DOI: 10.1002/cphc.201101007 A DFT Study of the Structures of Au x Clusters on a CeO 2 (111) Surface Bo-Tao Teng,* [a] Feng-Min Wu, [a] Wei-Xin Huang,* [b] Xiao-Dong Wen, [c] Lei-Hong Zhao, [a] and Meng-Fei Luo [a] 1. Introduction The structure of metal clusters supported on metal oxides strongly affects their catalytic performance. Therefore, it is of great importance to identify the structure of supported metal clusters, but this remains a great challenge. Experimentally, routine methods for structural characterization of ultrafine sup- ported metal clusters are still lacking. Theoretically, the system is very complicated, and many factors must be considered during calculations. For free metal clusters, numerous theoretical works with dif- ferent levels of theory and methods have been performed to identify their optimal structure. Empirical and semi-empirical potentials as well as DFT were used to describe the atomic in- teractions. [1, 2] Molecular dynamics simulations, [3] genetic algo- rithms, [4] basin hopping algorithm, [5] and so on were used to approach the global minimum of the potential-energy surface (PES). The empirical and semi-empirical potential methods that use empirical parameters in the calculation can converge rap- idly, even for large clusters, but with low precision. However, it may be difficult to apply solely the DFT method, which is much more time-consuming, to the global calculation of struc- tures of large metal clusters. Therefore, several mixed methods have been developed, in which the global optimization is car- ried out by means of semi-empirical potential models to screen a series of promising structures, which are then reopti- mized by DFT methods. [6–10] The system of metal clusters supported on oxides (M x /oxide) is much more complicated than that of free metal clusters due to the involvement of substrates. Owing to the lack of reliable empirical potentials for metal and oxygen ions in M x /oxide sys- tems, the application of empirical potential models and optimi- zation algorithms are substantially limited in the calculation of M x /oxide structures. [11] Density functional calculations thus become the main reliable method to theoretically consider the interaction between metal clusters and oxide support. Employ- ing DFT methods, Valero et al. [12] systematically investigated the nucleation and wetting of Pd x (x = 1–5) clusters on g-Al 2 O 3 ; C ¸ elik et al. [13] studied the adsorption structures and electronic properties of Pt x (x = 1–4) clusters on the stoichiometric, re- duced, and reconstructed rutile TiO 2 (110) surfaces; Yang et al. [14] reported the structures of Cu x (x = 1–4) clusters on CeO 2 (111); Zhao et al. [15] calculated Sn x (x = 1–4) on CeO 2 (111) ; Gong et al. [16] systematically compared the structures of Au x and Pt x (x = 1–3) clusters on anatase TiO 2 (101) surfaces; Barcaro and Fortunelli [17] systematically calculated coinage-metal clus- ters on Mg(100); Pabisiak and Kiejna [18] calculated the struc- tures of Au x nanorows (x = 1–7) and clusters (x = 1–12) ad- [a] Prof.Dr. B.-T. Teng, F.-M. Wu, Prof. L.-H. Zhao, M.-F. Luo Institute of Physical Chemistry Zhejiang Normal University Jinhua 321004 (China) E-mail : tbt@zjnu.cn [b] Prof. Dr. W.-X. Huang Hefei National Laboratory for Physical Sciences at the Microscale CAS Key Laboratory of Materials for Energy Conversion and Department of Chemical Physics University of Science and Technology of China Hefei 230026 (China) E-mail : huangwx@ustc.edu.cn [c] Dr. X.-D. Wen Theoretical Division Los Alamos National Laboratory Los Alamos, NM 87545 (USA) Supporting information for this article is available on the WWW under http://dx.doi.org/10.1002/cphc.201101007. Studying the structures of metal clusters on oxide supports is challenging due to their various structural possibilities. In the present work, a simple rule in which the number of Au atoms in different layers of Au x clusters is changed successively is used to systematically investigate the structures of Au x (x = 1– 10) clusters on stoichiometric and partially reduced CeO 2 (111) surface by DFT calculations. The calculations indicate that the adsorption energy of a single Au atom on the surface, the sur- face structure, as well as the Au Au bond strength and ar- rangement play the key roles in determining Au x structures on CeO 2 (111). The most stable Au 2 and Au 3 clusters on CeO 2 (111) are 2D vertical structures, while the most stable structures of Au x clusters (x > 3) are generally 3D structures, except for Au 7 . The 3D structures of large Au x clusters in which the Au number in the bottom layer does not exceed that in the top layer are not stable. The differences between Au x on CeO 2 (111) and Mg(100) were also studied. The stabilizing effect of surface oxygen vacancies on Au x cluster structures depends on the size of Au x cluster and the relative positions of Au x cluster and oxygen vacancy. The present work will be helpful in improving the understanding of metal cluster structures on oxide sup- ports. ChemPhysChem 2012, 13, 1261 – 1271  2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 1261