DFT calculations of NMR JC–H coupling constants: An additional tool to characterize the α-agostic interaction in high oxidation state M-alkylidene complexes (M = Re, Mo and Ta)
Graphical abstract
DFT calculations have been used to calculate the spin–spin JC–H coupling constants for the C–H alkylidene bond in Re(CR)(CHR)(X)(Y), Mo(NR′)(CHR)(X)(Y) and , where the experiment suggests a weak agostic interaction for Re and Mo and a strong agostic interaction for Ta. The calculated JC–H are in good agreement with the experiment when the IGLO basis sets are used. The calculations reproduce that JC–H is lower for the syn than for the anti isomer. They also show that JC–H depends on the metal even in the absence of agostic interaction.
Introduction
High oxidation state (d0) transition metal-alkylidene complexes were discovered 30 years ago, and since this time, a large variety of them have been synthesized, in part because they are potential olefin metathesis catalysts [1], [2], [3], [4], [5], [6], [7]. In several of these complexes, the JC–H NMR spin–spin coupling constant of the α-alkylidene C–H bond is low. This has been attributed to a weakening of the C–H bond due to the presence of an α-C–H agostic interaction, i.e., an intramolecular interaction between the C–H bond, acting as a Lewis base, and the electron deficient metal center acting as a Lewis acid [8], [9], [10], [11].
The NMR spin–spin coupling constant is a key tool to characterize the C–H⋯M agostic interactions because the JC–H varies over a rather large range of values whereas the C–H bond length varies little. The JC–H value of a non-agostic C(sp2)–H bond is around 160 Hz [12] whereas values as low as about 75 Hz have been observed in some complexes (see later). In contrast, the C–H bond is lengthened by 0.05 Å at most [13]. Furthermore, the C–H bond length cannot be determined experimentally with high accuracy by other techniques than the expensive and difficult neutron diffraction method. Vibrational stretching frequencies of the agostic C–H bond may be hard to identify because the band can be weak and broad [13]; computational studies are therefore very useful. The geometrical optimizations give C–H values, which usually compare very well with experiment in the few complexes whose structure has been studied by neutron diffraction [10]. Associated with the lengthening of the C–H bond, the stretching frequency is lower than that of a non-agostic C–H bond. However, direct comparison between the calculations and the experiment requires the calculations of JC–H.
In general, calculations of NMR chemical shifts and spin–spin coupling constants require considerable computational effort [14], [15], [16], [17], [18]. However computations of NMR parameters for transition metal complexes have been made possible because of the successful development of the DFT methods as described in the recent review by Autschbach [17]. The key point is that calculations using the GIAO [19], [20] or IGLO [21], [22] methods, along with extended basis sets, give NMR chemical shifts and spin–spin coupling constants that can be reasonably compared with experiment. In the case of the spin–spin coupling constant, the calculated values differ from the experimental ones by 10–20% [17], [23]. Because of the computational effort involved, spin–spin C–H coupling constants have so far only been calculated for various relatively small organic molecules [23], [24], [25], [26] and few metal complexes [27], [28], [29]. In the case of transition metal complexes, NMR spin–spin coupling constant calculations have been focused on ligand–ligand, metal–ligand and metal–metal coupling constants [30], [31], [32], [33], [34], [35], [36], [37].
It has been shown that the spin–spin coupling constant is influenced both by the solvent and by vibrational effects [23], [38], [39] and that the solvent effect is especially important in the case of electron deficient complexes [33], [34], [40]. In the case of transition metal complexes, the solvent effect has been shown to be due in large part to the coordination of molecule of solvent, which changes the coordination number of the metal center. Anharmonicity and vibrational effects have been shown to be most important in the case of two atoms with distance that are far from usual distance. For instance these effects will be more important in the case of compressed dihydride and stretched dihydride complexes than in the case of regular dihydrogen or dihydride complexes [30], [37]. In such cases, a temperature effect is expected [41].
These limitations being known, it should be possible to calculate JC–H for agostic C–H bonds. An agostic interaction requires an electron deficient metal but is usually observed in weakly or non-coordinating solvent. The C–H bond is only very slightly elongated by the agostic interaction so that the vibrational effect should be negligible since temperature effects have never been reported. We can thus study if the calculations reproduce, with reasonable computational effort (DFT with electronic core potential on heavy atoms and limited basis sets), the changes in the JC–H values in systems where it should be acceptable to neglect the temperature and solvent effects. To better test the validity of the computational method, we have studied two sets of complexes. One set is the pseudotetrahedral series, Re(CtBu)(CHtBu)(X)(Y) (X = Y = CH2tBu; X = Y = OtBu; X = CH2tBu, Y = OSiPh3) [42], [43], [44], [45], [46] and Mo(NR)(CHtBu)(X)(Y) (R = 2,6-iPr2-(C6H3) or CPh3; X = Y = CH2tBu; X = Y = OtBu) [47], [48], [49], [50], [51] where the agostic interaction is weak and has been observed only in one of the two isomers. The second set is the pseudooctahedral cyclopentadienyl tantalum alkylidene series, Ta(C5R5)(CHtBu)(X)(Y) (R = H (Cp) or CH3(Cp*), X = Y = Cl; X = Y = CH2tBu) where the alkylidene C–H α-agostic interaction is strong [52]. The alkylidene group has a relatively high barrier for rotation in all these complexes [1], [27], [44], [48], [53], [54], which avoids averaging problems. The following pattern emerges from the experimental results. Changing the metal and the coordination number changes JC–H significantly but changes of only a few hertz result from a substitution of ligands in a given complex. Experimentalists consider, nevertheless, these small changes as meaningful and it is therefore of interest to establish if the computations are able to reproduce the detailed changes in JC–H as well as the absolute values in a quantitative way.
The GIAO method with basis sets including functions with large exponents for accurate representation of the core electrons such as the IGLO-II and IGLO-III basis sets, elaborated by Kutzelnigg et al. [55] give very good results for NMR properties including spin–spin coupling constants. Good results have been also obtained with aug-cc-pXVZ basis sets after addition of supplementary core functions [14], [23]. However, the vast number of calculations of transition metal complexes have been carried out with 6-31G(d,p) Pople type basis set on the organic ligand [56]. We will thus compare the results obtained with the Pople basis set and the IGLO basis set in the case the rhenium complexes. Only the IGLO basis sets will be used for the other systems.
Section snippets
Computational details
In one set of calculations, the experimental rhenium Re(CtBu)(CHtBu)(X)(Y) (X = Y = CH2tBu, Re-1; X = CH2tBu, Y = OSiPh3, Re-2; X =Y = OtBu, Re-3) complexes, the molybdenum Mo(NR)(CHtBu)(X)(Y) (R = 2,6-iPr2-(C6H3), X = Y = CH2tBu, Mo-1; R = CPh3, X = Y = CH2tBu, Mo-2; R = 2,6-iPr2-(C6H3),X = Y = OtBu, Mo-3) complexes and the tantalum Ta(C5R5)(CHtBu)(X)(Y) (R = CH3, X = Y = CH2tBu, Ta-1; R = CH3, X = Y = Cl, Ta-2 and R = H, X = Y = CH2tBu, Ta-3; R = H, X = Y = Cl, Ta-4) complexes have been represented by small models, constructed by substitution
The Re(CtBu)(CHtBu)(X)(Y) complexes: methodology and calibration
The geometries of Re complexes, presented in Fig. 1, have been discussed at length in our previous contribution [27]. Thus, only key points will be mentioned here when needed. The overall geometries of the complexes for the small models are in good agreement with the available experimental data. The presence of the full ligands introduced only minor changes in the structures, the main difference residing in a slight change in the difference in the energy between the syn and anti isomers, the
Conclusion and perspective
This study shows that NMR JC–H coupling constants can be calculated accurately without excessive computational effort. B3PW91 NMR calculations with IGLO-II basis sets for main group atoms on geometries optimized with the same functional and 6-31G(d,p) basis sets is accurate enough to reproduce the experimental JC–H values associated with either weak or strong α-C–H agostic interactions in d0 metal–alkylidene complexes (Re, Mo and Ta). It appears that the computed values with the full systems
Acknowledgements
It is my great pleasure to thank Malcolm Chisholm, an old friend, who initially attracted me to Indiana University as visiting professor. He has always been an enthusiastic supporter of the dialogue between experimental and computational chemists. The present work is far removed from the simple MO pictures that we used to draw over a cup of tea in his office while looking out at the autumn foliage turning red during an Indian summer.
X.S.M. thanks the CNRS for a post-doctoral fellowship. We are
References (76)
- et al.
J. Organomet. Chem.
(1983) - et al.
Prog. Nucl. Magn. Res. Spectrosc.
(2002) - et al.
Chem. Phys. Lett.
(2003) - et al.
J. Organomet. Chem.
(1993) Chem. Rev.
(2002)J. Chem. Soc., Dalton Trans.
(2001)New J. Chem.
(2005)Angew. Chem. Int. Ed.
(2000)Top. Organomet. Chem.
(1998)Chem. Rev.
(2000)
New J. Chem.
Prog. Inorg. Chem.
Angew. Chem. Int. Ed.
The Organometallic Chemistry of the Transition Metals
Metal Dihydrogen and σ-Bond Complexes: Structure, Theory and Reactivity
Chem. Rev.
Chem. Rev.
Struct. Bond.
J. Chem. Phys.
J. Chem. Phys.
J. Chem. Phys.
J. Am. Chem. Soc.
Magn. Reson. Chem.
J. Chem. Theory Comput.
J. Am. Chem. Soc.
Organometallics
Chem. Commun.
Dalton Trans.
J. Am. Chem. Soc.
J. Am. Chem. Soc.
Organometallics
J. Am. Chem. Soc.
J. Am. Chem. Soc.
J. Am. Chem. Soc.
Magn. Reson. Chem.
J. Chem. Theory Comput.
Cited by (35)
Group 6 High Oxidation State Alkylidene and Alkylidyne Complexes
2022, Comprehensive Organometallic Chemistry IV: Volume 1-15Schrock vs Fischer carbenes: A quantum chemical perspective
2019, Advances in Inorganic ChemistryCitation Excerpt :Interestingly, the electronic features of the carbene ligand became notably different compared to previously observed Fischer type carbenes depending on the other ligands, and molybdenum and tungsten carbene complexes that are now thought of being Schrock type MC bond have been developed extensively, especially for olefin metathesis catalysis. Molybdenum or tungsten imido alkylidene complexes of the type M(CHR)(NR′)(OR″)2 (M = Mo, W) are generally prepared via the sequence of reactions shown in Scheme 5B.65 Because of the intense interest in the olefin metathesis reaction for constructing carbon–carbon bonds, theoretical studies of Mo- and W-carbene catalysts have been carried out as early as 1992 and have continued to attract much attention.66–69 The key results are summarized below in greater detail.
The missing agostomer in the fluxionality of cyclohexenylmanganese tricarbonyl
2018, Journal of Organometallic ChemistryCitation Excerpt :The lower relative energy of TS-1-2 (11.4 kcal mol−1, Fig. 3) compared to TS-2 (12.9 kcal mol−1, Fig. 2) and TS-3 (13.8 kcal mol−1, Fig. 2) would mean that TS-1-2 contributes to the observed spectra and equivalence processes. The most widely accepted criteria for an agostic bond include [6,53–55]: (1) M-H-C 3-center 2-electron interaction, (2) short M − H distances (1.8–2.3 Å), (3) small M-H-C bond angle (90–140°), (4) up field-shift of the agostic hydrogen, and (5) low spin-spin coupling JCH (50–100 Hz). Based on these criteria, the agostic parameters of structure 1 and structure 2 may be compared (Table 1).
Transition Metal Alkane-Sigma Complexes: Synthesis, Characterization, and Reactivity
2016, Advances in Organometallic ChemistryCitation Excerpt :Structures derived from single-crystal diffraction studies are obviously very useful in this regard with regard to both M⋯E distance and M–E–H bond angle. Solution NMR data can also provide more accessible data, via a significant changes in δ(E/H) chemical shift on coordination, or reduction in E–H coupling constants as the E–H orbitals become coopted into interaction with the metal.33,36,64 Ultimately detailed computational studies using techniques such as quantum theory of atoms in molecules (QTAIM)65 and Natural Bond Order (NBO) analyses may be required to unpick the precise nature of the bonding.66–73