A density functional theory benchmark of the formation enthalpy and first CO dissociation enthalpy of hexacarbonyl complexes of chromium, molybdenum, and tungsten Richard Sniatynsky, David L. Ceden ˜o * Department of Chemistry, Illinois State University, P.O. Box 4160, Normal, Il 61790-4160, USA Received 9 September 2004; accepted 21 September 2004 Abstract Reaction energies have been calculated for the hexacarbonyl metal complexes of chromium, molybdenum, and tungsten using six different Density Functional Theory (DFT) methods. Two gas phase reactions have been utilized to benchmark the computed results. The reactions are the heat of formation of the hexacarbonyl complex from its metal dioxide, dicarbon and dioxygen; and the first dissociation enthalpy of the carbonyl ligand. It was found that all DFT methods agreed well with the available experimental data. For the formation reaction, all methods underestimate the formation enthalpy, while non-hybrid methods do better than the methods involving functional hybridization. Including a diffuse function in the basis set decreases the magnitude of the formation enthalpy. For the CO dissociation reaction the LYP methods underestimate the bond enthalpy consistently for all the metals. There was not much difference between using a hybrid and a non-hybrid functionals, except when comparing the PWP91 and B3PW91 methods, but it was found that the level of exchange and correlation affects the result, especially when using the LYP-based methods. A comparison of calculated geometries and frequencies is also made with available experimental and computational data, confirming that the ground state of WO 2 is a singlet. q 2004 Elsevier B.V. All rights reserved. Keywords: Density functional theory; Metal hexacarbonyls; Benchmarking 1. Introduction In the latest years, the use of computational chemistry methods has become common practice in all fronts of chemical research. The availability of both sophisticated methods and efficient hardware has definitively contributed to the increased popularity of computational chemistry. The field of organometallic chemistry has not been shy of this trend and both theoreticians and bench chemists rely on computations to analyze a variety of systems from both mechanistic and thermodynamic fronts. Among the differ- ent methods available, density functional theory (DFT) have gained popularity because of their reliable estimate of molecular geometries, energies, and frequencies (among other properties) at a relatively low computational price [1–6]. The development of efficient and inexpensive ways to obtain accurate results includes the implementation of relativistic effects, frequency scaling, reliable effective core potentials (ECPs) to use with heavy elements, and the development of fast DFT algorithms. In spite of the benefits provided by DFT methods, anyone attempting to use them to study organometallic systems is posed with the following question: what combination of available density functionals would be the most suitable to represent most accurately such a system? Traditionally, the BP86 and B3LYP functionals have been the most common methods of choice. However, it has been shown by previous benchmarking studies [7–9] that, for example, B3LYP does not reproduce properly bond energetics of systems involving iron, and chromium while BP86 tend to overestimate certain metal–ligand bond energies and reaction enthalpies. So far, benchmarking is the best way available to establish which functional might be the most appropriate to use in a specific system, but it relies on the availability of experimental data 0166-1280/$ - see front matter q 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.theochem.2004.09.018 Journal of Molecular Structure (Theochem) 711 (2004) 123–131 www.elsevier.com/locate/theochem * Corresponding author. Tel.: C1 309 438 5595; fax: C1 309 438 5538. E-mail address: dcedeno@ilstu.edu (D.L. Ceden ˜o).