Typically, organometallic catalyst selectivity and reactivity have been controlled by variations in the metal and ancilliary ligands used, focusing on steric and electronic properties of the latter.^1,2 In contrast, nature's metalloenzymes use secondary interactions such as hydrogen bonding or proton transfer in the active site.³ An increasing number of synthetic catalysts and related systems show the benefits of secondary interactions,⁴ which are generally difficult to control. Using hydrogen bonding or proton transfer from or to imidazolyl⁵ and pyridylphosphines,^4d,6,7 our group has reported rate enhancements in anti-Markovnikov alkyne hydration and alkene isomerization by factors of up to 10000. In studying the origin of remarkable rate enhancements in alkyne hydration, we have discovered hydrogen bonding between the terminal alkyne proton and a heterocyclic nitrogen. To probe this unusual structural feature, we report the first use of scalar coupling across a hydrocarbon C-H bond (^2hJ_CN), a technique used to characterize hydrogen bonds in biological samples.⁸

Initial DFT calculations on the alkyne hydration pathway first suggested the importance of structures like 5 (Scheme 1).⁹ However, in 5 the orientation of the alkyne alone does not allow us to conclude that there is C-H-N hydrogen bonding: in 5c, a complex lacking the nitrogens, the same alkyne orientation is seen in both an X-ray crystal structure (with BF₄^- counterion)¹⁰ and our calculations. It appeared from the literature¹⁰ that to make an alkyne complex like 5 and study its structure before it isomerized to a vinylidene complex, we would need the smallest alkyne, phosphine, and ancilliary Cp ligand possible, making 5a,b with ligands 1a and 1b our synthetic targets, with 1c as a control.

Scheme 1. Precursor Synthesis

Thus, ligands 1a,¹¹ 1b,⁹ and 1c were each converted to 2 using CpRu(COD)Cl. In the case of 2c, ionization of the chloride (Scheme 1) was done in the presence of HCCH to give 5c, which could be isolated and stored cold, but it converted smoothly to 6c after 3 h at 50 C (Scheme 2).

Scheme 2. Alkyne Hydrogen Bonding and Enhanced Reactivity of Bifunctional Complexes

In contrast, early experiments made it very clear that extraordinary measures would be needed to isolate or even observe 5a or 5b. First (Scheme 1), the chlorides in 2a and 2b were ionized in the presence of water¹² to give aquo complexes 3a or 3b. Storage of 3a or 3b under vacuum led to loss of water and mixtures of 3 and 4, with a greater amount of 4 in the case of 4b. When cooled below -40 C in CD₂Cl₂, chelate complex 4a did not react at an appreciable rate with acetylene (Scheme 2), but near -40 C was converted to symmetrical complex 5a, which isomerized smoothly to vinylidene 6a at higher temperatures (0 C, 3 h). In the imidazole case, 4b was inert to acetylene until ca. 10 C, where unsymmetrical addition product 7b was the only product detected. Complex 7b reasonably arises from unseen vinylidene 6b. Interestingly, the NMR spectrum of 7b (e.g., two sets of sharp resonances for the imidazoles at -40 C) shows coalescence behavior at higher temperatures consistent with an equilibrium between 7b and 6b with E_a = 15.9 kcal mol^-1.

In summary, when imidazole derivative 4b reacts with acetylene, alkyne complex 5b is not seen, because of rapid conversion to 7b, whereas, when starting with pyridyl analogue 4a, alkyne complex 5a can be observed at -40 C, but at temperatures closer to 0 C it evolves to 6a. In contrast, control alkyne complex 5c requires temperatures 50-90 C higher for conversion to vinylidene 6c, all showing the profound effects and assistance of the pendant bases on these transformations.

Bonding in vinylidenes 6a and 6c was compared using isotopomers derived from H¹³C¹³CH. Significantly, nearly identical values for ¹J_CC (Scheme 2) are consistent with the same C-C bond length and degree of backbonding in each case.¹³

To clarify the structure of 5a, various unsuccessful attempts were made to grow crystals at -40 C. Instead, spectroscopic characterization¹⁴ of alkyne bonding in 5a and 5c was made using H¹³C¹³CH, resulting in 5a- and 5c-(¹³C)₂. At -40 C the ten-line AA'XX' pattern could be analyzed¹³ to show that there were significant differences in couplings involving the hydrogens (Scheme 2). Data above for ¹J_CC in 6a and 6c show that the electronic effects of the two ligands are identical, consistent with almost identical ¹J_CC values in the complexes. We could find no experimental data on effects of hydrogen bonding on alkyne NMR coupling constants and only a single theoretical paper (on HCCH-OH₂).¹⁵ Comparing alkyne and hydrogen-bonded alkyne, the changes in ¹J_CH and ²J_CH seen by us experimentally (+1.5 and +1.9 Hz) resemble those predicted¹⁵ for HCCH-OH₂ (+2.55 and +1.65 Hz).¹⁶ In short, the differences in NMR couplings are for those couplings to hydrogen, and we attribute this to effects of hydrogen bonding in 5a. Attempts were made to engage the CH bonds of complex 5c in hydrogen bonding. Instead, addition of imidazole 8 led to addition to the alkyne (9).

Because NMR coupling constants are a novel tool for studying alkyne hydrogen bonding, we wanted to provide additional evidence for structure 5a. The remarkable discovery of scalar couplings across hydrogen bonds has stimulated a great deal of experimental work on proteins, DNA, and supramolecular interactions⁸ and theoretical investigations of simple systems, but there appears to have been no use in coordination or organometallic chemistry. Ligand 1a-¹⁵N was made⁹ and converted to 3a-(¹⁵N)₂ and 4a-(¹⁵N)₂. Addition of acetylene at low temperature led to 5a-(¹⁵N)₂, whose ¹³C NMR spectrum acquired at -50 C with decoupling of both ¹H and ³¹P allowed observation of a somewhat broadened¹⁷ doublet (^2hJ_CN = 3 ± 0.5 Hz) as expected for coupling of one natural abundance acetylene ¹³C to the nearest ¹⁵N nucleus, whereas a similar spectrum of 5a showed a singlet.

Previous theoretical work by Del Bene et al. on a simple system¹⁸ showed that when the C-H-N angle is near 135, ^2hJ_CN would be approximately one-third the magnitude when the hydrogen bond is linear. Indeed, DFT calculations on 5a itself show a slightly unsymmetrical structure with two C-H-N angles (124.5 and 135.7).¹⁹ Determining the strength of the hydrogen bonds in 5a remains a subject for future study, but as an estimate, we note that calculations on conversion of 5c to 6c indicate that G = -17.9 kcal mol^-1 whereas for similar conversion of 5a to 6a, G = -12.7 kcal mol^-1. The difference may be attributed to thermodynamic stabilization of 5a by two hydrogen bonds. Despite this effect, experiments show a pronounced kinetic acceleration by the heterocycles on isomerization of alkyne to vinylidene (5 to 6), and importantly, 3a is a very competent catalyst (>99% yield of hexanal from 1-hexyne after 5 h at 70 C using 2 mol % 3a).

In summary, the presence of C-H-N hydrogen bonding in an alkyne complex was revealed using NMR coupling information, from both data within the alkyne ligand as well as between alkyne and pyridine (^2hJ_CN). Ongoing experimental and theoretical studies are designed to elucidate further the favorable effects of bifunctional ligands in alkyne hydration and related reactions, and these will be reported in due course.

Acknowledgment

We thank the NSF for continuing support under Grants CHE 0415783 and 0719575, Cambridge Isotope Laboratories, Inc. for labeled building blocks, Professor John Love for first pointing us to biomolecular NMR literature, and Dr. LeRoy Lafferty for assisting with NMR experiments.