(Invited) DLTS Studies of Defects in n-GaN

Abstract

Introduction There exist various defects in GaN depending on growth methods or growth conditions or substrates used for the epitaxial growth. There are point defects, such as vacancy-related defects, antisite defects, interstitials as well as line defects, i.e. dislocations. These affect the performance of GaN-based devices, since they create deep levels in the bandgap and act as traps and generation-recombination centers. Foreign atoms, such as carbon and iron, also create deep levels. It is important to characterize these defects and identify their origins to realize the high performance of GaN-based devices. Current collapse in AlGaN/GaN high electron mobility transistors is well known to be caused by trapping of carriers [1]. In this work, we present our results on the characterization of traps in Si-doped n-GaN grown by metal organic chemical vapor deposition (MOCVD) on different substrates, that is, sapphire, n+-GaN, SiC and Si [2]. Experimental Lateral and vertical Schottky diodes were fabricated using Pt/Au or Ni/Au as a Schottky contact and Ti/Al/Ni/Au as an ohmic contact for n-GaN on sapphire, Si and SiC substrates and for n-GaN on n+-GaN substrates, respectively. Carrier concentrations determined from capacitance-voltage measurements at room temperature are in the range from 1016 to 1017 cm-3. Deep level transient spectroscopy (DLTS) using bias pulses has been applied to estimate electron traps for fabricated Schottky diodes. Minority carrier transient spectroscopy (MCTS) using above-band-gap light pulses has been used to characterize hole traps. Results and Discussion Temperature-scan DLTS and MCTS have been performed in the temperature range from 90 to around 400 K. DLTS measurements detected four electron traps labeled E1 (Ec-0.24 eV), E3 (Ec-0.57 eV), E6 (Ec-0.40 eV) and E7 (Ec-0.73 eV). E1 and E3 might be ascribed to VGa-VN pairs and NGa antisite defects, respectively. E6 and E7 are observed for n-GaN on SiC and Si, suggesting their correlation to dislocation-related defects. MCTS measurements detected three hole traps labeled H1 (Ev+0.86 eV), H2 (Ev+0.25 eV) and H3 (Ev+0.25 eV). Trap E3 and H1 concentrations are as high as 1016 cm-3 and show a large variation of trap concentration on each wafer. Moreover, trap H1 concentration increases in n-GaN intentionally doped with carbon, suggesting its correlation to carbon-related defects, possibly in addition to VGa-related defects. We have studied a distribution of trap E3 and H1 concentration for a quarter of 2-inch n-GaN on n+-GaN using isothermal DLTS and MCTS measurements at room temperature, respectively. A variation of trap E3 concentration over four orders of magnitude is found in the range from ~1012 to ~1016 cm-3 with the average concentration of 4.6x1014 cm-3, while trap H1 shows a relatively small variation in the range from~1015 to ~1016 cm-3 with the average concentration of 5.9x1015 cm-3. Average trap concentration of trap H1 is higher in n-GaN doped with carbon of 6x1016 cm-3 and is found to be 4.2x1016 cm-3. It is thought that the increased concentration of trap H1 in carbon-doped GaN corresponds to the concentration of carbon-related defects. In this context, we will present a method to distinguish between VGa-related and carbon-related defects with similar thermal emission activation energies by a combination of MCTS and optical deep level transient spectroscopy (ODLTS) using below-band-gap light pulses. Further, we will discuss on the identification of these traps by varying MOCVD growth conditions, namely, V/III ratios, growth pressures and growth temperatures. Acknowledgments This work has been performed as an MEXT-Supported Program for the Strategic Research Foundation at Private Universities (2010-2014).