(Invited) Vacancy-Type Defects and Their Carrier Trapping Properties in GaN Studied by Monoenergetic Positron Beams

Abstract

GaN and its related alloys are the most important semiconductors in the post-scaling-era. The excellent physical properties of GaN make it a promising semiconductor as well for electronic devices used in high-power and high-frequency applications [1]. Recently, techniques to grow GaN on Si substrates (GaN-on-Si) have been intensively studied. Among many requirements that must be met for practical GaN-on-Si devices, reduced vertical leakage current and the ability to block vertical breakdown are the most important. Carbon doping of the GaN layer effectively increases the layer resistance. A high concentration of carbon (> 1019 cm−3) is commonly required to obtain highly resistive GaN layers. This quite high carbon-doping is also expected to introduce point and structural defects in the GaN layer. Positron annihilation is a powerful technique to evaluate vacancy-type defects in semiconductors [2-5]. In the present study, we used monoenergetic positron beams to study the carrier-trapping mechanism of vacancy-type defects in GaN layers grown on Si substrates. Positron annihilation is a non-destructive tool for investigating vacancies in materials. When a positron is implanted into solids, it annihilates with an electron and emits two γ-quanta (Fig. 1). The energy distribution of the γ-rays is broadened by the momentum component of the annihilating electron-positron pair p L. A freely diffusing positron may be localized in a vacancy because of Coulomb repulsion from charged ion cores. Because the momentum distribution of the electrons in the defects differs from that of electrons in the bulk material, these defects can be detected by measuring the Doppler broadening spectra of the annihilation radiation. The change in the Doppler broadened spectra due to the trapping of positrons is shown in Fig. 2 schematically. The resultant changes in the spectra are characterized by the S parameter. Because the electron density in vacancies is lower than that in the bulk, the positron lifetime increases in vacancies. Thus, measurements of positron lifetime spectra are also used to detect vacancy-type defects. Figure 3 respectively show the photoluminescence spectra and the S values for the GaN layers with [C] = 2 × 1016, 9 × 1017, and 1× 1018 cm−3 as a function of the photon energy. The S value started to increase at a photon energy of 2.7 eV and saturated above 3.2 eV. The observed increase in S can be attributed to the change in the defect charge state from positive to neutral due to the trapping of the excited electrons by the defects (V + + e‒ →V 0). The species of positively charged defects was identified as a Ga vacancy (V Ga) coupled with multiple nitrogen (V N) vacancies [V Ga(V N)n]. Figure 4 shows a schematic band diagram of possible transition levels for the vacancy-type defects. The charge transition could occur through the direct capture of electrons emitted from the valence band. Using the onset photon energies which cause the increase/saturation in the S value (1.8, 2.7, and 3.2 eV), the possible transitions are shown as (a), (b), and (c) in Fig. 4. The energy required for the optical transition from CN - to CN 0 and that from CN 0 to CN + is 2.81 eV and 3.18 eV, respectively [5]. Thus, assuming that the electron emitted from CN is captured by vacancy-type defects, the following reaction is possible: CN ( -/0) + V + → CN (0/+) + V 0. The corresponding electron capture processes involving the charge transition of CN are shown as (d), (e), and (f) in Fig. 4. These results suggest that positron annihilation is a useful tool to study the carrier trapping phenomenon by vacancy-type defects in group-III nitride semiconductors. References Krause-Rehberg and H. S. Leipner, Positron Annihilation in Semiconductors, Solid-State Sciences (Springer-Verlag, Berlin, 1999) vol. 127Uedono et al., APEX 8, 051002 (2015). Uedono et al., J. Appl. Phys. 120, 215702 (2016).Uedono et al., Phys. Stat. Sol. B 2017, 1700521. A. Reshchikov et al., Phys. Rev. B 90 (2014) 235203. Figure 1