III-nitrides wide bandgap semiconductors currently offer a future alternative to maintain the growing demand for high power electronic devices. Current III-nitride device architectures already show promising results, offering higher breakdown voltages and reduced on resistances compared to traditional semiconductor alternatives. However, these devices are still limited by the classical Baliga figure of merit (BFOM) that consider the main properties of a power device, breakdown voltages and on- series resistance. In order to surpass the BFOM, novel device geometries have been introduced such as the superjunction (SJ). The concept of the SJ has been demonstrated in silicon, with devices such as CoolMOS and attempted in SiC. These have demonstrated capabilities which have surpassed the BFOM. Nevertheless, there are no demonstrations of III-nitride SJs in the literature based on one major limitation, the difficulty in obtaining selective area doping.In order to achieve III-nitride SJs, there are several research barriers to address concerning selective area doping. A SJ device basically requires alternating, lateral n- and p-type doping regions with zero net charge. Two approaches can be considered to address this challenge: ion implantation and polarity control. The first approach, ion implantation, is one of the basic tools for semiconductor device fabrication and as such any maturing semiconductor system should be capable of being processed in this form. Currently, III-nitrides dot not possess a robust ion implantation toolbox that allows for reliable implantation control and activation. Recent advances in ion implantation for the realization n-type AlN and p-type GaN will be discussed. For the case of n-type AlN, Si implantation was realized and with implementation of defect quasi Fermi level control, self-compensation via DX and vacancy complex formation was inhibited. Thus, the highest reported n-type conductivity in AlN was achieved. For the case of p-type GaN, we demonstrate the ability to successfully achieve p-type conductivity in GaN films via room temperature Mg implantation and a post-implantation annealing procedure at high pressure (1 GPa). The highest recorded p-type conductivity (~0.1 Ω-1·cm-1) was measured on the Mg implanted GaN and annealed at 1300 °C and 1 GPa.The second approach to realize the lateral doped regions is by polarity control. The inherent polar doping selectivity of GaN can be used to achieve the doping scheme for a lateral GaN p/n junction. Oxygen, which unintentionally incorporates into N-polar GaN at levels >1019 cm-3, acts as the n-type dopant, whereas Ga-polar GaN does not readily incorporate oxygen. Accordingly, lateral polarity junctions (LPJs) with alternating domains of O doped N-polar and Mg doped Ga-polar GaN have been fabricated to realize lateral p/n junctions. In addition to this lateral patterning, the proper doping profiles must be attained in the N- and Ga-polar domains for SJ operation. For drift regions, the n-type doping in the N-polar domain (and p-type doping in Ga-polar domain) must be reduced to ~1017 cm-3 for typical micron wide domains. By implementing the chemical potential control (CPC) framework, MOCVD process conditions were designed in order to decrease the oxygen concentration by increasing the growth supersaturation. This led to a reduction in oxygen from >1019 cm-3 to low 1017 cm-3.Due to the difficulty of making reliable Schottky rectifying contacts to the N-polar surface, the alternating Schottky/ohmic contacts on a typical superjunction are replaced with alternating p+/n- junction (N-polar) and p+/p-junctions (Ga-polar). In this direction, a N-polar p/n junction with rectifying behavior comparable to a Ga-polar p/n junction was achieved. Finally, the growth of a GaN LPJs must exhibit N-/Ga-polar domains with a smooth surface and equal growth rates. To achieve these requirements, we demonstrate a supersaturation modulated growth (SMG) where the V/III ratio, and thus supersaturation, was modulated between low and high values. A GaN LPJ with a smooth surface and equal domain heights, with the necessary doping profile is hence demonstrated.All these results show the pathway for realizing superjunction device structures in III-nitrides and the possibility of such device architecture.
Read full abstract
Aug 19, 2024
H Yazdani + 4
Open Access