An Atomistically Informed Device Engineering (AIDE) Method Realized: A case study in GaAs

TL;DR

The paper introduces the AIDE method that combines first-principles defect data with experimental parameters to simulate defect chemistry in semiconductors, demonstrated through a GaAs case study.

Contribution

It presents a novel, generalizable AIDE approach that integrates atomistic defect properties into device modeling, enabling dynamic simulation of defect chemistry in semiconductors.

Findings

Fermi level dynamics and equilibrium in GaAs were successfully modeled.

Identification of arsenic vacancy charge states consistent with experimental data.

Demonstration of Coulomb attraction leading to defect formation in GaAs.

Abstract

Radiation-induced defects can have a significant impact on the longevity and performance of semiconductor devices. We present an Atomistically Informed Device Engineering (AIDE) method that integrates first-principles defect properties and experimentally measured parameters into a device model to dynamically simulate the defect chemistry in semiconductors. For a silicon-doped gallium arsenide (GaAs) material, we showcase three capabilities: (i) Fermi level $E_F$ movement including its component electron and hole Fermi levels, (ii) dynamical charge equilibration with the arsenic vacancy serving as an example, and a (iii) diffusion-driven reaction between Coulomb attracted gallium interstitial ($Ga_i$) and arsenic vacancy ($v_{As}$). Governed by charge carrier reactions, the electron and hole Fermi levels remained dissimilar until equilibrium was achieved at $E_F\approx1.32$ eV. The equilibrium Fermi level was verified by successfully identifying $v_{As}^{3-}$ as the most populated charge state within the arsenic vacancy defect. Lastly, a Coulomb attraction, created by the shifted Fermi level and the charge equilibration process, between $Ga_i^{1+}$ and $v_{As}^{3-}$ resulted in the formation of a doubly negative gallium antisite ($Ga_{As}^{2-}$). The AIDE method can access experimentally inaccessible short-time and low-concentration regimes, is generalizable to other more complex systems (e.g., indium gallium arsenide), and, after solving open problems in GaAs, will serve as a virtual experiment to bound estimates for difficult-to-measure physical quantities.