Cuprophilicity, a still elusive concept: a theoretical analysis of the ligand-unsupported CuIndash;CuIinteraction in two recently reported complexes

摘要

C1 C2 C3 C4 C5 N N Fe Cu Cl Cl Cu Cuprophilicity, a still elusive concept: a theoretical analysis of the ligand-unsupported CuIndash;CuI interaction in two recently reported complexes Josep-M. Pobleta and Marc B�enard*bdagger; a Departament de Quimica, Universitat Rovira i Virgili, Pc. Imperial Tarraco 1, E-43005-Tarragona, Spain b Laboratoire de Chimie Quantique, UMR 7551, CNRS et Universit�e Louis Pasteur, 4 rue B. Pascal, F-67000 Strasbourg, France Density functional theory (DFT) calculations explain the short ligand-unsupported CuIndash;CuI contact recently reported for the CuL+CuCl22 complex L = 1,1A-bis(2-pyridyl) octamethylferrocene by a strong electrostatic attraction (264 kcal mol21) between the two moieties and rule out the initially suggested metallophilic interaction, but cuprophilicity might account for the dimerization occurring in a family of trimetallic complexes.Weak attraction between transition metal atoms with closedshell electronic configuration was first evidenced by Schmidbaur et al. in the cases of intra- and intermolecular AuIndash;AuI contacts.1 The term aurophilicity was coined to describe these interactions, but it soon became clear that similar aggregation processes could also involve metal atoms other than gold, such as TlI or HgII.2 These closed-shell interactions in inorganic chemistry have been reviewed by Pyykk�o.2 The occurrence of analogous metallophilic3 effects involving lighter metal atoms and more specifically CuI has been the subject of a long debate due to the intramolecular character of the reported interactions. 2,4 Recently, two examples of unsupported CuIndash;CuI contacts with metalndash;metal distances of 2.905 Aring;5 and 2.810 Aring;6 have been tentatively assigned to cuprophilic interactions.On the theoretical side, recent studies by Pyykk�o et al. suggest that the stabilization of the ClCuPH3 model dimer due to the metallophilic interaction between the CuI atoms does not exceed 21.5 kcal mol21 when extrapolated to the best level of theory, and is associated with a rather long metalndash;metal distance of 3.143 Aring;.7 The goal of this study was to investigate by means of DFT and extended H�uckel calculations other possible origins for the unsupported CuIndash;CuI interactions in the two complexes for which cuprophilicity has been addressed.Metallophilic interactions are not easy to characterize from quantum chemical calculations. Pyykk�o et al.2,3 have demonstrated that metallophilicity is due to attractive dispersion forces that should overcome the Pauli repulsion between the d10 or the d10 s2 closed shells.For metals of the third transition row, the attractive forces are greatly enhanced by relativistic effects.2,7 Since neither ab initio Hartreendash;Fock nor DFT calculations account for dispersion-type R26 terms these levels of theory unavoidably predict repulsive behaviour between unsupported metallophilic fragments.2,3 Conversely, if these methods are able to account for an attractive interaction, it should be clear that it is not of the metallophilic type.This is the principle of the investigations performed on CuL+CuCl22 L = 1,1A-bis(2-pyridyl)octamethylferrocene 1 for which a cuprophilic interaction had been tentatively suggested to explain the short CuIndash;CuI contact (2.810 Aring;) observed between the two copper subsystems.6 Complex 1 has been slightly modeled by replacing octamethylferrocene with ferrocene (1A) and by assuming perfect C2v symmetry, which implies that the coordination axes of the anion and the cation are perpendicular (Fig. 1). We then carried out a full geometry optimization of 1A by means of gradient-corrected DFT calculations.Dagger; Selected geometrical parameters obtained from the calculation are reported in the caption of Fig. 1 and compared to experiment. The observed environment of the copper atoms is reproduced by the calculation with great accuracy, including the Cundash;Cu bond length (calc. 2.822 Aring;, exptl. 2.810 Aring;). The interaction energy between the two fragments is calculated to be 264.1 kcal mol21, after BSSE correction. The presence of a bonding interaction at this level of theory and its order of magnitude clearly show that the attraction between the CuL+ and the CuCl22 subunits should not be assigned to cuprophilicity. Since the interaction involves charged moieties, the bonding may instead be due to Coulombic forces. Mulliken population analyses, carried out either from the extended H�uckel (EHT) or from the DFT orbitals, indicate that the charge transfer between the two moieties is negligible (Table 1).The negative charge in the (CuCl2)2 fragment is distributed between the chlorine atoms while the Cu atom remains either neutral (+0.04e, DFT), or significantly positive (+0,24e, EHT). Even though the point charge distributions in the cationic fragment is noticeably different for EHT and DFT (Table 1), the (CuCl2)2/(CuL)+ electrostatic attraction computed from the point charge model are similar (62.0 kcal mol21 with EHT, 267.9 kcal mol21 with DFT) and practically reproduce the fragment interaction energy obtained from DFT calculations. Other models of space partitioning8,9 applied to the DFT wave function, however, predict some charge transfer toward the (CuL)+ moiety and the fragment electrostatic energies computed from those models are scaled accordingly (Table 1).In order to obtain an estimate of the fragment/fragment Coulombic interaction independent of space partitioning, we relied on standard energy decomposition analysis10,11 and computed the total energy starting from the wave functions computed for the (CuCl2)2 and (CuL)+ fragments assumed isolated, but occupying their geometrical positions in the complex.The interaction energy is now 269.4 kcal mol21. The difference with respect to the value of 264.1 kcal mol21 reported above corresponds to the fragment relaxation energy.The fragment interaction energy is made up of: (i) the Pauli Fig. 1 Molecular structure of 1A optimized from gradient corrected DFT calculations (C2v symmetry assumed). Selected bond lengths (Aring;) and angles (deg;): Cundash;Cu 2.822 (2.810); Cundash;Cl 2.119 (2.095); Cundash;N 1.895 (1.925); Cumiddot;middot;middot;Fe 3.74; Fendash;W 1.668 (W centroid of a Cp ring); Nndash;Cundash;Cu 95.0 (94.3); Clndash;Cundash;Cu 88.8 (89.1). Numbers in parentheses are the averaged experimental values. Chem.Commun., 1998 1179repulsion, +38.8 kcal mol21; (ii) the Coulombic attraction, 286.4 kcal mol21, and (iii) the energy associated with electron reorganization in the complex, which is also attractive and reaches 221.8 kcal mol21. This latter term includes the stabilization due to the mutual polarization of the two fragments which is a purely electrostatic effect that can be distinguished from charge transfer and orbital interaction.11 This energy decomposition analysis stresses the importance of the Pauli repulsion which should not be exclusively assigned to the CuIndash; CuI contact, but also to the two short Cumiddot;middot;middot;H1 distances (2.28 Aring;).It also proves, without assuming any space partitioning, the prominent influence of the Coulombic interaction. Siemeling and colleagues6 noted that a complex closely related to 1, Cu(C5H3NMe3-2,4,6)2CuCl2 2,12 does not display a similar Cundash;Cu interaction. The structure of 2 is characterized by the stacking of planar Cu(C5H3NMe3- 2,4,6)2+ fragments separated by (CuCl2)2 moieties perpendicular to the Nndash;Cundash;N axis, but the Cumiddot;middot;middot;Cu distance is now 3.61 Aring;.12 This increase of the interfragment separation may be tentatively assigned to steric crowding induced by the presence of four Me substituents.However, providing a final answer to this problem will require the geometry optimization of models of 2, with and without the Me substituents. The case of Cu3LA32 {LA = 2-3(5)-pyrazolylpyridine} 35 and related dimers of CuI 3 and AgI 3 complexes13 seems more relevant to metallophilic interactions.Preliminary calculations of the extended H�uckel type carried out on these molecules indicate that the Multhe copper atoms is close to zero and confirm that no significant orbital interaction is at work between the two monomers. However, a conclusive argument proving the existence of metallophilic interactions on such large systems is at present impossible to obtain from quantum chemical calculations.It is however of interest to extrapolate from Pyykk�orsquo;s calculations on ClMPH32 7 the order of magnitude of the metallophilic stabilization in 3 and in its silver counterpart. Pyykk�orsquo;s potential energy curves were obtained at the ab initio MP2 level of calculation with very large basis sets.7 They display energy minima at 3.208 Aring; for Au, 3.113 Aring; for Ag and 3.137 Aring; for Cu. The curves are however extremely shallow, especially for copper.The stabilization energy computed at the minimum is 23.07 kcal mol21 but a separation of 4.5 Aring; still provides a favourable interaction which amounts to 21.2 kcal mol21. The crystal structure of 3 displays two short-range Cundash; Cu interactions (2.905 Aring;) between the two monomers, but also six Cumiddot;middot;middot;Cu distances between 4.44 Aring; and 4.75 Aring;.5 Most of the metallophilic stabilization (ca. 60) might then originate in these long distance interactions. However, providing a quantitative estimate for the overall stabilization energy requires caution.A comparison between MP2 calculations and more elaborate methods carried out for XAuPH32 (X = H, Cl) indicates that MP2 overestimates the real stabilization energy by a factor of 2.2,7 Scaling down accordingly the value deduced from Pyykk�orsquo;s potential energy curves provides an overall stabilization of ca. 26 kcal mol21 due to the cuprophilic effect between the two monomers, 60 of which is assigned to the intermediate range Cumiddot;middot;middot;Cu interactions. A similar reasoning applied to the silver equivalent of 3 yields an estimate of ca. 27.5 kcal mol21 for the metallophilic interaction, but in this case the intermediate range interactions account for no more than one third of the global stabilization. All calculations were carried out on workstations purchased with funds provided by the DGICYT of the Government of Spain and by the CIRIT of Generalitat of Catalunya (Grants no. PB95-0639-C02-02 and SGR95-426). We are pleased to thank Dr Pierre Braunstein for stimulating discussions. Notes and References dagger; E-mail: benard@quantix.u-strasbg.fr Dagger; Computation: gradient-corrected DFT calculations on complex 1 have been carried out by means of the ADF program.14 We used the local spin density approximation characterized by the electron gas exchange (Xa with a = 2/3) together with Voskondash;Wilkndash;Nusair15 parametrization for correlation.Beckersquo;s nonlocal corrections to the exchange energy16 and Perdewrsquo;s nonlocal corrections to the correlation energy17 were added.Slater basis sets of triple-z+ polarization quality were used to describe the valence electrons of C, N, O and H. For first-row atoms, a 1s frozen core was described by means of a single Slater function. For copper, the frozen core composed of the 1s to 2sp shells was also modelled by a minimal Slater basis; 3sp electrons were described by double-z Slater functions, 3d and 4s by triple-z functions and 4p by a single orbital.18 1 (a) H.Schmidbaur, W. Graf and G. M�uller, Angew. Chem., Int. Ed. Engl., 1988, 27, 417; (b) H. Schmidbaur, Gold Bull., 1990, 23, 11. 2 P. Pyykk�o, Chem. Rev., 1997, 97, 597. 3 P. Pyykk�o, J. Li and N. Runeberg, Chem. Phys. Lett., 1994, 218, 133. 4 See, for example: S. P. Abraham, A. G. Samuelson and J. Chandrasekhar, Inorg. Chem., 1993, 32, 6107 and references therein. 5 K. Singh, J. R. Long and P. Stavropoulos, J. Am. Chem. Soc., 1997, 119, 2942. 6 U. Siemeling, U. Vorfeld, B. Neumann and H.-G. Stammler, Chem. Commun., 1997, 1723. 7 P. Pyykk�o, N. Runeberg and F. Mendizabal, Chem. Eur. J., 1997, 3, 1451. 8 F. L. Hirshfeld, Theor. Chim. Acta, 1977, 44, 129. 9 G. Vorono�amp;yacute;, Journal f�ur die reine und angewandte Mathematik, 1908, 134, 198. 10 K. Kitaura and K. Morokuma, Int. J. Quantum Chem., 1976, 10, 325. 11 P. S. Bagus, K. Herrmann and C. W. Bauschlicher, Jr., J. Chem. Phys., 1984, 80, 4378; 1984, 81, 1966. 12 P.C. Healy, J. D. Kildea, B. W. Skelton and A. H. White, Aust. J. Chem., 1989, 42, 115. 13 (a) M. K. Ehlert, S. J. Rettig, A. Storr, R. C. Thompson and J. Trotter, Can. J. Chem., 1990, 68, 1444; 1992, 70, 2161; (b) N. Masciocchi, M. Moret, P. Cairati, A. Sironi, G. Ardizzoia and G. L. Monica, J. Am. Chem. Soc., 1994, 116, 7668. 14 (a) ADF 2.3 Userrsquo;s Guide, Chemistry Department, Vrije Universiteit, Amsterdam, The Netherlands, 1997; (b) E. J. Baerends, D. E. Ellis and P. Ros, Chem.Phys., 1973, 2, 41; (c) G. te Velde and E. J. Barrends, J. Comput. Phys., 1992, 99, 84. 15 S. H. Vosko, L. Wilk and M. Nusair, Can. J. Phys., 1980, 58, 1200. 16 (a) A. D. Becke, J. Chem. Phys., 1986, 84, 4524; (b) A. D. Becke, Phys. Rev. A, 1988, 38, 3098. 17 J. P. Perdew, Phys. Rev. B, 1986, 33, 8882; 1986, 34, 7406. 18 J. G. Snijders, E. J. Baerends and P. Vernooijs, At. Nucl. Data Tables, 1982, 26, 483. Received in Basel, Switzerland, 24th February 1998; 8/01560K Table 1 Point charge distributions (electrons) computed for 1A using (i) the Mulliken space partitioning applied to the EHT and to the DFT wave functions, and (ii) the Hirshfeld8 and the Voronoi9 space partitionings, both applied to the DFT wave function. The electrostatic interaction energy between the two fragments is calculated from those point charges Mulliken Mulliken Hirshfeld Voronoi Atoms (EHT) (DFT) (DFT) (DFT) Fragment (CuL)+: Cu +0.032 +0.482 +0.247 +0.240 N 20.336 20.461 20.086 20.100 Cl + H1 +0.333 +0.352 +0.058 +0.148 C2 + H2 20.037 +0.005 +0.034 +0.175 C3 + H3 +0.115 +0.034 +0.044 +0.052 C4 + H4 20.038 +0.019 +0.023 20.091 C5 +0.433 +0.178 +0.076 +0.020 Fe 20.243 20.012 +0.048 20.036 Cp +0.134 +0.119 +0.027 +0.060 Total (CuL)+ +0.997 +0.962 +0.647 +0.732 Fragment (CuCl2)2: Cu +0.239 +0.037 +0.111 +0.088 Cl 20.618 20.499 20.379 20.410 Total (CuCl2)2 20.997 20.961 20.647 20.732 Electrostatic interaction energy/kcal mol21 262.0 267.9 228.9 241.6 1180 C