Metal – ligand cooperative activation of nitriles by a ruthenium complex with a de-aromatized PNN pincer ligand †

The pincer complex (PNN)RuH(CO), with a de-aromatized pyridine in the ligand backbone, is shown to react with nitriles in a metal–ligand cooperative manner. This leads to the formation of a series of complexes with new Ru–N(nitrile) and C(ligand)–C(nitrile) bonds. The initial nitrile cycloaddition products, the ketimido complexes 3, have a Brønsted basic (nitrile-derived) Ru–N fragment. This is able to deprotonate a CH2 side-arm of the pincer ligand to give ketimine complexes (4) with a de-aromatized pyridine backbone. Alternatively, the presence of a CH2 group adjacent to the nitrile functionality can lead to tautomerization to an enamido complex (5). Variable-temperature NMR studies and DFT calculations provide insight in the relative stability of these compounds and highlight the importance of their facile interconversion in the context of subsequent nitrile transformations.


Introduction
Organometallic complexes with tridentate pincer-type ligands have found widespread application in catalysis due to their tuneable nature and the robustness of the resulting transition metal complexes. 1Moreover, ligands of this type have attracted increasing attention in recent years also as a result of their ability to activate relatively strong bonds via a pathway that directly involves a (non-spectator) ligand site in bond breaking, called 'metal-ligand cooperation' (MLC). 2 Using this strategy, a reduction in bond order within the substrate is achieved which leads to either complete bond cleavage (in case of a single bond) 3 or weakening (in case of a CO or CN multiple bond).As examples for the latter, the CvO bond in CO 2 4 or organic carbonyl compounds 5 may be activated using metal complexes with de-aromatized pyridine-based pincer ligands.Similarly, metal-ligand cooperative binding/activation of nitriles was reported by Milstein, 6 Pidko, 7 and our group, 8 and this has led to new reactivity involving organic nitriles as either Michael donors 6 or acceptors 8 (Chart 1).The synergistic effect of a Lewis basic fragment (the deprotonated pincer 'arm') and a Lewis acid (the metal centre) is reminiscent of Frustrated Lewis Pairs (FLPs), 9 and nitrile activation using an Al/P-based FLP was recently reported. 10n many of the (catalytic) reactions involving metal-ligand cooperative bond activation, several different species are observed in solution due to competition between different potential substrates, tautomerization/rearrangement reactions of the activated substrates, or both.For example, our group reported that catalytic oxa-Michael addition of alcohols to unsaturated nitriles using 1 proceeds via metal-ligand cooperative activation of the nitrile substrate, but may be inhibited by the competing MLC activation of the alcohol. 8Milstein and co-workers showed the presence of various tautomers in equilibrium upon binding of organic nitriles to de-aromatized PNP Re and Mn pincer complexes. 6Key to the observed chemistry is that these compounds can readily interconvert and have similar energies so that the catalyst does not get trapped in a thermodynamically stable but catalytically inactive state.A better insight in the factors that govern these equilibria is important for understanding and designing improved catalysts that make use of MLC.

Results and discussion
Reaction of 1 with nitriles: identification of products Building on our recently reported results on the activation of 2-and 3-pentenenitrile towards oxa-Michael addition reactivity with Milstein's de-aromatized PNN Ru complex 1, 8 we tested the reactivity of the less activated substrate cinnamonitrile as the Michael acceptor in isopropanol with a catalytic amount of 1.While no oxa-Michael addition was observed, NMR spectroscopy indicated the catalyst 1 to be fully consumed.A stoichiometric reaction between 1 and cinnamonitrile in a teflonsealed NMR tube in C 6 D 6 , either in the presence or absence of isopropanol, indicated clean conversion of 1 to a single ruthenium complex (4a) that contains a cinnamonitrile-derived fragment.The 1 H NMR spectrum of the reaction mixture showed a broadened singlet at δ 9.68 ppm which is assigned to a NH moiety.A set of resonances at δ 6.54, 6.38 and 5.71 ppm due to the pyridine backbone of the pincer ligand indicates that it remains de-aromatized.A doublet at δ 3.77 ppm in the 1 H NMR shows coupling to 31 P (J = 2.8 Hz) and is correlated to a 13 C resonance at δ 69.07 ppm ( J P,C = 54 Hz), suggesting the presence of a vCHP t Bu 2 moiety.The NMR data allow us to assign the structure of 4a as the NH-ketimine complex shown in Scheme 1.In analogy with the reactivity of 1 towards 2-and 3-pentenenitrile 8 and the data of Milstein and co-workers for related nitrile reactivity of PNP Re and Mn complexes, 6 the following pathway for the formation of 4a is proposed.Initial interaction of the nitrile with compound 1 results in the formation of the Lewis acid-base adduct 2a.Tautomerization of the reactive, unsaturated moiety in the pincer ligand from the P t Bu 2 to the NEt 2 side-arm 11 results in metal-ligand cooperative nitrile binding (via C-C and Ru-N bond formation) to give 3a.That nitrile addition takes place at the NEt 2 side-arm is in agreement with previous studies in which CvO/CuN addition reactions are shown to be thermodynamically more favourable at this site.4e,8 Subsequent transformation of the intermediate 3a occurs via deprotonation of the pincer CH 2 P sidearm by the Brønsted basic Ru-N fragment to give the de-aromatized compound 4a as the final product.Attempts to isolate the compound in pure form met with failure: after several hours in solution, decomposition is noticeable in the 1 H NMR spectrum and also quick workup of the mixture by removal of the volatiles resulted in an intractable mixture.
The reaction of compound 1 with 4-pentenenitrile in C 6 D 6 was investigated.Whereas the unsaturated compounds 2-and 3-pentenenitrile (with the nitrile conjugated with the CvC bond or in the allylic position, respectively) both lead to formation of compound A (Chart 1) via a series of tautomerization reactions, 4e,8 the reaction with 4-pentenenitrile reproducibly leads to a different set of products.Analysis of the reaction mixture by NMR spectroscopy indicated the presence of more than one species, as evidenced for example by the appearance of two new major 31 P NMR resonances at 106.0 and 96.1 and a minor species at δ 121.0 ppm, that account for ca.49%, 43% and 8%, respectively, of the total signal intensity.The reaction is complete within 10 min as judged by NMR spectroscopy, and the composition of the mixture remains unchanged for at least 24 h.The 1 H NMR spectrum contains resonances that can be attributed to two major Ru-H species at −13.85 and −21.07 ppm, and a minor component with a Ru-H resonance at −11.88 ppm with integrations that match those observed in the 31 P NMR spectrum.The new Ru-H signal at −21.07 ppm is broadened, which is likely due to chemical exchange: the position of this Ru-H signal is highly dependent on the concentration of 4-pentenenitrile (Fig. 1; vide infra), suggesting that it is due to an equilibrium that involves 4-pentenenitrile.Analysis of 2D NMR data (COSY, HSQC, HMBC) allowed the assignment of the major sharp Ru-H resonance at −13.85 ppm to the 4-pentenenitrile cycloaddition product 4b, a ketimine adduct with a de-aromatized pincer backbone analogous to that observed in the reaction with cinnamonitrile (4a).For the other species that give rise to the exchange broadened Ru-H resonance at −21.07 ppm, Scheme 1 Products obtained from the binding of nitriles to complex 1. 1 H NMR resonances at 6.53, 6.42 and 5.36 ppm due to the central pyridine ring also indicate a de-aromatized ligand.This is most consistent with a rapid equilibrium (K 1 in Scheme 1) between the starting materials 1 + free 4-pentenenitrile and the nitrile adduct 2b.Unfortunately, the minor species could not be further characterized by 1 H NMR spectroscopy due to overlap, but its similarity in 31 P and 1 H (Ru-H) NMR shifts with those of compound A suggest it to be the rearomatized ketimido product 3b.EXSY NMR spectroscopy (mixing time 0.8 s) did not provide evidence for interconversion between the Ru-H species, although it is plausible that 2b and 3b are intermediates in the formation of 4b and therefore involved in an equilibrium.The absence of exchange crosspeaks suggests that the barrier for interconversion between these compounds is sufficiently high that it is slow on the timescale of the NMR experiment.Upon irradiation of the Ru-H NMR signal assigned to 3b in a 1D-NOESY experiment results in a decrease in intensity of the Ru-H moiety of 4b but not 2b, which qualitatively suggests that the rate of 3b reacting to form 4b is faster than its conversion to 2b.Similarly, treatment of 1 with 1 equiv. of acetonitrile led to formation of a reaction mixture that consists of 1, free acetonitrile, and adduct 2c in rapid exchange (broad Ru-H: −24.02 ppm), in addition to the ketimine compound 4c (Ru-H: −13.97 ppm).In this case, signals for 3c are not observed.EXSY spectroscopy at 50 °C for the mixture in this case did show exchange crosspeaks between the Ru-H moieties of 2c and 4c, lending credence to an equilibrium between the various compounds in solution.
The reaction between 1 and benzonitrile on NMR scale was analysed 5 min after mixing, which indicated initial formation of the nitrile adduct 2d.In ca. 1 hour, this was fully converted to a mixture of two new products, which appear in a 1 : 4 ratio.The major species showed a Ru-H signal at −13.22 ppm ( 31 P: 105.8 ppm), indicative of the ketimine complex 4d.The minor species was assigned as the re-aromatized ketimido 3d on the basis of a diagnostic 31 P NMR resonance at 120.9 and Ru-H at −10.02 ppm.
Finally, addition of benzyl cyanide to 1 on NMR scale results in clean transformation of the starting materials in ca. 1 hour to give the enamido complex 5e as the sole product (Ru-H: −12.0 ppm; 31 P: 120.6 ppm).Diagnostic for the formation of the enamido group are the appearance of singlets at δ 5.35 and 5.10 ppm corresponding to the vinylic CH and the Ru-NH moieties, respectively.Formation of 5e is the result of tautomerization of the intermediate 3e, as was reported by Milstein for a Re PNP pincer complex.6a Monitoring the progress of the reaction by NMR spectroscopy allowed observation of an intermediate that is tentatively assigned as 3e on the basis of its NMR data (Ru-H: −13.71 ppm; 31 P: 109.8 ppm).Its rapid disappearance, however, precluded a full NMR assignment of this intermediate.The identity of the final product 5e was confirmed by X-ray crystallography.‡ The molecular struc-ture shows a ruthenium centre in a pseudo-octahedral environment with the tridentate PNN pincer ligand and a CO ligand in the same plane (Fig. 2).The position trans to the hydride is occupied by the nitrile-derived N atom, with a H-Ru-N(3) angle of 164.2( 8 From the data discussed above, it is clear that the reaction of 1 with nitriles can result in a variety of products that are in equilibrium, and the final product (mixture) is highly dependent on the substituents on the nitrile.In order to gain more insight in these equilibria, we studied the system 1 + 4-pentenenitrile in more detail.Addition of increasing amounts of 4-pentenentrile to a C 6 D 6 solution of 1 (see Fig. 1) resulted in a gradual change of the broadened Ru-H resonance from −26.69 ppm in the starting material 1 to −15.07 ppm in the presence of 32 equiv. of 4-pentenenitrile.The composition of the equilibrium mixture changes also upon addition of 4-pentenenitrile: the percentage of ketimine complex 4b increases from 31% of the total Ru-H integration when only 0.5 equiv. of substrate are present, to 64% with 32 equiv.4-pentenenitrile.In all spectra, there is a small amount of 3b present as well (between 6 and 11%).From the data at high 4-pentenenitrile concentration it is possible to estimate the chemical shift of the nitrile adduct 2b as ca.−15.0 ppm.From the position of the exchange-averaged peak for the Ru-H the 1 : 2b ratio may then be determined. 1H NMR integration of the remaining Ru-H signals (from 2b and 3b) allows a complete description of the equilibria involved.Fitting of the 1 H NMR data to a 1 : 1 binding model for the (rapidly exchanging) equilibrium 1 + 4-pentenenitrile ⇄ 2b results in a binding constant of 41 (±20%) M −1 (see ESI † for details). 12The equilibrium constants K 2 and K 3 (for the equilibria 2b ⇄ 3b, and 3b ⇄ 4b, respectively) are determined using the 1 H NMR integrations as 0.5 and 6 M −1 , respectively.
Study of equilibria in the 1 + 4-pentenenitrile product mixture temperature dependence NMR data were also collected at various temperatures for the product mixture obtained from 1 with 1 equiv. of 4-pentenenitrile in toluene-d 8 .At temperatures of −25 °C or lower, the predominant species in solution are 3b and 4b, which are observed in a ca. 1 : 6 ratio.Increasing the temperature results in a gradual appearance of signals due to the nitrile adduct 2b (which again is in fast exchange with 1 + 4-pentenenitrile).At +75 °C or above, <10% of the Ru-H signal intensity is due to 3b/4b, and the remaining signal approaches the chemical shift of pure 1. Keeping the sample at elevated temperature results in the gradual decrease of the total signal intensity, suggesting that the species decompose.The line broadening at low temperature and the decomposition at high temperature only allow reliable determination of concentrations in a limited temperature range (−5 to 55 °C).The temperature dependence of the 1 H NMR spectra (Fig. 3) is in agreement with the components 1 + free 4-pentenenitrile being entropically favoured at high temperature, whereas binding of the nitrile in a metal-ligand cooperative manner (3b and 4b) is preferred at low temperature (exergonic) due to a favourable enthalpy contribution that results from Ru-N and C-C bond formation in these compounds.A Van 't Hoff plot afforded an estimate of the thermo-chemical parameters for the first equilibrium (1 + 4-pentenenitrile ⇄ 2b; K 1 ) of ΔH = −11 kcal mol −1 and ΔS = −32 cal mol −1 K −1 , and for the second equilibrium (2b ⇄ 3b; K 2 ) it gives ΔH = −3 kcal mol −1 and ΔS = −13 cal mol −1 K −1 .The negative values for ΔS are expected for the formation of the nitrile adduct (2b), and its subsequent transformation to the ketimido compound (3b).The equilibrium between 3b and 4b (K 3 ) is hardly affected by changes in temperature, consistent with these having similar structures and thus very similar entropy.

DFT calculations
A key factor that leads to formation of the various products is the Brønsted basicity of the Ru-N moiety that results from metal-ligand cooperative nitrile activation with 1.Based on the experimental data, the ketimido compounds 3 that are initially formed are either fully consumed or only present as a minor component of the reaction mixture.The geometries of the compounds 3-5 was optimized using DFT calculations at the TPSSTPSS level of theory using a LANL2DZ basis set (with effective core potential) on Ru and 6-31G(d,p) for all other atoms.Calculations were carried out in the gas phase, and the stationary points were confirmed to be local minima by subsequent frequency analyses.The energies of these geometries were further refined by carrying out single-point TPSSTPSS calculations using Alrichs' def2-TZVP basis set 13 and employing Grimme's D3 empirical dispersion correction (Table 1). 14We initially focussed on the nitriles that lead to a single product according to NMR spectroscopy, e.g., cinnamonitrile (4a) and benzyl cyanide (5e).Geometry optimizations of the possible products (3a/4a and 3e/4e/5e) from the addition of these nitriles to 1 converge on the desired structures as local minima on the potential energy surface.In case of cinnamonitrile, the difference in free energy between 3a and 4a (which is experimentally observed) is calculated to be 1.3 kcal mol −1 in favour of the latter.Evaluation of the Gibbs free energies for the benzyl cyanide addition products identifies the enamido compound 5e as the most stable product (ΔG = −6.0 and −4.6 kcal mol −1 relative to 3e and 4e, respectively), in agreement with empirical data.The stability of 5e is presumably due to conjugation with the aromatic ring.For the products of 1 with 4-pentenenitrile, the DFT calculations suggest that the   energies of 3b, 4b and 5b differ by only ca. 1 kcal mol −1 , with the latter being the most stable isomer computationally.This is in contrast to the experimental data, which show 4b as the major species.The relative Gibbs free energies of the reaction of 1 with the other nitriles (c/d) included in this study are shown in Table 1.In case of acetonitrile, the calculations correctly predict the de-aromatized ketimine complex 4c to be the most stable.For benzonitrile, the DFT calculations suggest that 3d is favoured by 0.3 kcal mol −1 , which contrasts the experimental observation of 4d as the major isomer.Given the small computed differences in stability (and neglecting solvation in our computational model) we feel that these discrepancies are not unexpected given the complexity of the system. 15In addition, it is likely that (stabilizing) H-bonding interactions with the solvent 16 exist in case of NH-containing ketimine compounds 4, which are absent in 3. We have not calculated the transition states connecting the various species, but note that recent computational studies suggest that involvement of adventitious water or other proton-shuttling agents ( possibly including the nitrile substrates) 17 may lead to significant lowering of the activation energies for these tautomerization steps.6b,18 In connection to this, it should be noted that during the study of the concentration-or temperature-dependence of the system 1 + 4-pentenenitrile, we observed that equilibrium is established within minutes (the time required to setup the NMR experiment).

Conclusions
We have studied how nitrile activation by a de-aromatized pincer ruthenium complex leads to a variety of different products, the stability of which depends on the substitution pattern on the nitrile.Based on our combined experimental/ computational results and those reported in the literature for similar systems, 6 it is clear that the highly dynamic nature of these systems is a key factor in their unusual reactivity: the barriers for interconversion between the various species is low and their Gibbs free energies are close.This allows the system to select the reaction pathway that leads to the most exergonic products, for example resulting in unusual Michael addition reactivity with 1. 8 The data presented here provide additional insight in metal-ligand cooperative substrate activation pathways and subsequent reactivity ( proton-transfer/tautomerization) that results from the Brønsted basicity of the bound nitrile fragment.

Experimental
NMR scale reaction between 1 and cinnamonitrile to give 4a In a J. Young NMR tube, a 1/1 mixture of 1 and cinnamonitrile was dissolved in benzene-d 6 .After shaking for 15 min at RT, the reaction mixture was analysed by NMR spectroscopy.Full conversion to 4a was observed.Minor species (3b): Diagnostic signals: 1 H-NMR: doublet at −11.88 ppm ( J = 28.5 Hz), 31 P-NMR: 121.0 ppm.Further NMR assignment was unsuccessful due to overlap with 4b.

NMR scale reaction between 1 and acetonitrile
In a J. Young NMR tube, an equimolar mixture of 1 and acetonitrile was dissolved in C 6 D 6 .After shaking for 90 min at room temperature, the reaction mixture was analysed by NMR spectroscopy which showed the presence of 2 species: 2c (in rapid exchange with 1 + 4-pentenenitrile):

Fig. 1
Fig. 1 Hydride region of the 1 H NMR spectrum in toluene-d 8 solvent for 1 with varying amounts of 4-pentenenitrile added.
)°. Metal-ligand cooperative activation of the CuN is evidenced by the formation of the C(15)-C(21) and Ru(1)-N(3) bonds.The bond lengths within the central pyridine ring are indicative of re-aromatization, with C-C distances ranging between 1.387(2)-1.393(2)Å and C-N bond lengths of 1.350(2) and 1.347(2) Å.The Ru(1)-N(3) distance is relatively long (2.2033(14) Å) and from the difference Fourier map it was clear that a hydrogen atom is attached to N(3).The nitrilederived N(3)-C(21) bond length of 1.341(2) Å and the short adjacent C(21)-C(22) bond (1.379(2) Å) are consistent with the enamido structure with a re-aromatized pyridine ring, as assigned on the basis of the NMR data for 5e.Study of equilibria in the 1 + 4-pentenenitrile product mixture nitrile concentration

Table 1
DFT calculated relative Gibbs free energies of MLC nitrile activation products 3-5 in kcal mol −1 and experimentally observed percentage of each based on 1 H NMR integration

a
Gibbs free energies in kcal mol −1 relative to the lowest energy isomer.bRelative amount of each component as determined by1  H NMR integration of 1 : 1 mixture 1 + nitrile at room temperature (remainder is 1 + 2).