Band Alignment in Quantum Wells from Automatically Tuned DFT+ U
I am happy to update with the news that our article has been published in Physical Chemistry Chemical Physics by the Royal Society of Chemistry, see https://pubs.rsc.org/en/content/articlelanding/2019/cp/c9cp00122k/
Congratulations to my co-authors from Mitsubishi Electric: Grigory Kolesov, Chungwei Lin, Keisuke Kojima, Joseph Katz, Koichi Akiyama, Eiji Nakai, and Hiroyuki Kawahara! Great work, finished well!
Quantum wells are formed in semiconductors by having a material, like gallium arsenide, sandwiched between two layers of a material with a wider bandgap, like aluminium arsenide. (Other example: layer of indium gallium nitride sandwiched between two layers of gallium nitride.) These structures can be grown by molecular beam epitaxy or chemical vapor deposition with control of the layer thickness down to monolayers. Because of their quasi-two dimensional nature, electrons in quantum wells have a density of states as a function of energy that has distinct steps, versus a smooth square root dependence that is found in bulk materials. Additionally, the effective mass of holes in the valence band is changed to more closely match that of electrons in the valence band. These two factors, together with the reduced amount of active material in quantum wells, leads to better performance in optical devices such as laser diodes. As a result quantum wells are in wide use in diode lasers, including red lasers for DVDs and laser pointers, infra-red lasers in fiber optic transmitters, or in blue lasers. They are also used to make HEMTs (High Electron Mobility Transistors), which are used in low-noise electronics. Quantum well infrared photodetectors are also based on quantum wells, and are used for infrared imaging.
My collaborators and I have posted a new report. We demonstrate that Density functional theory (DFT) calculations using DFT+U can be an efficient way to determine the band alignments between two alloys.
The full procedure can be divided into two steps. The first step is to determine U values of a bulk alloy by automatically optimizing atomic orbital-specific values of U so that the experimental bandgap and the lattice constant agree with the values obtained in the simulation. The second step is to use these fitted U values in a superlattice calculation (with lattice relaxation), and the valence and conduction band offsets (VBOs and CBOs) are then determined from the projected Density of states (DOS) away from the interface.
We apply this procedure to InGaAs/InP, InGaAs/InAlAs, and InAlAs/InP, and are able to obtain both VBOs and CBOs consistent with experiments. In addition the computed quantum-well width-dependent bandgaps of InGaAs/InP are in excellent agreement with the photoluminescent measurements.
The proposed method is semi-empirical, because optimization of U values requires knowledge of experimental bandgaps and lattice constants. However, it provides meaningful valence and conduction band offsets between two alloys, with the interface strain taken into account.
The use of a compact numerical atomic orbital basis sets as implemented in SIESTA package makes this method quite lightweight, amenable to large (200+ atoms) supercell computation on a single workstation. Because lattice relaxation is taken into account, the proposed procedure can serve as a practical method to explore the band alignments between complicated alloys.