing before." This dream has become reality thanks to an opportunity through his friendship with the academic Dimiter Alexandrov from Lakehead University. Alexandrov offered Butcher a role at Meaglow.
Recently, Meaglow has been wrestling with approachesto improve the crystal quality of nitride material grown atlow temperatures. One issue is the presence ofimpurities – dopants or otherwise. At concentrationsabove the solubility limit, these impurities lead toinclusions, extended defects, and, in extreme cases,crystalline boundaries. Of all the impurities, oxygen isthe biggest concern with plasma-based deposition. Themain source of oxygen contamination is from thedielectric windows typically used for both microwaveplasma sources and RF induction plasma sources,which can be addressed with a routine of passivationthat can take up to three days after opening the systemto air. However, Meaglow has learnt to sidestep theseissues by developing a high-density plasma source thatdoes not use a dielectric window and is scalable to verylarge areas.
Metalorganic vapour delivery system used for the Meaglow reactor. Operating at close to 1 Torr carrier gases are not required for most metalorganics - greatly simplifying the vapour delivery system compared to MOCVD
The Canadian outfit has also overcome a more fundamental challenge. Reducing growth temperature slashes the surface mobility of the gallium atoms, which drop to a fraction of that associated with MOCVD growth temperatures. Ideally, these atoms must diffuse far enough to reach the end of an atomic terrace before combining with nitrogen in order to realize two dimensional epitaxial growth. “If the surface mobility is too low, and the time to form a GaN molecule is too short, the diffusion length of the gallium atoms may be well short of this distance," explains Butcher. “Three dimensional growth then dominates and polycrystalline N-face material is formed."
It is not possible to overcome this limitation with conventional crystal growth methodologies, unless unreasonably slow growth rates are employed. “However, physics let’s you cheat this process," says Butcher, who explains that the trick to overcome this problem has its roots in migration enhanced epitaxy methods developed for MBE. “The idea is to saturate the surface of the substrate with a pulse of so much metal that it’s basically everywhere on the surface, while maintaining at most a low flux of active nitrogen." The metal is then slowly consumed after the metal pulse is ended by the plasma species introduced during the growth. “However, there is enough time for the gallium atoms to find energetically favourable lattice positions that allow two dimensional growth to proceed," adds Butcher. A series of such pulses can build up a film of desired thickness. Butcher points out that this form of epitaxy has been around for several years, but was traditionally much slower than normal film growth. “Recent advances have overcome this limitation and Meaglow has taken advantage of that in its new film growth system."
The company is now looking to win sales for these tools and develop its epiwafer services. The strengths of the novel reactor are not just its ability to grow InN and GaN films at low temperatures: Running costs are low, because the deposition process is free from ammonia; and it is relatively easy to control the diffusion length of gallium atoms, which hold the key to the growth of nanostructures such as quantum dots and nanowires. The Thunder Bay start-up is also demonstrating and refining its technology via a prototype tool housed at Lakehead University, and it has a research contract with its alumni to develop the migration enhanced afterglow technology for high-speed FETs, and other device applications. Continued efforts will help to develop a commercial, alternative technology to the wellestablished MOCVD process, which has weaknesses that growth by migration enhanced afterglow addresses.
A wide view of the prototype Meaglow reactor, showing the UHV central growth chamber with load lock, RGA and RHEED analysis chamber,MO delivery system (on left), control electronics (on right)
© 2011 Angel Business Communications. Permission required.