We discuss how to include our recently proposed thermopotentiostat technique [Phys. Rev. Lett. 126, 136803 (2021)] into any existing ab initio molecular dynamics (AIMD) package. Using thermopotentiostat AIMD simulations in the canonical NVTΦ ensemble at constant electrode potential, we compute the polarization bound charge and dielectric response of interfacial water from first principles.

Applying state-of-the-art first-principles calculations we study atomic geometry, electronic structure, and energetics for all native defects and for several donor impurities (0, C, Si) in GaN. An analysis of these results gives direct insight into the defect concentrations and the solubility of impurities with respect to the growth conditions (temperature, chemical potentials) and into possible mechanisms forpassivation and compensation. Particularly, we discuss in detail the roleofthe nitrogen vacancy, which is commonly assumed to be the source for the "auto-doping" of GaN. Our results show that GaN has distinctively different defect properties compared to more "traditional" semiconductors such as Si, GaAs or ZnSe. This is explained in terms of the large mismatch in the atomic radii of Ga and N.

A review of surface structures of bare and adsorbate-covered GaN (0001) and (000) surfaces is presented, including results for In, Mg, Si, and H adsorbates. Emphasis is given to direct determination of surface structure employing experimental techniques such as scanning tunneling microscopy, electron diffraction, and Auger electron spectroscopy, and utilizing first principles computations of the total energy of various structural models. Different surface stoichiometries are studied experimentally by varying the surface preparation conditions (e.g. Ga-rich compared to N-rich), and the stoichiometry is included in the theory by performing calculations for various chemical potentials of the constituent atoms. Based on the work reviewed here, surface reconstructions for plasma-assisted molecular beam epitaxy growth of GaN (0001) and (000) surfaces are fairly well understood, but reconstructions for reactive molecular beam epitaxy or for metal-organic vapor phase epitaxy (both involving H, at moderate and high temperatures, respectively) are less well understood at present.

No abstract