Determination of the concentration of impurities in GaN from photoluminescence and secondary-ion mass spectrometry

Abstract

photoluminescence (pL) was used to estimate the concentration of carbon in Gan grown by hydride vapor phase epitaxy (HVPE). The PL data were compared with profiles of the impurities obtained from secondary ion mass spectrometry (SiMS) measurements. comparison of pL and SiMS data has revealed that apparently high concentrations of C and O at depths up to 1 µm in SIMS profiles do not represent depth distributions of these species in the Gan matrix but are rather caused by post-growth surface contamination and knocking-in impurity species from the surface. in particular, pL analysis supplemented by reactive ion etching up to the depth of 400 nm indicates that the concentration of carbon in nitrogen sites is below 2-5 × 10 15 cm −3 at any depth of Gan samples grown by HVpe. We demonstrate that pL is a very sensitive and reliable tool to determine the concentrations of impurities in the Gan matrix.GaN is a key material in modern solid-state lighting technology. The investigation and identification of point defects in GaN is very important, with immediate technology relevance to longer lifetime of light-emitting devices. Photoluminescence (PL) is a powerful tool for investigation of point defects in GaN 1 . Only few point defects in unintentionally doped GaN are well understood and reliably identified. One of them is the C N defect with the −/0 and 0/ + thermodynamic transition levels at 0.916 eV and ~0.3 eV, respectively, above the valence band 2-5 . This defect is responsible for the yellow luminescence (YL1) band with a maximum at 2.2 eV and zero phonon line (ZPL) at 2.59 eV in n-type GaN. Transitions via the 0/ + level of this defect can be observed as the blue luminescence (BL C ) band with a maximum at 3.3 eV at high excitation intensity 4 . The YL1 band is commonly observed in n-type GaN layers grown by metalorganic chemical vapor deposition (MOCVD), where the concentration of unintentionally introduced carbon is usually in the range of 10 16 -10 18 cm −3 , depending on growth conditions 6,7 . Undoped or Si-doped GaN samples grown by hydride vapor phase epitaxy (HVPE) contain much less carbon (10 15 -10 16 cm −3 ) 8-11 , yet still the YL1 band may be strong in these samples due to high hole-capture cross-section for the C N defects 12 .We have recently demonstrated 12 that PL can be used for determination of concentrations of defects responsible for PL bands. However, the concentrations of impurities found from PL were inconsistent with the concentrations found from secondary-ion mass spectrometry (SIMS) analysis in that study. Comparison of these two techniques is complicated by the following ambiguity. SIMS measurements provide depth profiles of impurity concentrations regardless of positions of impurity species (lattice sites or precipitations), whereas PL signal, excited with above-bandgap light, is collected from impurity atoms at lattice sites in the near-surface region (roughly, top 0.4 µm layer in GaN). The problem is that close to the surface, where PL is collected, very high concentrations of carbon an...

Comprehensive characterization of vertical GaN-on-GaN Schottky barrier diodes

“…Close to the surface, very high Si concentration is noticed from the SIMS profile in Fig. 5, as similar to the literature reports [16][17][18]. The depth of this dopant tail from the surface to the bulk is about 100 nm.…”

“…The depth of this dopant tail from the surface to the bulk is about 100 nm. The tail formation may be visualized as follows [16]: During SIMS measurements, the primary ion-beam sputtering can knock the dopant atoms from surface contaminations into a certain depth (about 50 nm) in the epilayer. The tails can be further extended due to the out-diffusion process through the dislocations and grain boundaries.…”

Photoluminescence from Be‐Doped GaN Grown by Metal‐Organic Chemical Vapor Deposition

“…The significant uncertainties for the concentrations of C and O in samples R26, R29, and R48 are related to the presence of a measurement artifact characterized by tails in the concentration–depth plots caused by kick‐on of the impurity atoms from the surface and/or relatively large surface roughness. [ 43 ] Sample R50, an undoped ≈1400 nm thick layer, was contaminated with Be due to the “memory effect” after several MOCVD growths with Be source. A spike of Be concentration can be seen at the interface with the GaN substrate (Figure 8).…”

Mass Spectrometry‐based Techniques for Direct Quantification of High Ionization Energy Elements in Solid Materials—challenges and Perspectives

“…Rather high oxygen intensity was achieved reducing the effect of spectral interferences. Several groups intensively used SIMS to investigate oxygen impurities in gallium nitride (Reedy et al, 2004; Michałowski, Złotnik, & Rudziński, 2019; Reshchikov et al, 2020). SIMS was also shown to be applicable for the quantification and isotopic analysis of oxygen (18/16 isotope ratio) in bones with acceptable internal precision of 0.25‰ (Maggiano et al, 2019).…”