Observation of Interfacial Strain Relaxation in High Indium, AlInN/GaN Heterostructures by Transmission Electron Microscope

Abstract

Abstract Body: AlInN has gathered attention as a candidate material for power electronics and optoelectronics. Historically, there were many challenges in growing AlInN by traditional MBE or MOVCD due to conflicting issues of a) phase separation for high temperature growth, and b) low adatom surface mobility for low temperature growth. Metal Modulated Epitaxy (MME) has proven useful for this challenging material by improving adatom kinetics at low temperature and was shown to improve X-Ray diffraction figures of merit by as much as 11X1. However, the source of the film improvement has been unknown up to now. TEM is a useful tool to investigate defect structures but can itself cause damage and induce phase separation if not performed under proper conditions. Of the few TEM reports of AlInN, most focus attention to AlInN around 18% Al and electron-beam induced degradation on AlInN is not investigated. Few reports exist for imaging high In AlInN. The present work focuses on the investigation of MME grown high In AlInN/GaN heterojunctions, explanation of MME’s improved quality compared to prior literature and its potential false defect observation because of damage induced by the electron beam in the TEM. Unintentionally doped (UID) MME AlInN films were grown on MME GaN on hydride vapor phase epitaxy (HVPE) GaN templates on c-plane sapphire substrates as described elsewhere1, using plasma-assisted molecular-beam-epitaxy (PA-MBE) with a target composition of 70% indium. Strain relaxation at a heterojunction of AlInN/GaN in this epitaxial film is investigated by TEM and Scanning TEM (STEM). The cross-section sample is prepared by standard mechanical polishing methods, followed by argon ion milling. Lattice-expansion mapping (or strain mapping) obtained by STEM nano-beam diffractions showed an intermediate lattice constant between AlInN and GaN. This implies that there is a thin intermediate layer where gradual strain relaxation happens from GaN towards AlInN. Moire fringes are detected in the TEM with two-beam condition of g = [1 -1 0 0], indicating full strain relaxation beyond ~10 nanometers via the generation of misfit dislocations at the interface. Moire fringes in MME grown InGaN/GaN resulted from strain exceeding the critical thickness in the first monolayer2,3. Thus, like MME grown InGaN which also has high adatom mobility at low temperatures, high In AlInN appears to relax via the same misfit dislocation mechanism resulting in higher film quality above the misfit array. In order to validate the results, it was also necessary to ensure the electron beam did not damage the films under examination. It was found that a low beam dose of 5.7 A/cm2 was necessary to prevent damage in thicker regions of the AlInN but still appreciably damaged thin sections of the AlInN even for less than 20 seconds exposure. This beam damage threshold is far lower than the ~35 A/cm2 work on InGaN/GaN4 suggesting high In AlInN is more sensitive to beam damage than InGaN. In-situ TEM measurements on g = [0 0 0 2] detected decomposition and In clustering in/on the AlInN layer. The beam-induced damage was not observable in the low-resolution strain map obtained by STEM micro-beam diffraction in comparably thicker regions. This extremely low beam damage threshold complicates TEM analysis of high In AlInN and results in less clear images and quicker analysis time requirements. Further work quantifying the degradation from the electron beam as well as varied compositional studies are ongoing. 1Engel, et. al., J. Appl. Phys., 127, 125301, 2020 2Fischer, et. al., Appl. Phys. Lett., 103, 131101, 2013 3Clinton, et. al,. Solid-State Electronics, vol.136, pp. 3, 2017 4Smeeton, et. al., Appl. Phys. Lett., 83, 5419, 2003