Process engineers put down nitride films by cranking MOCVD reactors up to 1100°C and cracking ammonia and metal precursor molecules on the substrate. But this type of growth can be performed at far lower temperatures to create epitaxial films with minimal hydrogen by switching to a table-top Nanomaster tool that employs an RF source to split nitrogen gas into a plasma. Richard Stevenson talks to Nanomaster’s CEO, Birol Kuyel, about the pros and cons of this alternative growth technology.

Q What kick-started your development of a plasmaassisted MOCVD tool?

A I met Hae-Won Seo, a professor in the department of physics and astronomy at the University of Arkansas at Little Rock. She was just getting a new appointment, had limited funds, and wanted to replicate the clumsy, big MOCVD systems that she had seen or worked with in Taiwan.

For the money she had, she couldn’t buy or do anything with those systems. So I proposed that we could do something table-top. The temperature will not be as high, but at these lower temperatures she could still do MOCVD because we could provide her with plasma enhancement. With her funding we designed and built the system, delivering it in September 2008.

Q Your interest in plasma-assisted growth of nitride films goes back a long way, doesn’t it?

A Yes, I have 30 year-old patents for silicon nitride deposition from AT&T Bell Labs. At that time it was difficult to get the stochiometry and other properties of silicon nitride uniform throughout the film simultaneously. But we found that by activating nitrogen in a separate chamber, you could control the stochiometry independent of the composition and other physical parameters, like thickness, density and so on. Those experiments were done with nitrogen rather than ammonia.

So I had a similar expectation from gallium nitride – that plasma enhancement could help incorporate nitrogen at lower temperatures than is possible by the pyrolitic way that is widely used today.

Q What are the benefits of growing nitride films by plasma-assisted MOCVD?

A Nitrogen has some metastable states with long lifetimes, and plasma enhancement of nitrogen allows you to lower deposition temperatures. Reducing the temperature wherever you can is important, because higher temperature reactors are more difficult to build and require longer times to cool.

Another benefit is using nitrogen instead of ammonia. Then you don’t need the hydrogen source. Hydrogen is not a good thing in these films, whether its silicon nitride or gallium nitride. It forms a vacany, making an intermediate state that will interfere with the proper operation of the device. In fact, in my earlier silicon nitrogen work, hydrogen would even migrate from the capping material to the gate area and cause threshold shifts. Plasma-enhancement gives you an opportunity to reduce the hydrogen content and the temperature at which the films can be grown.

Q Are there other benefits to using nitrogen gas, rather than ammonia, for the nitrogen source?

A Yes, it has big impact on running costs. Dealing with ammonia requires process abatement. In the case of nitrogen, you don’t have that.

Q Are you the only manufacturer of a plasma-enhanced MOCVD tool?

A Yes. I think there is some work in Japan and Australia, but I don’t know the nature of this work. I’d like to know more, but I have not run into any detailed information.

Q What were the big challenges in making your first tabletop MOCVD tool?

A The most difficult part was designing everything to work together – putting everything into one system was very, very complicated. Our customer wanted features and upgrades, so we made the tool flexible for that – making it possible to inject liquids directly into the chamber. That adds to the complexity of the small volume inside the plasma source. Vacuum technology, plasma technology and delivering liquid precursors - we had done that before with a number of different systems. The small volume was a big challenge. In this small volume interactions between heater plate and plasma can take place. However, the heater plate can be powered from outside, making it possible to maintain the plasma in isolation.

Q Could your describe the gas flows in your reactor?

A We designed our own showerhead plasma source. Some of the gases go through the showerhead and some of them through a gas ring put in the downstream of the plasma source. Fitting all of this in a small volume was a nightmare. We did that and built a system that is PC controlled. A person doesn’t have to do anything manually. This system also has a turbo-molecular pump. Unlike other systems where you have pressures only going to milliTorrs, with this system you can go to the 10-6 to 10-7 range. So it is incredibly clean. All the turbo molecular pump heaters, chambers, power supply and RF connections are in the table-top cabinet.

Q Does the cabinet accommodate absolutely everything?

A We provided bubblers for our first customer. But the volume of some of the bubblers was too big for their applications and it also required filling at their location. They went to different bubblers and chillers, and all that assembly went down on the floor. The only other thing outside the cabinet is the mechanical pump.

Q Is it possible to put in-situ monitoring equipment into your reactors?

A Yes, we don’t have any limitations in terms of chamber size. What’s more, there is no limitation in this geometry for diagnostics compared to any other tool. Although you have a plasma source on top, you can always have a small area where you can look at the surface through an optical window. It is also possible to do optical thickness measurements at oblique angles, and for temperature uniformity, you can put a pyrometer into the chamber. It important to remember that the substrate is not rotating. In that sense, all the measurements will be even easier.

Q Today you are offering two tools: the NMC-3000, which accommodates a 2-inch wafer, and the NMC-4000, which can hold a 6-inch wafer. Both of these are intended for R&D work. If they were scaled, would they be suitable for high-volume manufacture?

A The way that the machine operates currently is to put down thin films at slow rates. Manufacturing applications need more throughput and higher deposition rates. That would require operating at higher pressures, which is counter-productive, because plasmas like to operate at pressures in the sub-Torr range. If you go to 10 Torrs or 100 Torrs, like you do in an industrial machine, you may not be able to enhance with the plasma. So the window where plasma enhancement may be applicable is not as broad as all applications of MOCVD. However, it may that for a subset of certain MOCVD applications, where you may get certain benefits, you might comprise on throughout for quality.

Q Do you have any plans to develop a multi-wafer tool?

A What we would like to do – and this requires the availability of funds – is to build a cluster tool using an NMC-4000. A number of tools, served with a robotic load-lock, could be run in parallel to reach production levels of throughput.

Q What are you short term goals?

A Six months down the road we will have a lot more information because our demonstration system will be complete by then. Then we will try MOCVD at much higher pressures. Today, you can grow clean, uniform, high-quality films, but the rates are low. They can be enhanced by special design of the plasma source.

We will also be able to obtain uniformity data for our tool. I wouldn’t expect any problems with that because the plasma source, the chamber and heating system are all much larger than the substrate size. And there is a perfect circle of symmetry in our chamber.

© 2011 Angel Business Communications. Permission required.

 



Nanomaster’s plasma-enhanced MOCVD tool is capable of producing nitrides with very low levels of hydrogen impurities

 



Birol Kuyel is president and CEO of Nanomaster.

He has a broad portfolio of expertise, including

high-temperature plasma physics, Si3N4 film deposition

and characterisation, X-ray and deep UV source

development. He has been awarded 9 patents and

published numerous papers.

 

Growing InN nanorods by plasma-enhanced MOCVD

Hae-Won Seo’s research team at the University of Arkansas at Little Rock, in partnership with researchers at National Sun Yat-Sen University, Taiwan, have employed the Nanomaster plasma-enhanced MOCVD tool for the growth of InN nanorods. These nanostructures have interesting characteristics, including strong quantum confinement and very high surface-to-volume ratio, but their growth is challenging. If the growth temperature is 500 °C or more decomposition of InN occurs, but switching to lower temperatures tends to increase the density of defects and dislocations in the crystal, leading to high carrier concentrations. The USTaiwanese partnership addressed this issue with a MOCVD-based technique that employs a 50 W, 13.9 MHz nitrogen plasma, a flow ratio of nitrogen to indium of 6000:1 and base and chamber pressures of 5 x 10-6 Torr and 2 x 10-1 Torr, respectively. Turning to this approach led to an optimum growth temperature of 500 °C, which enabled the growth of highquality InN nanorods on silicon (111) substrates.

Scanning electron microscopy revealed that the nanorods are straight, have a diameter of 90-120 nm, are typically 3.2-3.5 μm in length, and are uniformly dispersed on the substrate with an areal density of 3-5 x 107 cm-2. The researchers have also studied the photoluminescence spectra produced by their nanorods at a range of temperatures between 7K and 160 K. At the lowest temperature, the full width half maximum of the photoluminescence peak is just 27 meV. The narrow peak is claimed to be indicative of a high material quality and a low intrinsic carrier concentration.

The work of the researchers is reported in detail in J. Nanosci. Nanotechnol. 10 6783 (2010)