Epitaxial growth and benefits of GaN on silicon
Armin Dadgar and Alois Krost
GaN epitaxial growth has a long history reaching back to the late 1960s , and even growth on silicon reaches back to the early 1970s . But, apart from a few devices demonstrated, it took more than twenty years to enter mass production. The biggest difficulty for GaN growth at the beginning of GaN research was the lack of suitable homosubstrates for epitaxial growth, leading to low material quality and a high background carrier concentration. Added to this was the difficulty of p-type doping, which required thermal annealing to activate the Mg acceptor. The most commonly used substrate for GaN growth has been, and still is, sapphire, but also SiC is used, mostly for high-power or high- frequency applications. Silicon, which offers the best crystalline properties, large substrate diameters, and lowest substrate cost, was always in principle attractive for GaN growth, but many difficulties appear when it is used as a substrate in GaN epitaxy. Around 2000, GaN-on-silicon technology made some important breakthroughs: the demonstration of an MBE-grown LED structure [3, 4], the development of thicker crack-free GaN layers using an AlGaN intermediate layer , and LT-AlN interlayers  grown by metalorganic vapor phase epitaxy (MOVPE). Since then it has developed rapidly, and in the meantime it has been proven as a competitive technology to the growth on sapphire in the case of LEDs and even as an advantageous technology in the segment of high-voltage electronics. We discuss the major problems of GaN epitaxy on silicon with a focus on the growth by MOVPE, and discuss solutions for epitaxy and device design to achieve state-of-the-art GaN-on-silicon layers, and demonstrate the benefits of such layers for device applications.