Phonon dispersions in AlGaN with varying Al content
Title: Phonon dispersions in AlGaN with varying Al content
SNIC Project: SNIC 2021/22-989
Project Type: SNIC Small Compute
Principal Investigator: Vanya Darakchieva <vanya.darakchieva@liu.se>
Affiliation: Linköpings universitet
Duration: 2021-12-15 – 2022-05-01
Classification: 10304
Keywords:

Abstract

The goal pf this project is to calculate phonon dispersions in AlGaN with different Al content. These are needed for the non-Debye relaxation time approximation model of thermal conductivity we developed and which will be used to model our experimental data [1]. As an outcome we expect to create a predictive calculation model of thermal conductivity for real device structures. GaN, AlN and AlGaN have large bandgaps and high figure-of merit (FOM), which makes them suitable for high-power (HP) and high-frequency (HF) electronic device applications. In such devices Joule-heating becomes a critical issue as it degrades their performance. Therefore, thermal management and optimization are critically needed for the device design. The semiconductor layers used in the devices commonly contain defects such as dislocations and unintentional impurities. Consequently, determination of thermal conductivity of such layers by first-principle or molecular dynamics calculations is very challenging, because these methods are currently limited to idealized pure materials systems. Analytical model such as relaxation time approximation (RTA) turns to be a flexible tool to assess the information of thermal conductivity for not only pure but also defective semiconductor layers [2]. Because the heat is mainly carried by phonons in semiconductors, phonon dispersion is the most critical as well as basic parameter needed for the calculation. The use of Debye approximation is the main limitation of the model because it overestimates phonon frequency and group velocity, rising a concern about the accuracy of the calculation. The use of mode Gruneisen parameters as fitting variables is another drawback making this model unable to predict thermal conductivity of materials. We have successfully derived a non-Debye RTA model for thermal conductivity calculation which excludes the Debye approximation. In this model, exact phonon dispersion and other phonon properties such as group velocity and mode Gruneisen parameters are needed as input parameters. With the use of exact phonon parameters this method could show its capability of predicting thermal conductivity of pure semiconductor materials with high accuracy. The possibility of implementation of various scattering rates in the model such as phonon-boundary, phonon-point defects, phonon-dislocation, phonon-alloy, and so forth allows a predictive calculation of thermal conductivity of real device structures. To achieve accurate phonon properties of the (Al)GaN semiconductors, the phonon dispersions of these materials need to be calculated by ab initio using supercell approach and displacement method. The atomic forces of different atomic displacements of supercells will be calculated using density-functional-perturbation theories (DFPT) implemented in Quantum Espresso (QE). The second harmonic interaction force constants (IFCs) are then produced from the obtained forces and the phonon dispersions are calculated using PHONOPY. From the results on the phonon properties calculated in this project we will establish a predictive model of thermal conductivity calculation which is going to be used for material development and devices. References [1] Dat Q. Tran, et al., Appl. Phys. Lett. 117. 252102 (2020) [2] J. Callaway, Phys. Rev. 113, 1046 (1959)