Computational thermochemistry of iron–platinum carbonyl clusters Michael Bühl a,⇑ , Herbert Früchtl a , Pascal André b a EaStCHEM School of Chemistry, University of St. Andrews, North Haugh, St. Andrews, Fife KY16 9ST, United Kingdom b SUPA, School of Physics and Astronomy, University of St. Andrews, North Haugh, St. Andrews, Fife KY16 9SS, United Kingdom article info Article history: Received 22 March 2011 In final form 2 May 2011 Available online 5 May 2011 abstract Structures of Fe 3 Pt 3 (CO) 15 (1), Fe 2 Pt 5 (CO) 12 (COD) 2 (2), Fe 2 Pt(CO) 8 (COD) (3) and Fe 2 Pt 2 (CO) 10 (4), as well as the driving forces for their formation from Fe(CO) 5 and Pt(COD) 2 (COD = 1,5-cyclooctadiene) have been computed at the PBE0-D3 level of density functional theory. Judged from a comparison of computation vs. experiment, this level should be well suited to study structures and thermochemistry of mixed Fe– Pt clusters, which may occur at the early stages of FePt nanoparticle synthesis. Ó 2011 Elsevier B.V. All rights reserved. 1. Introduction Due to their high magnetic anisotropy, Curie temperature and chemical stability [1], binary iron–platinum nanoparticles are an increasingly popular research topic with promising applications ranging from biomedicine [2–5], to magnetic storage [6–8]. Tai- lored production of nanocrystalline material with low size polydis- persity and well-ordered local structure is highly desirable for subsequent technological applications, with face-centered tetrago- nal FePt being of prime interest to obtain hard magnet nanoparti- cles with a grain diameter apparently as small as 2.8 nm at 300 K [9]. Starting materials which have been explored so far are some- how limited and usually involve organometallic iron(0) species, mostly Fe(CO) 5 , and various Pt(0) complexes or Pt(II) salts follow- ing experimental protocols that rely on reducing agents, or thermal decomposition [1,6,10,11]. Formation and growth of the nanopar- ticles indeed occurs under rather drastic conditions, with reaction temperatures sometimes exceeding 300 °C. While some progress has been made in controlling size and structure of the FePt parti- cles, the chemical processes that lead to their formation are only poorly understood, and only very few numerical studies have been reported, which have focused, for instance, on cluster composition [12]. It is frequently assumed that at the high temperatures, the pre- cursor complexes decompose into bare metal atoms that assemble into nanoparticles [1]. However, the rate of temperature increase is also known to be a critical parameter implying that the formation of complexes and clusters at the very early stage of the syntheses needs to be carefully considered and understood. Because mixed iron–platinum carbonyl complexes display a rather rich chemistry (see Refs. [13–15,41] for some recent examples), it is conceivable that they can be involved in the growth process, at least when car- bonyl complexes are used as metal source. For instance, reaction of Fe(CO) 5 and Pt(COD) 2 (COD = 1,5-cyclooctadiene) at room temper- ature readily affords a mixture of iron–platinum carbonyl com- plexes, of which Fe 3 Pt 3 (CO) 15 (1; 9%), Fe 2 Pt 5 (CO) 12 (COD) 2 (2; 40%), and Fe 2 Pt(CO) 8 (COD) (3; 3%) have been isolated [16]. If these or related clusters are involved in the growth process, e.g. through adsorption on a growing surface and subsequent decarbonylation, their speciation in the reaction mixture could directly affect the structure and rate of formation of the nanoparticles. If this specia- tion could be reliably modeled computationally, with similar suc- cess as, for instance, displayed in quantum dot syntheses [17– 19], rational design of reaction conditions for FePt nanoparticle growth may be envisioned. The modern tools of density functional theory (DFT) can de- scribe structures, properties, and reactivities of most transition- metal compounds very well. Nonetheless, thorough validation is usually required in order to ensure that the flavor of DFT to be ap- plied is suited for a particular target system. It is one purpose of this communication to provide such a validation for the Fe(CO) 5 and Pt(COD) 2 system, i.e. to identify a computational protocol that yields results in agreement with experimental observations. Be- cause of its good performance in many areas of chemistry, the PBE0 hybrid functional has been chosen for geometry optimiza- tions and energy evaluations [20,21], and the latter have been cor- rected for dispersion interactions and enthalpic and entropic contributions (see Section 4). 2. Results and discussion The target species of this study are displayed in Figure 1. To as- sess the quality of the DFT-derived structural parameters, we com- pare the optimized bond distances to experimental data from X- ray crystallography (both included in Figure 1) [16,22]. As is com- monly observed with gradient-corrected and hybrid functionals, bond distances between heavy atoms are generally overestimated. 0009-2614/$ - see front matter Ó 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.cplett.2011.05.002 ⇑ Corresponding author. E-mail address: buehl@st-andrews.ac.uk (M. Bühl). Chemical Physics Letters 509 (2011) 158–161 Contents lists available at ScienceDirect Chemical Physics Letters journal homepage: www.elsevier.com/locate/cplett