Silicon-based devices are approaching their physical limits in applications like data transmission, satellite communication, radar systems and autonomous driving. Silicon devices can hardly be scaled down beyond the current state of research. And if the ever-increasing amounts of data should be processed using state-of-the-art silicon technology, the server rooms would occupy such a large area that it would be economically and ecologically unsustainable.
So-called HEMTs (High Electron Mobility Transistors) surpass the capabilities of silicon devices manifold. The properties of the materials on which they are based are crucial for the success of HEMT structures. Aluminum scandium nitride (AlScN) has excellent properties that enable higher carrier concentrations compared to any other material. In future, significantly more powerful and efficient HEMTs may be realized with AlScN.
However, manufacturing AlScN entails significant challenges. Cutting-edge technologies permit sputtering of AlScN layers. Yet their quality is not sufficient for electronic applications. Alternatively, AlScN can be grown by molecular-beam epitaxy (MBE). Using this method, high scandium content can be obtained in the compound. Also the quality is sufficient for manufacturing microelectronic devices. However, the process is complex and the productivity is too low to implement it on an industrial scale.
Manufacturing AlScN by metal-organic chemical vapour deposition (MOVCD) promises both sufficient quality and productivity for industrial use. In this process, gas is passed over a heated wafer. As a result, certain molecules are released from the gas and incorporated into the crystal structure of the wafer. By regulating gas flow, temperature and pressure, the formation of the crystal can be precisely adjusted, and the rapid exchange of gases allows different layers of material to be grown on top of each other.
"We knew from the past that researchers had tried to produce gallium scandium nitride using MOCVD, but were not successful. We also know that many researchers around the world are working to develop AlScN transistors, but no one else has managed to use MOCVD before us, although this could be a promising path for the industry," explains Stefano Leone, group leader at the Fraunhofer Institute for Applied Solid State Physics IAF.
The challenge for the researchers at Fraunhofer IAF has been that there is no gas available for scandium. The molecules (precursors) for scandium are very large and difficult to get into the gas phase. "We investigated which could be the best precursor for scandium and considered how we could modify our MOCVD reactor for the necessary processes. We did a lot of research and discussions and finally developed a setup. Ultimately, we managed to grow AlScN layers via MOCVD with very high crystal quality and the right amount of scandium to develop the next generation of electronic power transistors," comments Leone, pleased with the outcome.
Following the successful deposition of the AlScN by the MOCVD system, also the first layers for transistors were manufactured. With a sheet resistance of about 200 Ω/sq., an electron mobility of about 600 cm²/Vs and a charge carrier density of about 4.0 × 1013 cm-2, these layers already achieve promising results.
Now, the researchers' goals are to reduce resistance, increase mobility and further optimize material quality. The objective is to further improve the performance of future transistors and to bring Fraunhofer IAF closer to its goal of providing AlScN HEMTs for industrial power electronic applications.