Solubility Limit of Ge Dopants in AlGaN: A Chemical and Microstructural Investigation Down to the Nanoscale

Abstract

Attaining low resistivity AlxGa1−xN layers is the keystone to improve the efficiency of light emitting devices in the ultraviolet spectral range. Here, we present a microstructural analysis of AlxGa1−xN:Ge samples with 0 ≤ x ≤ 1, and nominal doping level in the range of 10 20 cm 3 , together with the measurement of Ge concentration and its spatial distribution down to the nm scale. AlxGa1−xN:Ge samples with x ≤ 0.2 do not present any sign of inhomogeneity. However, samples with x > 0.4 display µm-size Ge crystallites at the surface. Ge segregation is not restricted to the surface: Ge-rich regions with a size of tens of nanometers are observed inside the AlxGa1−xN:Ge layers, generally associated with Ga-rich regions around structural defects. With this local exceptions, the AlxGa1−xN:Ge matrix present an homogenous Ge composition which can be significantly lower than the nominal doping level. Precise measurements of Ge in the matrix provide a view of the solubility diagram of Ge in AlxGa1−xN as a function of the Al mole fraction. The solubility of Ge in AlN is extremely low. Between AlN and GaN, the solubility increases linearly with the Ga mole fraction in the ternary alloy, which suggests that the Ge incorporation takes place by substitution of Ga atoms only. The maximum percentage of Ga sites occupied by Ge saturates around 1%. The solubility issues and Ge segregation phenomena at different length scales likely play a role in the efficiency of Ge as n-type AlGaN dopant, even at Al concentrations where Ge DX centers are not expected to manifest. Therefore, this information can have direct impact in the performance of Ge-doped AlGaN light emitting diodes, particularly in the spectral range for disinfection ( 260 nm), which requires heavily-doped alloys with high Al mole fraction.

Role of Metallic Adlayer in Limiting Ge Incorporation into GaN

“…The “memory effect”, if not properly managed, will have even more severe repercussions on the growth of heavily doped heterostructures, such as tunnel junctions, where precise control of doping profiles and interfaces determines the device performance. For thick, heavily doped GaN:Ge layers, it also leads to Ge precipitates, as reported before by others [ 25 ] and observed also for our crystals.…”

“…In spite of that, early efforts to obtain GaN:Ge by plasma-assisted MBE reported the formation of both Ge and Ge 3 N 4 precipitates for concentrations exceeding 4 × 10 20 atoms/cm 3 as reported by Hageman et al [ 24 ]. More recently, the doping limit for plasma-assisted MBE of AlGaN:Ge was identified by Bougerol et al [ 25 ]. The authors claim to have found the pronounced solubility limit for germanium in AlGaN layers with Al contents higher than 0.2, leading to the formation of Ge crystallites and Ge inclusions, mainly formed near defects or metal droplets on the surface.…”

“…During GaN growth, a considerable number of germanium atoms tended to stay at the surface rather than being incorporated. The use of indium (instead of gallium) as a surfactant during growth resulted in no Ge-related precipitates [ 25 ]. By growing low-indium-content InGaN:Ge, we present one of the highest electron concentrations obtained in nitrides.…”

Inversion of the Internal Electric Field due to Inhomogeneous Incorporation of Ge Dopants in GaN/AlN Heterostructures Studied by Off-Axis Electron Holography

“…To attain doping concentrations beyond the Mott transition (≈10 19 cm –3 ), the use of Ge is preferred, − due to the proximity of the Ge–N and Ga–N binding energies. However, the solubility limit for Ge in GaN is attained around 4 × 10 20 cm –3 , , and it decreases with the incorporation of Al in the material …”

Behavior of the ε-Ga2O3:Sn Evaporation During Laser-Assisted Atom Probe Tomography

“…In the last decade, laser-assisted atom probe tomography (La-APT) was widely employed in order to analyze the composition of dielectrics and semiconductors at the nanometer scale, including dopant atom distribution and chemical mapping of defects, providing fundamental new insights into the atom-scale mechanisms taking place during material growth and device fabrication (Grenier et al, 2014; Rigutti et al, 2016; Di Russo et al, 2017 a , 2017 b , 2020 a , 2020 b ; Dumas et al, 2019; Amichi et al, 2020; Cuduvally et al, 2020; Dimkou et al, 2020; Bougerol et al, 2021; Dumas et al, 2021). Such local quantitative analyses require to measure elemental composition with high accuracy (i.e., sensitivity in the range of 10 18 –10 19 cm 3 ) and near-atomic scale resolution.…”