We identify the atomic structure of the Au/Si(111)-5×2 surface by using density functional theory calculations. With seven Au atoms per unit cell, our model forms a bona fide 5×2 atomic structure, which is energetically favored over the leading model of Erwin, Barke, and Himpsel [Phys. Rev. B 80, 155409 (2009)] and well reproduces the Y-and V-shaped 5×2 STM images. This surface is metallic with a prominent half-filled band of surface states, mostly localized around the Au-chain area. The correct identification of the atomic and band structure of the clean surface further clarifies the adsorption structure of Si adatoms and the physical origin of the intriguing metal-to-insulator transition driven by Si adatoms.PACS numbers: 68.43. Bc, 73.20.At, 81.07.Vb The Au/Si(111)-5×2 surface is a representative selforganized one-dimensional (1D) metal chain system [1][2][3][4] and has served as a rich source of intriguing 1D phenomena such as atomic-scale Schottky barriers [5], 1D domain-wall hoppings [6], and confined doping on a metallic chain [7]. Its atomic structure, however, is not yet solved in spite of extensive experimental studies [8][9][10][11][12][13][14][15][16][17][18][19] and density functional theory (DFT) calculations [20][21][22][23][24][25]. This long-standing surface science problem is thus considered a touchstone of our ability to look into the structure of complex surface/nano systems at truly atomic scale.In particular, the structural debate was recently reignited by two conflicting reports [26,27]. Figure 1(a) shows the best structural model so far, proposed by a combined theoretical-experimental study of Erwin, Barke, and Himpsel (hereafter, EBH) in 2009 [25]. It features a Si honeycomb chain and an Au triple chain and contains six Au atoms and twelve top-layer Si atoms in a 5×2 unit cell, well reflecting the experimental estimations of 5.6-6.7 Au atoms [17][18][19] and 11-14 Si atoms [28,29]. The EBH model was recently challenged by a new model derived by Abukawa and Nishigaya (hereafter, AN) in a reflection high-energy electron diffraction (RHEED) study [26]. The AN model has the same Au and Si coverages as the EBH model but quite distinct structural features: It has no Si honeycomb chain, and the Au chain contains four Au rows rather than three. More recently, however, the AN model was excluded by Hogan et al.[27]: The AN model was energetically and microscopically unfavored in DFT calculations, and their analysis of previous reflectance anisotropy spectroscopy data [18] strongly supported the presence of Si honeycomb chains, thereby being in favor of the EBH model.One inherent problem with the EBH model is that it is basically a 5×1 structural model as seen in Fig. 1(a) and therefore is incompatible with the observed, distinct 5×2 STM images [11,[14][15][16]. EBH argued by DFT calculations [25] that 5×2 STM images may be possible from the 5×1 structure with the aid of either Si adatoms or electron doping, but the origin of such electron doping is not clarified yet.In this Letter, we report...