Deep-Level Traps in AlGaN/GaN- and AlInN/GaN-Based HEMTs With Different Buffer Doping Technologies
P. Vigneshwara Raja, Mohamed Bouslama, Sujan Sarkar, Khade Ramdas Pandurang, Jean‐Christophe Nallatamby, Nandita DasGupta, Amitava DasGupta
Abstract
Deep-level traps in AlGaN/GaNand AlInN/GaN-based HEMTs with different buffer doping technologies are identified by drain current transient spectroscopy (DCTS) and low-frequency (LF) output admittance (Y <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">22</sub> ) dispersion techniques. TCAD simulations are also carried out to determine the spatial location and type of traps. The DCTS and LF Y <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">22</sub> measurements on Al <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">0.25</sub> Ga <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">0.75</sub> N/GaN HEMT (Fe-doped buffer) reveal a single electron trap at EC - 0.47 eV. On the other hand, an electron trap at E <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">C</sub> - (0.53-0.59) eV and a deep hole trap at EV + 0.82 eV are detected in Al <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">0.845</sub> In <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">0.155</sub> N/AlN/GaN HEMT with unintentionally doped (UID) buffer, while a slow detrapping behavior is noticed at E <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">C</sub> - 0.6 eV in Al <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">0.83</sub> In <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">0.17</sub> N/AlN/GaN HEMT with C-doped buffer. The DCTS and LF Y <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">22</sub> measurements yield nearly the same trap signatures, indicating the reliability of the trap characterization techniques used in this article. The simulated LF Y <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">22</sub> characteristics show that all these traps are acceptor-like states located in the buffer layer. The identified trap parameters in various buffers may be helpful to improve the crystalline quality of the epitaxial buffer layers.