Prediction of distinct surface segregation effects due to coordination-dependent bond-energy variations in alloy nanoclusters Leonid Rubinovich and Micha Polak* Department of Chemistry, Ben-Gurion University, Beer-Sheva 84105, Israel Received 30 March 2009; revised manuscript received 27 May 2009; published 7 July 2009 The first implementation of a recently introduced method based on the extraction of the coordination dependence of surface bond-energy variations CBEVfrom density-functional theory DFTcomputed pure- metal surface-energy anisotropy is reported. In particular, polynomial functions fitted to DFT data computed previously for Pt, Pd, and Rh are used as input energetics for statistical-mechanical computations of Pt-Pd 923-atom cuboctahedron-cluster compositional structures and Pt-Rh111as a test caseusing the free-energy concentration expansion method FCEM. The major findings concern the roles of preferential strengthening of intrasurface and surface-subsurface interlayer bonds leading to quite unique segregation characteristics: i strong Pt segregation at certain 111surface sites of the Pt-Pd clusters, accompanied, at relatively high overall Pt composition, by weaker Pt segregation forming Pt-Pd ordered 100structure, whereas Pd segregates mainly at the edge and vertex sites; iidominant Pd subsurface segregation. The high computation efficiency of the CBEV/FCEM approach allows the determination of the complete temperature dependence of atomic-exchange processes among surface sites, as well as between subsurface and deeper sites, reflected in the corresponding configurational heat-capacity curves. Compared to other approaches, the high transparency of this method helps to elucidate the origin of the distinct bond-energy-variation effects on site-specific segregation in alloy nanoclusters. DOI: 10.1103/PhysRevB.80.045404 PACS numbers: 61.46.Df, 68.35.Dv, 82.60.Qr, 68.35.Md I. INTRODUCTION Considerable efforts have been made to model composi- tional and geometrical structures specific to alloy nanoclus- ters compared to the corresponding bulk alloys, with empha- sis on the outer surface shell. 1 These theoretical- computational studies are motivated by the relevance of chemical and physical properties, which depend on the clus- ter structure and size, for applications in catalysis, magnetic media, or electronic devices. Pt-Pd and Pt-Rh modeled in this work are widely used as catalysts in the detoxication of car exhaust gases. Limitations of current experimental char- acterization techniques make modeling almost the only tool to obtain atomic-scale geometrical and compositional infor- mation concerning nanosized alloy clusters. Several ap- proaches that differ in the level of accuracy and efficiency have been reported in the literature. 1 In particular, as far as energetics are concerned, the currently most common meth- ods are based either on the density-functional theory DFT or on empirical or semiempiricalmany-body potentials EPsuch as the second moment approximation to the tight- binding theory TB-SMA, or preferably, on a combination of both. 2 The former can be of high accuracy but the rather time consuming computations limit their practical applicabil- ity to relatively small size alloy clusters. For the study of medium or large clusters the generally less accurate but computationally much more efficientEP methods have been often used. For the exploration of cluster properties at finite temperatures, the interatomic energetics has to be combined with thermodynamic/statistical-mechanical formalism or computer simulations. Employing statistical-mechanical ex- plicit expressions for the alloy free energy has certain advan- tages. In particular, the free-energy concentration expansion method FCEM, which takes into account short-range order SRORef. 3in a system of equilibrated atom-exchanging clusters, has proven to be highly efficient in computation of site-specific concentrations obtainable by F minimization vs overall composition, temperature and cluster size, 4 as compared to Monte Carlo simulations, for example. Our first attempts to go beyond simple empirical energet- ics as input in the FCEM for the study of alloy nanoclusters incorporated 4,5 elemental bond energies and their surface- induced variations as obtained by means of the Naval Re- search Laboratory-tight-binding method NRL-TB. 6 How- ever, it appears that quantitative evaluation of the role of elemental bond-energy variations in surface segregation in bulk alloys and alloy clusters necessitates more reliable en- ergetics. Therefore, we introduced most recently another ap- proach to the issue of near-surface interatomic bonding. 7 Our aim was to develop a DFT-based method of high efficiency, namely, to retain a reasonable accuracy in relatively fast computations of medium- and large-cluster compositional structures. The idea was to derive a procedure for extracting the coordination dependence of intrasurface and surface- subsurface pair-bond-energy variations CBEVfrom DFT- computed surface-energy anisotropy. The paper is organized as follows. After describing briefly the introduced method, the results computed for Pt-Pd 923- atom clusters are divided between surface and subsurface segregation effects. Pd clusters of this size were predicted to have the cuboctahedron Ohshape. 8 Computational vs ex- perimental data for segregation in Pt 0.25 Pd 0.75 111are given as a test case. The discussion deals with the origins of the two respective driving forces in the context of bond-energy variations involving the two outer shells and assesses previ- ous theoretical and experimental studies of the Pt-Pd system. A summary of the research results and prospects concludes the article. PHYSICAL REVIEW B 80, 045404 2009 1098-0121/2009/804/0454048©2009 The American Physical Society 045404-1