Optical and high-power electronic devices composed of gallium nitride (GaN) have attracted much attention because of its direct wide bandgap, high electron saturation velocity, and high thermal conductivity. [1][2][3] For applications, it is known that unintentionally doped impurities, such as oxygen, into GaN active layers deteriorate device performance. [4,5] Clarifying the impurity incorporation mechanism and controlling its concentration during growth are essential. Considerable research on the reduction of impurities in polar plane growth, such as þc(0001)and Àc(000À1)-plane growth, has been reported due to wide practical uses of the resulting materials. [6][7][8][9] Conversely, there have been few reports on impurity incorporation in the case of nonpolar plane growth, such as m(10À10)-plane growth, although layers grown without a polarization field along the growth direction are advantageous for device performances. [10] In 2018, Tanaka et al. [11] reported that the oxygen concentration in GaN films grown on 5 off m-GaN toward the Àc-direction (Àc 5 off ) by metal-organic vapor phase epitaxy (MOVPE) was below that on 5 off m-GaN toward the þc-direction (þc 5 off ). Understanding the reduction mechanism of the oxygen concentration in vicinal m-GaN layers is important for both science and technology. In 2020, via large-scale density functional theory (DFT) calculations and the theoretical method proposed by Kangawa et al., [12][13][14] our group clarified that the charge distribution near the step edge associated with surface reconstruction influences oxygen substitution at the nitrogen sites near the step edge. [15] In this study, we expand the DFT calculations to the entire range on the terrace and quantitatively analyzed the oxygen concentration in thin films by the newly proposed diffusion model.In a previous study, [15] it was reported that ideal and 3N-H structures appear for the þc 5 off and Àc 5 off step edges, respectively, under the following experimental growth conditions: [11] a H 2 carrier gas, a total pressure of 1 atm, T ¼ 1100 C, and V/III ¼ 1019. In this study, the predicted surface model was used as the starting model, and then, the formation energies of oxygen substituting nitrogen, O N , in the topmost layer were calculated by using real-space density functional theory (RSDFT) as implemented in the RSDFT package. [16][17][18] In this research, we considered only O N, because the formation of the point defect is most often predicted in the first principles calculations. [19] The detailed DFT calculation conditions are described in the literature. [15] The O N formation energy ΔE O N is written aswhere E substitute is the total energy of a system whose nitrogen site at the topmost layer was substituted by an oxygen atom, and E reference is the total energy of a system whose nitrogen site on the terrace was substituted by an oxygen atom. Figure 1a,b shows the side views of þc 5 off and Àc 5 off m-GaN models, respectively. The slab model comprised a vacuum layer of more than 20 Å, five GaN bil...