P–Ru-Complexes with a Chelate-Bridge-Switch: A Comparison of 2-Picolyl and 2-Pyridyloxy Moieties as Bridging Ligands

Abstract

Starting from [Ru(pyO)₂(nbd)] 1 and a N,P,N-tridentate ligand (2a: PhP(pic)₂, 2b: PhP(pyO)₂) (nbd = 2,5-norbornadiene, pic = 2-picolyl = 2-pyridylmethyl, pyO = 2-pyridyloxy = pyridine-2-olate), the compounds [PhP(μ-pic)₂(μ-pyO)Ru(κ²-pyO)] (3a) and [PhP(μ-pyO)₃Ru(κ²-pyO)] (3b), respectively, were prepared. Reaction of compounds 3 with CO and CNtBu afforded the opening of the Ru(κ²-pyO) chelate motif with the formation of compounds [PhP(μ-pic)₂(μ-pyO)Ru(κ-O-pyO)(CO)] (4a), [PhP(μ-pic)₂(μ-pyO)₂Ru(CNtBu)] (5a), [PhP(μ-pyO)₄Ru(CO)] (4b) and [PhP(μ-pyO)₄Ru(CNtBu)] (5b). In dichloromethane solution, 4a underwent a reaction with the solvent, i.e., substitution of the dangling pyO ligand by chloride with the formation of [PhP(μ-pic)₂(μ-pyO)Ru(Cl)(CO)] (6a). The new complexes 3a, 4a, 5a, 5b and 6a were characterized by single-crystal X-ray diffraction analyses and multi-nuclear (¹H, ¹³C, ³¹P) NMR spectroscopy. The different coordination behaviors of related pairs of molecules (i.e., pairs of 3, 4 and 5), which depend on the nature of the P–Ru-bridging ligand moieties (μ-pic vs. μ-pyO), were also studied via computational analyses using QTAIM (quantum theory of atoms in molecules) and NBO (natural bond orbital) approaches, as well as the NCI (non-covalent interactions descriptor) for weak intramolecular interactions.

1. Introduction

Hemilabile ligands are crucial in various kinds of homogeneous catalysis [1,2]. As an example of a relevant Ruthenium-based system, the introduction of a hemilabile site allowed for the development of the well-known Grubbs catalysts (I, Scheme 1), leading to the Grubbs–Hoveyda catalysts (II) [3]. The labile site of the hemilabile ligand stabilizes the metal complex in the absence of alternative electron pair donors, but it may give rise to a vacant coordination site for substrate binding “on demand”. The labile ligator function remains a dangling group in close proximity. Recently, we reported on Ru complexes with 2-pyridyloxy (pyO) ligands, in which a chelating pyO group may exert hemilability by means of a coordinative switch between two centers. Rather than simply opening the chelate (with the formation of a monodentate pyO ligand with a dangling second donor site), the κ²-pyO motif is converted into a bridging μ-pyO motif within a paddle-wheel-like complex, then buttressing the connection to another ligand site in the Ru coordination sphere (Scheme 1) [4,5]. Starting from the initial [PhP(μ-pyO)₃Ru(κ²-pyO)] system, we have shown that this switch also works for the related As–Ru system, and CO (for P–Ru and As–Ru) and NCMe (for As–Ru) were able to trigger this switch. In the related complex [PhSb(μ-pyO)₄Ru(NCMe)], the rather strong binding of pyO to the Sb atom did not allow for the release of NCMe with the formation of a corresponding [PhSb(μ-pyO)₃Ru(κ²-pyO)] chelate complex.

As only one pyO ligand is involved in switching, the aim of the current study was the exploration of the role of the remaining buttresses across the P–Ru core. In previous studies, we had explored systems with other (O,N)-bidentate ligands as well (anions of N-methylbenzamide and phthalimidine), which preferred the bridging position in Pn-Ru-complexes (Pn = pnictogen) and showed no tendency to form chelates at Ru [6,7,8]. However, for the current study with systems of well-defined combinations of bridging ligands, we wanted to retain the pyridine motif as the Ru-binding site, and we also wanted to avoid ligand scrambling, thereby using a bridging moiety tightly bound to phosphorus. Therefore, 2-picolyl (pic) bridges were introduced in this molecular system, and reactions and molecular structures were compared for pairs of corresponding pic-vs.-pyO-bridged compounds. In a similar manner as pyO had been used as a bridging ligand in dinuclear complex systems of transition metals and heavier main group elements such as (RuAs) [5,9], (CuSi), (PdSi) [10], (IrSi) [11], (CoSi) [12] and (RhBi) [13], picolyl bridges have also been successfully used as bridging moieties, e.g., for (CoSi) [14], (FeP) [15], (PdSe) [16] and (PdTe) [17] systems.

2. Results and Discussion

2.1. Syntheses

The starting materials, i.e., [Ru(pyO)₂(nbd)] 1 [4], phosphanes 2a (i.e., phenylbis(2-pyridylmethyl)phosphane) [15] and 2b (i.e., phenylbis(2-pyridyloxy)phosphane) [8], and complex 3b ([PhP(μ-pyO)₃Ru(κ²-pyO)]) [4], have been reported in the literature and were prepared following the protocols reported previously. As 3b was accessible in a straightforward manner from 1 and 2b, our synthesis of 3a ([PhP(μ-pic)₂(μ-pyO)Ru(κ²-pyO)]) followed the same route (Scheme 2). Whereas 3b was crystallized by vapor diffusion of diethyl ether into a dichloromethane (DCM) solution of 3b, crystallization of compound 3a was successful upon vapor diffusion of n-pentane into a DCM solution of 3a (isolated yield 68%). Additionally, synthesis of compound 4a ([PhP(μ-pic)₂(μ-pyO)Ru(κ-O-pyO)(CO)]) was carried out following the protocol reported for the synthesis of compound 4b ([PhP(μ-pyO)₄Ru(CO)]) [4], i.e., a dispersion of 3a in toluene was exposed to CO atmosphere at 60 °C for 3 days (isolated yield < 35%, product still contained starting material 3a). The reaction of 3a (in DCM) or 3b (in toluene) with CNtBu afforded crystals of isonitrile complexes 5a ([PhP(μ-pic)₂(μ-pyO)₂Ru(CNtBu)], upon vapor diffusion of n-pentane, a few crystals only) or 5b ([PhP(μ-pyO)₄Ru(CNtBu)], upon cooling of the toluene solution, isolated yield 53%), respectively. In an attempt to recrystallize a small amount of 4a from DCM (because 4a was contaminated by 3a, see NMR Section 2.3, and because the crystal structure of the toluene solvate of 4a suffered from heavy disorders, see Section 2.2), a few crystals of 6a ([PhP(μ-pic)₂(μ-pyO)Ru(Cl)(CO)]) were obtained.

2.2. Single-Crystal X-ray Diffraction

Compound 3a crystallized in the orthorhombic space group Pna2₁ with two independent (but conformationally very similar) molecules in the asymmetric unit (Figure 1). The molecular configuration of 3a is the same as in the known compound 3b [4], i.e., the Ru atom is located in a distorted octahedral coordination sphere with a RuN₄ equatorial plane and trans-situation of P- and O-donor sites. In both compounds, one pyO ligand chelates at Ru, and the Ru–P bond is bridged by three buttresses ((pic)₂(pyO) in 3a, (pyO)₃ in 3b).

Two noteworthy differences between 3a and 3b are found in the Ru-P-C(phenyl) angles (136.5(2)° in 3a, 147.3(1)° in 3b) and in the P–O separations (Figure 2). The pic substituted P atom of 3a is less Lewis acidic than the pyO-substituted counterpart in 3b. The bridging pyO moiety in 3a establishes a rather weak P···O contact (2.792(5) Å), whereas in 3b the longest P–O separation (1.952(2) Å) is much shorter. Thus, the less pronounced widening of the Ru-P-C(phenyl) angle in 3a can be attributed to this limited P···O interaction. Nonetheless, in 3a the P–CH₂ trans-bond to this weak P···O coordination is slightly elongated. Interestingly, both the Ru–P and Ru–O bonds of 3a are longer than their corresponding bonds in 3b. We attribute this to the weaker π-acceptor phosphane in 3a, which may weaken both bonds at the same time by causing lower Ru→P π-back-bonding contributions and lower resulting O→Ru π-donation.

Compound 4a crystallized from the mother solution as a toluene solvate in the monoclinic space group P2₁/c (Figure 3a). The molecular configuration of 4a exhibits two striking differences to the presumed analog 4b [4]. Whereas the latter resembles a paddle-wheel-shaped molecule with trans situation of P- and CO-ligands at Ru, with all pyO moieties bound to Ru through Ru–N bonds, compound 4a exhibits cis arrangement of P- and CO-ligands at Ru with an only three-fold buttressed Ru–P bond. Furthermore, the monodentate pyO in 4a is Ru–O-bound with a dangling N donor site. This configuration about the Ru–P core is essentially retained in compound 6a, which is crystallized from dichloromethane solution as a DCM solvate in the monoclinic space group P2₁/n (Figure 3b). Even though the molecular configuration of 4a was unexpected at first sight (with respect to the known paddle–wheel shape of 4b), [Ru(dppe)(CO)(NCMe)₃][OTf]₂ [18] is an example of another octahedral Ru(II) complex in which the CO ligand occupies a cis-P,trans-N position, even though sterics would allow for the alternative cis-N,trans-P arrangement. As for the Ru-O-bound monodentate pyO moiety, a paddle-wheel complex with a Ru≡Ru axis, which has been reported by Powers et al., also bears a Ru-O-bound pyO derivative [19]. Because of the configurational analogies of 4a and 6a, Figure 4 provides a direct comparison of these two molecular structures. Their Ru-P-C(phenyl) angles (132.0(2)° in 4a and 132.3(1)° in 6a) are essentially identical, and replacement of pyO by Cl (i.e., replacing of one π-donor ligand by another π-donor) did not alter the Ru–P bond length in a noteworthy manner. Additionally, in both compounds the dangling pyO oxygen atom coordinates the P atom from a remote position with slightly different separations of the P···O contacts (2.701(4) in 4a vs. 2.637(2) Å in 6a), with also only a small difference between the molecular shapes of 4a and 6a and similar to the corresponding P···O interatomic distance in 3a. Also related to 3a is the slight P–CH₂ bond elongation trans to this P···O contact. In the Ru coordination sphere, the corresponding Ru–N and Ru–C bonds also exhibit similar lengths. In this regard, the Ru–N trans-bond to the CO ligand is significantly longer than the other two Ru–N bonds of the mutually trans-situated pyridine moieties. We attribute this structural feature to weakened Ru→N π-back-bonding trans to the CO ligand, which itself causes strong Ru→C π-back-bonding. The latter is indicated by the C≡O stretching vibration at 1931 cm⁻¹, which gives rises to a strong band in the IR spectrum of compound 4a. This is just at slightly higher wave numbers than the C≡O stretch found for compound 4b (which is at 1921 cm⁻¹) and indicative of strong Ru→C π-back-bonding. Additionally, this structural feature of a long Ru–N(pyridine) bond trans to a CO ligand can be found in other octahedral Ru(II) complexes, e.g., in the cationic complex [Ru(phen)(bpy)(CO)Cl]⁺ with cis-arranged monodentate ligands [20].

Compound 5a crystallized as a DCM solvate in the triclinic space group P$\overline{1}$ (Figure 5a) and the analogous isonitrile complex 5b crystallized in the orthorhombic space group Pnma (Figure 5b). In this set, the molecules adopt related configurations with trans-situated P- and isonitrile-ligands at the RuN₄ core. The most striking difference is associated with the different P-O-binding of the pyO bridging ligands. Even though the pyO oxygen atoms approach the P atom in a very direct manner (resulting in O-P-C angles in 5a and O-P-O angles in 5b wider than 170°), in 5a two rather long P···O contacts (2.814(2) and 2.770(2) Å) leave room for a smaller Ru-P-C(phenyl) angle (143.9(1)°), whereas in 5b the stronger binding of the pyO oxygen atoms (P–O separations of 1.713(3) and 2.309(3) Å) causes noticeable widening of the Ru-P-C(phenyl) angle (158.1(2)°), thereby approaching a distorted octahedral coordination sphere of the P atom in 5b. In accordance with the pair 3a/3b, compound 5a exhibits a longer Ru–P bond than compound 5b (2.308(1) vs. 2.270(2) Å), which again can be attributed to the weaker π-acceptor phosphane in 5a. The trans-disposed Ru–C bond, however, is slightly shorter in 5a (1.957(2) vs. 1.995(6) Å). As isonitriles are ligands with significant π-acceptor features, the Ru1–C29 bond in 5a is likely to respond to the weaker competing π-acceptor phosphane with enhanced Ru→C π-back-bonding. This explanation is supported by the slightly longer Ru–C bond (2.009(6) Å) in complex [Ru₂(CO)₅(tBuNC)(bpcd)] (bpcd = 4,5-bis(diphenylphosphino)-4-cyclopenten-1,3-dione) [21], which also features a tBuNC-Ru-P trans-arrangement and additional strong π-acceptor ligands. Furthermore, it is backed by IR spectroscopic data (bands of the CN stretch) for complexes 5a and 5b (see Appendix B).

2.3. Solution and Solid State NMR Characterization

In the solid state, compound 3a bears a set of two chemically non-equivalent pyO-ligands (a chelating and a bridging group), and resulting therefrom two non-equivalent pic-moieties. In dichloromethane solution, the two types of pyO ligands undergo rather rapid exchange (with respect to the NMR time scale), as both the ¹H and ¹³C NMR spectra of 3b exhibit one set of signals for pyO- and one for pic-ligands (for ¹H, ¹³C and ³¹P NMR spectra of 3a and of compounds 4a, 5a, 5b and 6a, see Figures S1–S19 in the Supplementary Materials). Especially in the ¹H spectrum, the former signals are noticeably broadened and do not exhibit coupling patterns (related exchange of pyO-ligands was found in 3b [4]). As to the pic-ligands, only their CH₂ groups suffer severe signal broadening both in ¹H and ¹³C spectra. With respect to the phosphorus coordination sphere, the average coordination number in solution is close to the situation in the solid state, the ³¹P NMR shift in CD₂Cl₂ solution (80.8 ppm) is just slightly shifted downfield relative to the values found for the two crystallographically independent P sites in solid 3a (76.3 and 73.8 ppm). With respect to the ³¹P NMR shift of the free phosphane PhP(pic)₂ (δ³¹P = −13.7 ppm in CDCl₃), the corresponding NMR signal of 3a is shifted downfield by about 90 ppm. The products arising from 3a (i.e., 4a, 5a and 6a) may exhibit different configurations, which will be indexed with superscript-1 and superscript-3 in accord with the isomers under investigation by computational analyses (vide infra). In the crystal structures (Figure 3 and Figure 5), we encountered 4a¹, 5a³ and 6a¹. In solution ¹H NMR spectra, the pic-CH₂ groups are a convenient probe for assigning configurations 1 and 3, as in the former the molecules bear four chemically non-equivalent CH₂ protons, while in the latter the molecules bear two symmetry-related CH₂ groups with diastereotopic protons. Figure 6 shows the section of CH₂ signals of the ¹H NMR spectra of compounds 3a, 4a, 5a and 6a.

Compound 6a (devoid of a second pyO group) must exhibit a configuration with four chemically independent CH₂ protons. Its ¹H spectrum clearly shows the four signals, which are “dd” patterned by ²J(³¹P¹H) and ²J(¹H¹H) coupling for each proton (Figure 6, red). Apart from the signals of small amounts of 3a, the spectrum of 4a exhibits essentially the same signal pattern as 6a, showing retention of configuration 4a¹ in the solution (Figure 6, blue). The ³¹P NMR signal of 4a in the CD₂Cl₂ solution (δ³¹P = 68.6 ppm) is slightly shifted upfield with respect to the signal of 3a. In the solid state, ³¹P NMR signals of 4a were found at 63.5 and 68.1 ppm (Figure 7). Again, the set of crystallographically independent molecules gives rise to more than just one signal. Furthermore, the ³¹P CP/MAS NMR spectrum of 4a indicates the presence of 3a in the solid product. Hence, the signals of 3a encountered in solution spectra of 4a arose from impurities in the solid product rather than from decomposition of 4a with release of CO.

Upon addition of some drops of CNtBu to a CD₂Cl₂ solution of 3a, which corresponds to large excess of the isonitrile, spectra 5a (initial 1) and 5a (initial 2) were obtained (sample 5a (initial 1) contained somewhat larger excess of CNtBu, while in sample 5a (initial 2) there was only a 15% excess according to integral traces of Ru-bound and free CNtBu), cf. Figure 6 (olive, orange). The appearance of only two “dd”-patterned CH₂ proton signals indicates the formation of configurational isomer 5a³ under these conditions, which is in accordance with the structure of 5a found in the solid state. Within 20 h, the latter sample had undergone some decomposition (pale green spectrum 5a (initial 2′)). Large amounts of 3a formed (thus, 5a³ must have released CNtBu), and CH₂ signals of a new complex appeared. (Unfortunately, we were not able to isolate this new compound by layering the NMR sample solution with n-pentane at this stage.) Upon dissolution of some crystals of the isolated 5a in CD₂Cl₂, however, a ¹H NMR spectrum was obtained, which indicated the presence of both isomers 5a³ and presumably 5a¹ (the coupling patterns of the four CH₂ signals are very well in accordance with those of 4a and 6a), as well as significant amounts of 3a (Figure 6, purple). Again, the latter can be explained by dissociation of 5a in solution with formation of 3a and CNtBu in a dynamic equilibrium as the (corresponding to the proton intensities of 3a) relative intensity of 9 H atoms was observed for the signal of free CNtBu in the same spectrum. As compound 5a (isomer 5a³) forms in the first instance upon adding CNtBu to 3a, but also undergoes dissociation (with formation of 3a and CNtBu) as well as decomposition in solution (most likely driven by free CNtBu), the solution NMR data reported in the experimental section are based on spectra of 5a generated in situ. The identity of the decomposition products (apart from 3a) has not been established yet. This initial formation of 5a³ and rather quick decomposition observed for this compound in DCM solution (which should include formation of 5a¹) serves as an explanation as to why we obtained crystalline 5a (mixture of isomers) in very poor yield, and timely workup of the synthesis mixture was required to isolate 5a at all.

The related compound 5b, in CD₂Cl₂ solution, produced one set of ¹H and ¹³C signals of the pyO moieties, indicating both retention of the paddle-wheel isomer 5b³ in solution and either the transition toward a more symmetrical arrangement of ligands (with essentially equal P–O bond lengths) or rapid exchange of the pairs of short and long trans-situated P–O bonds. Additionally, the ³¹P NMR shift of 5b (28.0 ppm), which is noticeably shifted upfield with respect to the starting material 3b (135.0 ppm) [4] and much closer to the shift of 4b (−9.9 ppm) [4], indicates retention of the higher coordination number of the P atom of 5b in solution. Furthermore, the ¹J(P-C(Phenyl)) coupling (225 Hz for 5b in CD₂Cl₂ solution, 277 Hz for 4b [4]) indicates related coordination spheres about the P atoms of these two compounds. In contrast to 5a, the ¹H NMR spectrum of 5b did not hint at the liberation of CNtBu in solution. (Traces of 3b, the presence of which is apparent in the ³¹P NMR spectrum with a signal at 136.1 ppm, must have been contaminations in the solid product).

2.4. Computational Analyses

2.4.1. Relative Stability of Configurational Isomers

The question as to why compound 4a formed the isomer with cis-disposed P- and CO-ligands was addressed with the aid of computational analyses. For that purpose, the molecular structures of seven different isomers of 4a were optimized with consideration of the effects of solvent (COSMO model for toluene), dispersion and relativistic effects. Figure 8 shows the molecular structures of the isomers. (The atomic coordinates and total energies of isomers under investigation in this chapter are listed in the Supplementary Materials, see Figures S20–S32 and Tables S1–S13. Isomer 4a¹ corresponds to the isomer found in the crystal, cf. Figure 3a. Isomers 4a² and 4a³ resemble paddle-wheel-shaped isomers (P-trans-CO arrangement) with mutually trans- or cis-disposed pic-CH₂-groups, respectively. Another set of isomers is related to isomer 4a¹ but with Ru-N-bound monodentate pyO ligand (trans to the phosphane moiety.) The hindered rotation about its Ru–N bond gave rise to four different local minima 4a⁴–4a⁷. According to this analysis, the crystallographically encountered isomer 4a¹ represents the favored isomer, whereas the paddle-wheel-shaped isomers 4a² and 4a³ are about 35 and 15 kcal mol⁻¹ less stable, respectively. The trans-arrangement of the CH₂-P-CH₂ motif (in 4a²) exerts a particularly destabilizing effect. Isomers 4a⁴–4a⁷ (with relative energies ranging between 3 and 8 kcal mol⁻¹) are noticeably more stable than the paddle-wheel forms.

As CO complex 4b and isonitrile complexes 5a and 5b crystallized as isomers with configurations related to paddle-wheels (related to 4a³ in particular), Gibbs free energy values were analyzed for those isomers (denoted with superscript-3) in comparison with their respective isomer, which corresponds to the cis-P-Ru-CO arrangement and Ru-O-bound pyO in 4a¹ (isomers denoted with superscript-1). Surprisingly, 3b¹, 5a¹ and 5b¹ were found to be the thermodynamically favored isomers. However, their paddle-wheel-shaped alternatives were only 4.3 kcal mol⁻¹ (4b³) and 6.2 kcal mol⁻¹ (5b³) less stable for (μ-pyO)₄-bridged complexes. The pic-ligands, however, destabilize the paddle-wheel arrangement for isonitrile complex 5a too (5a³: +12.9 kcal mol⁻¹).

Even though solvent effects, dispersion effects and relativistic effects have been considered in the above analysis, we cannot rule out computational experimental errors which could amount to error bars of some kcal mol⁻¹ for the relative energies and isomer 3 to be slightly more stable than isomer 1 in some cases. However, in the context of the observations made in solution (cf. NMR spectroscopy Section 2.3), we attribute the formation of isomers 4b³, 5a³ and 5b³ to kinetic effects (preferred pyO-chelate opening with dissociation of the Ru–O bond and formation of the complex with trans-P-Ru-C arrangement, and if thermodynamically favorable and kinetically feasible, conversion into a different isomer). Rearrangement into the thermodynamically more stable isomer 1 would involve some steps such as changes of Ru–N- vs. Ru–O-coordination modes of a dangling pyO group and site exchange of the CO or isonitrile ligand. For complexes 4b and 5b, both steps are kinetically hindered to a greater extent. The O atom of the otherwise “dangling” pyO ligand is incorporated in a more or less tight bond with the P atom, and the unavailability of this O donor site appears to hinder Ru–C dissociation as well (no formation of free CNtBu observed in solution of 5b).

2.4.2. P···O and C–H···(O,C) Interactions

Comparison of pairs of related molecular structures (3a/3b and 5a/5b, see Section 2.2) allowed for the conclusion that pic-moieties at the phosphane ligand (in 3a and 5a) destabilize the bridging coordination mode of the pyO moieties in the same molecule. In order to elucidate the effect of pic- vs. pyO substitutions on different facets of bonding, computational analyses in this and the following sections were performed for structurally related molecules, which allowed for comparisons, i.e., the pair of related molecules 3a/3b as well as the more or less paddle-wheel-shaped complexes 4b, 5a and 5b. In the pic-functionalized compounds, bridging pyO groups establish P···O contacts with rather long interatomic separations only, which hint at weak electrostatic attraction. Therefore, we visualized the P-pic- and P-pyO-interplay with the non-covalent interactions descriptor (NCI, Figure 9). In addition to the Ru–N bond, the bridging pyO moieties establish two general types of further attractive interactions, which stabilize them in their bridging position: (A) an attractive P-O-contact and (B) an attractive hydrogen contact between H⁶ of the pyO group and the ligand atom trans-disposed to the phosphane. In Figure 9, these interactions are pointed out for 3a with red arrows. Whereas the P-O-contacts in pyO-bridged complexes 3b and 4b are dominated by covalent interactions, the longer P-O contacts in 5b (still exhibiting covalent contributions) are transitioning toward electrostatic interactions. In the pic-bridged complexes 3a and 5a, the P···O-contacts are non-covalent in nature, and according to the color scale are of similar intensity as the C–H···O contacts in 3a. Compounds 4b, 5a and 5b exhibit electrostatic attraction between the pyO-H⁶ and the C atom of the monodentate ligand trans to the phosphane (CO or isonitrile, respectively). This attraction, however, is less intense than the corresponding C–H···O interactions in 3a and 3b, which can be attributed to the lower electronegativity of C vs. O. Thus, one contribution to the driving force in the formation of isomer 4a¹ (cf. Figure 8) can be attributed to the retention of the trans-P-Ru-O arrangement, which electrostatically stabilizes three bridging ligands via C–H···O interaction at the cost of an only weakly attractive potential P···O interaction (which could have been established in paddle-wheel isomer 4a³).

2.4.3. Topological Analysis with Quantum Theory of Atoms-In-Molecules

For the analysis of certain characteristic bond features, to supplement the insights from Section 2.4.2, a quantum theory of atoms-in-molecules (QTAIM) analysis was performed for compounds 3a, 3b, 4a, 5a and 5b.

Table 1. Selected features of (3, −1) critical points in compounds 3a, 3b, 4a, 5a and 5b. Note: electron density ρ(r_b) in au, Laplacian of the electron density ∇²ρ(r_b) in au, Lagrangian kinetic energy density G(r_b) in au, potential energy density V(r_b) in au, ratios |V(r_b)|/G(r_b) and G(r_b)/ρ(r_b) in au, electron energy density H(r_b) in au, ellipticity of the electron density ε and Wiberg Bond Index (WBI).

Compd	Bond 1	ρ(rb)	∇2ρ(rb)	G(rb)	V(rb)	|V(rb)|/G(rb)	H(rb)	G(rb)/ρ(rb)	ε	WBI
3a	Ru–P	0.1264	0.1395	0.0968	−0.1588	1.640	−0.0620	0.766	0.099	1.407
	P···O	0.0233	0.0588	0.0148	−0.0149	1.005	−0.0001	0.635	0.344	0.126
3b	Ru–P	0.1393	0.1368	0.1086	−0.1830	1.685	−0.0744	0.780	0.099	1.491
	P–O 2	0.1558	0.4284	0.2222	−0.3372	1.518	−0.1151	1.426	0.045	1.167
	P–O	0.1088	0.0343	0.0911	−0.1735	1.906	−0.0825	0.837	0.248	0.854
	P–O	0.0960	−0.0240	0.0614	−0.1287	2.098	−0.0674	0.639	0.244	0.734
4b	Ru–P	0.1137	0.0844	0.0697	−0.1183	1.697	−0.0486	0.613	0.011	0.945
	Ru–C	0.1417	0.5397	0.1980	−0.2610	1.318	−0.0630	1.397	0.042	1.716
	P–O	0.1179	0.0878	0.1124	−0.2029	1.805	−0.0905	0.953	0.350	0.918
	P–O	0.1160	0.0671	0.1061	−0.1955	1.842	−0.0894	0.915	0.345	0.900
	P–O	0.1094	0.0093	0.0860	−0.1697	1.973	−0.0837	0.786	0.383	0.841
	P–O	0.1052	−0.0170	0.0748	−0.1539	2.057	−0.0791	0.711	0.412	0.813
5a	Ru–P	0.1018	0.1566	0.0792	−0.1192	1.506	−0.0400	0.777	0.040	1.034
	Ru–C	0.1331	0.4778	0.1741	−0.2287	1.314	−0.0546	1.308	0.061	1.424
	P···O	0.0218	0.0556	0.018	−0.0136	0.989	−0.0002	0.632	0.277	0.109
	P···O	0.0197	0.0528	0.0127	−0.0122	0.959	−0.0005	0.645	0.419	0.093
5b	Ru–P	0.1130	0.1581	0.0876	−0.1357	1.549	−0.0481	0.776	0.006	1.107
	Ru–C	0.1216	0.4633	0.1603	−0.2048	1.278	−0.0445	1.319	0.053	1.351
	P–O 3,4	0.1441	0.3535	0.1943	−0.3002	1.545	−0.1059	1.348	0.110	1.110
	P···O 4	0.0503	0.0652	0.0280	−0.0397	1.418	−0.0117	0.557	0.229	0.359

¹ P–O bonds (from top to bottom for each compound) are in the order of increasing interatomic separation. ² This P–O bond occupies an apical position in the square-pyramidal P coordination sphere and is noticeably shorter than the other two P–O bonds in the same molecule (1.68 vs. 1.88 and 1.95 Å) [4]. ³ This set of P–O bonds in 5b is noticeably shorter than the P–O bonds in compound 4b (1.71 vs. 1.83–1.90 Å) [4]. ⁴ Because of the symmetry of the molecule (cf. Figure 5b), it features two pairs of chemically equivalent P-O-interactions. Redundant data have been omitted from this table.

lists selected characteristic features for selected bonds at their (3,−1) critical points (i.e., bond critical points, BCPs). In all cases, BCPs were detected between the P atoms and the O atoms of the bridging pyO ligands; therefore, the weak P···O interactions are included in this discussion. The Ru–P bonds are similar to one another in terms of electron density ρ at the BCP. A slight decrease in this value is observed from compounds 3 via 4 to 5, which is in accord with the trend of increasing bond length in this order. The ratio of the modulus of the potential energy density per Lagrangian kinetic energy density is in the range of 1 < |V(r_b)|/G(r_b) < 2 in all cases and is indicative of an intermediate bond characteristic (i.e., closed-shell covalent bond with additional ionic contribution). The Ru–P bonds do not exhibit any noticeable ellipticity ε of the electron density (≤0.1 in all cases). However, the Wiberg bond index (WBI) indicates certain multiple bond characteristics for compounds 3 (WBI ≈ 1.4–1.5), whereas a WBI close to 1 is shown for Ru–P single bonds in compounds 4 and 5. In compounds 4 and 5, the Ru–C bonds to CO or isonitrile, respectively, exhibit some multiple bond characteristics, with WBI values ranging between 1.4 and 1.7. As for Ru–P, the Ru–C bonds’ ellipticity of electron density is also < 0.1, indicating the radial symmetry of the bonding contributions. The π-donor trans to Ru–P in 3a and 3b, as well as the higher multiple bond character of Ru–CO in 4b over Ru–CNtBu in 5a and 5b in combination with the lower WBI of Ru–P in 4b, indicate competing Ru→L π-acceptor ligand contributions as the origin of this variety of WBI values observed for compounds 3, 4 and 5.

The P–O bonds in compounds 3, 4 and 5 can be divided into three groups. The very long P···O contacts (as found in pic-bridged Ru-P-complexes 3a and 5a) exhibit very low electron density values at the BCP (ca. 0.02 au), a ratio |V(r_b)|/G(r_b) very close to 1 and a WBI close to 0.1, supporting the interpretation of weak donor–acceptor interactions. Furthermore, the electron energy density H(r_b) is close to zero, also supporting the absence of covalent bonding. In the context of their low total electron density, minor variations in electron density distribution already cause large effects on ε, while the values of ε > 0.3 encountered with these P···O contacts should not be interpreted as the results of multiple bonding. The second group of P-O-interactions corresponds to formally covalent P–O single bonds with a WBI close to 1. The short ones of this group (i.e., the apical P–O bond at the square–pyramidal-coordinated P atom in 3b and the shorter P–O bonds of 5b) exhibit electron densities in the range 0.14–0.16 au, a WBI slightly above 1 and a ratio |V(r_b)|/G(r_b) close to 1.5. The latter, as mentioned above, is characteristic of closed-shell covalent bonds with additional ionic contributions. The longer ones of this group, which are part of nearly symmetrical linear O–P–O-arrangements, exhibit WBI values slightly below 1, somewhat lower electron density values (in the range 0.10–0.12 au) and a ratio |V(r_b)|/G(r_b) closer to 2. Most of these bonds also exhibit noticeably enhanced ellipticity of their electron density at the BCP. The third “class” of P–O bonds encountered with these compounds is the pair of long P–O bonds in 5b. Their features are intermediate between those of the two former groups: an electron density value of 0.05 au, 1.5 < |V(r_b)|/G(r_b) < 1 and WBI of 0.36. This intermediate situation of this set of P–O bonds, as detected by this topological analysis, is in accord with the intermediate situation found for the same bonds in the NCI analysis (cf. Figure 9).

2.4.4. NBO-/NLMO-Analyses

For a closer view of π-back-bonding contributions and weak donor–acceptor-interactions between pyO-ligands and P atoms, as well as for insights into the atomic contributions to the Ru–P σ-bond, we analyzed natural bond orbitals (NBOs) and natural localized molecular orbitals (NLMOs) of compounds 3a, 3b, 4b, 5a and 5b.

Table 2. Natural charges (NCs) of Ru- and P-atoms and contributions to the NLMO of the Ru–P σ-bonds of compounds 3a, 3b, 4b, 5a and 5b, as well as energy levels of selected intramolecular donor–acceptor interactions (obtained from second-order perturbation theory) in kcal mol⁻¹ (∑E(π-Ru→P) = sum of π-back-bonding contributions into relevant σ-antibonding P–O- or P–C-based orbitals (∑E(π-Ru→C) = sum of π-back-bonding contributions into relevant π-antibonding C–O- or C–N-based orbitals of the CO or CNtBu ligand).

	3a	3b 1	4b 1	5a	5b
NC(Ru)	0.229	0.181	0.032	0.080	0.036
NC(P)	1.335	1.748	1.730	1.263	1.741
NLMO σ-Ru–P	36.7% Ru	36.2% Ru	45.9% Ru	36.6% Ru	35.1% Ru
	61.0% P	61.5% P	52.0% P	60.9% P	62.0% P
E(σ-O→P)	5.6	-	-	3.4, 3.7	17.8, 17.8
∑E(π-Ru→P)	11.5	30.4	13.4	4.5	7.3
∑E(π-Ru→C)	-	-	53.3	40.0	36.1

¹ Even though NLMO analyses had been performed for 3b and 4b previously [4], we repeated the calculations with the method–basis set combination used in the current paper for the sake of comparability.

lists the natural charges (NCs) of these compounds’ Ru- and P-atoms as well as these atoms’ contributions to the Ru–P σ-bond. The NCs of the Ru atoms are only slightly positive, and replacement of the Ru-bound O atom by a monodentate ligand (CO or CNtBu) lowers the Ru atom´s NC by ca. 0.15. Even though an anionic π-donor ligand atom is replaced by a charge-neutral π-acceptor ligand, the different electronegativities (O vs. C) appear to dominate the effect on the NC. In accord, replacing two P-bound pyO moieties by pic moieties (i.e., replacement of P–O by P–C bonds) results in lowering of the P atom´s NC by ca. 0.4. In spite of the variable number of additional O-donor sites in its proximity, the P-atom´s NC is almost identical for the PhP(pyO)_nRu-compounds 3b, 4b and 5b. For those compounds, which exhibit σ-O→P donor–acceptor interactions (3a, 5a, 5b), second-order perturbation theory analysis revealed the relevant orbital interactions (Figure 10) and the energies E(σ-O→P) listed in Table 2. The increasing intensity of those interaction energies (ca. 4, 6 and 18 kcal mol⁻¹ for 5a, 3a and 5b, respectively) resembles the increasing intensity indicated along this series in the NCI analyses (Figure 9).

The σ-Ru–P bonds, in most cases, can be interpreted as polar covalent bonds with ca. 2/3 phosphorus contributions. Comparison of the σ-Ru–P relevant NLMOs of compounds 3a and 3b shows that the exchange of bridging moieties (pic vs. pyO) has only a marginal influence on the Ru–P σ-bond. For compounds 4b and 5a, NBO/NLMO analyses initially afforded delocalized Ru–P σ-bonds (noticeable delocalization of the σ-Ru–P NLMO across the P-Ru-C axis, involving significant atomic orbital contributions of the trans-disposed ligand, CO or CNtBu). This was not unexpected, as in a previous analysis we had already encountered such a delocalized situation with 4b [4]. For the sake of comparability, in order to obtain corresponding NLMOs with predominant two-atom contributions in compounds 4b and 5a as well, the occupancy threshold of the Lewis structure search was adjusted (from an initial value of 1.65 to 1.57 or 1.51 for 4b or 5a, respectively). The NLMOs thus obtained revealed close similarities between 5a and 5b, underlining that pic- vs. pyO-exchange does not affect the σ-Ru–P bond significantly. (Graphical representations of the σ-Ru–P NLMOs can be found in the Supporting Information, Figure S34.) The composition of the σ-Ru–P NLMO of compound 4b hints at a more covalent situation (equal orbital contributions from both atoms involved).

Replacing pyO- with pic-bridges, however, has significant influence on the Ru–P bond with respect to Ru→P π-back-donation. For compound 3b, second-order perturbation theory analysis revealed a total of 30.4 kcal mol⁻¹ Ru→P π-back-bonding energy with two major contributions of 20.2 and 7.8 kcal mol⁻¹ into the P–O-based σ-antibonding orbitals and a minor contribution of 2.3 kcal mol⁻¹ associated with σ*(P–C(Ph)). In compound 3a, the energetically small back-bonding contributions into the σ*(P–C) NBOs result in a total of only 11.5 kcal mol⁻¹. With respect to 3b, the weaker Ru→P π-back-donation in 3a is in accord with the lower WBI and the longer Ru–P bond found for 3a. Introduction of trans-disposed π-acceptor ligands (CO or CNtBu) lowers the Ru-P π-back-donation, as expected. Interestingly, in spite of the stronger π-acceptor CO, π-back-bonding to the phosphorus site is still more efficient in 4b than in 5b. We attribute this to the more symmetrical (nearly square-shaped) PO₄ moiety of 4b, the σ*-O-P-O orbitals of which serve as π-acceptors. This formal flow of electron density in the Ru→P direction in the π-system appears to be compensated for by enhanced σ-Ru←P donation. The contributions of P and Ru to the NLMO, which is representative of the σ-bond, indicate a shift of the electron pair toward Ru. (Graphical representations of the NBOs involved in π-Ru→C, and where applicable π-Ru→P interactions can be found in the Supporting Information, Figures S35–S37).

3. Materials and Methods

3.1. General Considerations

Starting materials [Ru(pyO)₂(nbd)] 1 [4], 2a [15] and 3b [4] were prepared following the literature protocols. CNtBu (Sigma-Aldrich, Steinheim, Germany, 98%) was used as received without further purification. CD₂Cl₂ (Deutero, Kastellaun, Germany, 99.6%), acetonitrile (Roth, Karlsruhe, Germany, >99.95%) and n-pentane (Th.Geyer, Renningen, Germany, >99%) were stored over activated molecular sieves (3 Å) for at least 7 days and used without further purification. Dichloromethane was distilled from calcium hydride, while diethyl ether and toluene were distilled from sodium benzophenone. All reactions were carried out under an atmosphere of dry argon utilizing standard Schlenk techniques. Solution NMR spectra (¹H, ¹³C, ³¹P) were recorded on Bruker Avance III 500 MHz and Bruker Nanobay 400 MHz spectrometers. ¹H and ¹³C chemical shifts are referenced to Me₄Si (0 ppm) or to solvent signals of CHDCl₂ (¹H 5.32 ppm) and CD₂Cl₂ (¹³C 53.84 ppm) as internal references, ³¹P shifts are reported relative to 85% H₃PO₄ (0 ppm). For compound 3a, ¹H and ¹H COSY as well as ¹H,¹³C HMBC and HSQC spectra were recorded for signal assignment. ³¹P (CP/MAS) NMR spectra were recorded on a Bruker Avance 400 WB spectrometer with 2.5 mm zirconia (ZrO₂) rotors at an MAS frequency of υ_spin = 15 kHz (3a) or 10 kHz (4a·(toluene)). Infrared spectra of 4a, 5a and 5b were recorded on a Nicolet 380 FT-IR instrument in ATR mode. Elemental analyses were performed on an Elementar Vario MICRO cube. For single-crystal X-ray diffraction analyses, crystals were selected under an inert oil and mounted on a glass capillary (which was coated with silicone grease). Diffraction data were collected on a Stoe IPDS-2/2T diffractometer (STOE, Darmstadt, Germany) using Mo Kα-radiation. Data integration and absorption correction were performed with the STOE software programs XArea and XShape, respectively. The structures were solved by direct methods using SHELXS-97 or SHELXT and refined with the full-matrix least-squares methods of F² against all reflections with SHELXL-2014/7 or SHELXL-2018/3 [22,23,24,25,26]. All non-hydrogen atoms were anisotropically refined. Hydrogen atoms were isotropically refined in idealized position (riding model). For details on the data collection and refinement (incl. the use of SQUEEZE in the refinement of the structures of 4a and 6a), see Appendix A. Graphics of molecular structures were generated with ORTEP-3 [27,28] and POV-Ray 3.7 [29]. CCDC 2162022 (3a), 2162023 (4a·(toluene)), 2162024 (5a·1.5(CH₂Cl₂)), 2162025 (6a·1.5(CH₂Cl₂)) and 2162026 (5b) contain the supplementary crystal data for this article. These data can be obtained free of charge from the Cambridge Crystallographic Data Centre via https://www.ccdc.cam.ac.uk/structures/ (accessed on 14 April 2022).

The geometry optimizations were carried out with ORCA 5.0.2 [30] using the restricted PBE0 functional with a relativistically recontracted Karlsruhe basis sets ZORA-def2-TZVPP [31,32] (for H, C, N, O, P) and SARC-ZORA-TZVPP (for Ru) [33], the scalar relativistic ZORA Hamiltonian [34,35], atom-pairwise dispersion correction with the Becke–Johnson damping scheme (D3BJ) [36,37] and COSMO solvation (toluene, ε = 2.38, r_solv = 3.48). VeryTightSCF and slowconv options were applied and the DEFGRID3 was used with a radial integration accuracy of 10 for ruthenium for all calculations. Calculations were started from the molecular structures obtained by single-crystal X-ray diffraction analysis and isomers were created by modifying these structures. Numerical frequency calculations were performed to prove convergence at the local minimum after geometry optimization and to obtain the Gibbs free energy (293.15 K). The calculated C≡N stretching vibrations were taken from the numerical frequency calculations. On the final structures, single-point calculations were performed with a restricted B2T-PLYP functional with relativistically recontracted Karlsruhe basis sets ZORA-def2-TZVPP [31,32] (for H, C, N, O, P) and SARC-ZORA-TZVPP (for Ru) [33] and utilizing the AutoAux generation procedure [38], the scalar relativistic ZORA Hamiltonian [34,35], atom-pairwise dispersion correction with the Becke–Johnson damping scheme (D3BJ) [36,37] and COSMO solvation (toluene).

After optimization of the H-atom positions of the molecular structures obtained by single-crystal X-ray diffraction analyses, NBO and NLMO calculations were performed using ORCA 5.0.2 [30] with the NBO7.0 package [39] using the restricted PBE0 functional with relativistically recontracted Karlsruhe basis sets ZORA-def2-TZVPP [31,32] (for H, C, N, O, P) and SARC-ZORA-TZVPP (for Ru) [33], the scalar relativistic ZORA Hamiltonian [34,35], atom-pairwise dispersion correction with the Becke–Johnson damping scheme (D3BJ) [36,37] and COSMO solvation (toluene). QTAIM (quantum theory of atoms-in-molecules) [40], WBI [41] and NCI [42] calculations were performed with MultiWFN [43] at the same level of theory as used for NBO analysis. NBO/NLMO graphics were generated using Chemcraft [44] and visualization of the NCI results was carried out with VMD [45].

3.2. Syntheses and Characterization

Compound 3a ([PhP(μ-pic)₂(μ-pyO)Ru(κ²-pyO)], C₂₈H₂₅N₄O₂PRu). A Schlenk flask was charged with magnetic stirring bar, [Ru(pyO)₂(nbd)] (1) (0.525 g, 1.50 mmol) and PhP(pic)₂ (2a) (0.440 g, 1.50 mmol), evacuated and set under Ar atmosphere prior to adding acetonitrile (10 mL). The resultant dispersion was heated and stirred under reflux. Within the first ten minutes of heating, the color of the dispersion changed from yellow to orange. Upon 3 h of heating, the mixture was allowed to attain room temperature. The solid product was filtered off, washed with acetonitrile (2 × 3 mL) and dried in vacuum. (Crystals suitable for X-ray diffraction analysis were grown from a dichloromethane solution of this product upon diffusion of diethyl ether via gas phase in the course of one week.) Yield: 0.589 g (1.01 mmol, 68%). Elemental analysis for C₂₈H₂₅N₄O₂PRu (581.56 g·mol⁻¹): C, 57.83%; H, 4.33%; N, 9.63%; found C, 57.64%; H, 4.33%; N, 9.32%. ¹H NMR (CD₂Cl₂): δ (ppm) 9.01 (d, 2H, pic-6, 5.3 Hz), 7.83 (br, 2H, pyO-6), 7.78 (ddd, 2H, Ph-ortho, 1.5 Hz, 7.6 Hz, 11.7 Hz), 7.42–7.48 (m, 3H, Ph-meta/para), 7.32 (t, 2H, pic-4, 7.6 Hz), 7.24 (d, 2H, pic-3, 7.6 Hz), 7.07 (m, 2H, pyO-4), 6.89 (“t”-shaped dd, 2H, pic-5, 5.3 Hz, 7.6 Hz), 5.95–6.05 (m, 4H, pyO-5,3), 3.96 (br, 2H, CH₂), 3.65 (dd, 2H, CH₂, 17.0 Hz, 10.8 Hz); ¹³C{¹H} NMR (CD₂Cl₂): δ (ppm) 173.7 (pyO-2), 164.8 (pic-2), 152.6 (pic-6), 150.6 (pyO-6), 137.4 (d, Ph-ipso, 60 Hz), 136.2 (pyO-4), 134.2 (pic-4), 130.6 (d, Ph-ortho, 11 Hz), 130.0 (d, Ph-para, 2 Hz), 128.6 (d, 11 Hz, Ph-meta), 122.1 (d, 11 Hz, pic-3), 121.7 (pic-5), 112.0 (pyO-3), 107.7 (pyO-5), 46.2 (CH₂); ³¹P{¹H} NMR (CD₂Cl₂): δ (ppm) 80.8; ³¹P CP/MAS NMR: δ_iso (ppm) 73.8, 76.3.

Compound 4a ([PhP(μ-pic)₂(μ-pyO)Ru(κ-O-pyO)(CO)], C₂₉H₂₅N₄O₃PRu) and compound 6a ([PhP(μ-pic)₂(μ-pyO)Ru(Cl)(CO)], C₂₄H₂₁ClN₄O₃PRu). A Schlenk flask (volume ca. 15 mL) charged with magnetic stirring bar, compound 3a (0.116 g, 0.199 mmol) and toluene (2.5 mL) was cooled in liquid N₂ prior to evacuating the initial atmosphere and recharging with CO atmosphere in 3 cycles. (The gas volume in the Schlenk flask (>10 mL) corresponds to excess CO (>0.45 mmol).) Then, the contents were allowed to attain room temperature, and the resultant orange dispersion was stirred at room temperature for two days. On the third day, the contents were stirred at 60 °C for 6 h (the contents remained a dispersion) and then stored at room temperature for 3 days. (Some crystals suitable for X-ray diffraction analysis were taken from the crude product.) Thereafter, the contents were separated from the supernatant by decantation, washed with toluene (1.5 mL) and briefly dried in vacuum. Yield: 0.05 g (0.07 mmol, ca. 35%) of 4a·(toluene). ³¹P NMR spectroscopy of both the solid and CD₂Cl₂ solution of this product indicated the presence of some starting material (contains ca. 15% 3a). Therefore, elemental analysis data are not reported. An attempt at recrystallizing crude 4a·(toluene) from dichloromethane (with gas phase diffusion of n-pentane) afforded some crystals of complex 6a. (The formation of 6a may originate from traces of HCl contained in dichloromethane. Nonetheless, even though pyridine itself is not sufficiently nucleophilic to undergo nucleophilic substitution of chloride from DCM under mild conditions [46], metal-bound pyridyl groups have been shown to undergo nucleophilic attack at DCM [47,48]. Thus, the latter path cannot be ruled out at the current stage. However, side-products of the formation of 6a were not identified.) Analogous synthesis of stirring a dispersion of 3a (0.095 g, 0.16 mmol) in acetonitrile (2 mL) under an atmosphere of CO afforded a clear solution within one day, and the solution remained clear for one week. Gas-phase diffusion of diethyl ether did not result in crystallization of the target product, and a ³¹P NMR spectrum recorded from this crude solution (products in MeCN/Et₂O) indicated the presence of both 4a and starting material 3a with signals at 68.9 and 80.2 ppm, respectively, at an intensity ratio of 2:1. The limited amount of sample available (especially in case of 6a) and decomposition/reaction with solvent (in case of 4a), 1D ¹³C{¹H} and 2D ¹³C NMR spectroscopy did not allow for detecting all ¹³C signals or assignment of all signals observed. Thus, the ¹³C NMR shifts reported here basically serve as fingerprints of 4a and 6a.

NMR data for 4a (recorded from the crude product of 4a·(toluene)): ¹H NMR (CD₂Cl₂): δ (ppm) 9.71 (dd, 1H, 5.7 Hz, 1.6 Hz), 8.82 (dd, 1H, 5.8 Hz, 1.6 Hz), 8.02–8.06 (m, 2H), 7.75 (m, 2H, Ph-ortho), 7.62 (m, 1H), 7.37–7.56 (m, 7H), 7.30–7.35 (m, 2H), 6.86–6.97 (m, 2H), 6.46 (ddd, 1H, 8.6 Hz, 5.0 Hz, 1.0 Hz), 5.86 (“dt”-like m, 1H, 6.4 Hz, 1.5 Hz, 1.0 Hz), 5.68 (dd, 1H, 8.6 Hz, 1.0 Hz), 4.79 (dd, 1H, CH₂, 17.4 Hz, 13.8 Hz), 3.87 (dd, 1H, CH₂, 16.7 Hz, 12.2 Hz), 3.64 (dd, 1H, CH₂, 17.4 Hz, 8.4 Hz), 3.50 (dd, 1H, CH₂, 16.7 Hz, 7.2 Hz); ¹³C{¹H} NMR (CD₂Cl₂): δ (ppm) 172.5, 169.9, 162.3, 153.0, 152.8, 150.0, 148.4, 137.7, 137.4 (2×), 136.7, 131.2 (d, 11 Hz), 131.0, 128.9 (d, 12 Hz), 123.0, 122.6, 115.2, 114.3, 110.5, 105.9, 41.5, 41.1; ³¹P{¹H} NMR (CD₂Cl₂): δ (ppm) 68.6; ³¹P CP/MAS NMR: δ_iso (ppm) 63.5, 68.1.

NMR data for 6a: ¹H NMR (CD₂Cl₂): δ (ppm) 9.74 (ddd, 1H, pic-6, 5.9 Hz, 1.6 Hz, 0.7 Hz), 9.67 (ddd, 1H, pic’-6, 5.8 Hz, 1.7 Hz, 0.8 Hz), 8.64 (dd, 1H, pyO-6, 6.4 Hz, 2.1 Hz), 7.40–7.70 (m, 8H, pic-5, pic’-5, pic’-3, Ph-ortho/-meta/-para), 7.31 (ddd, 1H, pic-3, 7.9 Hz, 1.2 Hz, 0.7 Hz), 7.17 (“tt”-like m, 1H, pic-4, 6.6 Hz), 7.10 (“tt”-like m, 1H, pic’-4, 7.5 Hz, 6.7 Hz), 6.95 (ddd, 1H, pyO-4, 8.6 Hz, 6.4 Hz, 2.1 Hz), 6.00 (dt, 1H, pyO-5, 2 × 6.4 Hz, 1.5 Hz), 5.64 (dddd, 1H, pyO-3, 8.6 Hz, 1.5 Hz, 2 × 0.5 Hz), 4.80 (dd, 1H, CH₂, 17.3 Hz, 14.6 Hz), 3.84 (dd, 1H, CH₂, 16.8 Hz, 12.8 Hz), 3.60 (dd, 1H, CH₂, 17.3 Hz, 7.3 Hz), 3.46 (dd, 1H, CH₂, 16.8 Hz, 7.2 Hz); ¹³C{¹H} NMR (CD₂Cl₂): δ (ppm) 155.3, 153.9, 150.7, 137.9, 137.1, 131.1, 130.3 (d, 11 Hz), 129.1 (d, 12 Hz), 123.8, 123.4, 122.7, 114.3, 106.1; ³¹P{¹H} NMR (CD₂Cl₂): δ (ppm) 66.7.

Compound 5a ([PhP(μ-pic)₂(μ-pyO)₂Ru(CNtBu)], C₃₃H₃₄N₅O₂PRu). To a solution of compound 3a (0.062 g, 0.107 mmol) in dichloromethane (2 mL), CNtBu (0.011 g, 0.132 mmol) was added, whereupon the initially deep orange solution changed color to lighter orange. Thereafter, the volume of the solution was reduced to 0.5 mL (by condensation of parts of the solvent into a cold trap under reduced pressure), and the flask with the crude solution of 5a was connected to a second flask with 3 mL of n-pentane (for gas phase diffusion). Within one day, some crystals of 5a·1.5(CH₂Cl₂) had formed. One of the crystals was used for single-crystal X-ray diffraction analysis, while the remaining few crystals were separated from the supernatant by decantation and briefly dried in vacuum. The ¹H NMR spectrum of this crude product (cf. Figure 6) indicated the presence of two isomers of 5a (5a¹ and 5a³), as well as of starting materials 3a and CNtBu. The latter two are likely to have formed from 5a in a dissociation reaction. Elemental analysis data correspond very well to the composition of 5a·(CH₂Cl₂): Elemental analysis for C₃₄H₃₆Cl₂N₅O₂PRu (749.63 g·mol⁻¹): C, 54.48%; H, 4.84%; N, 9.34%; found C, 54.55%; H, 5.14%; N, 9.57%. Therefore, for NMR spectroscopic characterization a solution of 5a (isomer 5a³ initially formed) in CD₂Cl₂ was prepared in situ and NMR spectra were recorded within few hours. ¹H signals were assigned by considering shift ranges (of pic from 3a and pyO from 5b) and coupling patterns. Caution: Some couplings were not resolved, meaning the coupling constants reported here represent the superposition of two couplings of similar frequency. ¹H NMR (CD₂Cl₂): δ (ppm) 9.25 (dd, 2H, pyO-6, 5.9 Hz, 1.1 Hz), 7.97 (m, 2 H, Ph-ortho), 7.50 (dd, 2H, pic-6, 6.2 Hz, 2.2 Hz), 7.40–7.45 (m, 3H, Ph-meta/para), 7.36 (tt, 2H, pic-4, 7.6 Hz, 1.1 Hz), 7.17 (d, 2H, pic-3, 7.6 Hz), 6.88 (ddd, 2H, pyO-4, 8.6 Hz, 6.5 Hz, 2.3 Hz), 6.83 (“dt-like” m, 2H, pic-5, 7.5 Hz, 6.3 Hz, 1.3 Hz), 5.81 (dd, 2H, pyO-3, 8.6 Hz, 1.5 Hz), 5.62 (m, 2H, pyO-5, 1.5 Hz), 4.03 (dd, 2H, CH₂, 16.9 Hz, 8.6 Hz), 3.52 (dd, 2H, CH₂, 16.9 Hz, 10.5 Hz); ¹³C{¹H} NMR (CD₂Cl₂): δ (ppm) 172.5, 164.3 (d, 8 Hz), 157.1, 154.7 (d, 4 Hz), 135.7, 135.1, 133.5 (d, 9 Hz), 129.9 (d, 2 Hz), 128.4 (d, 9 Hz), 122.8 (d, 10 Hz), 121.6, 114.8, 105.1, 57.6 (CMe₃), 42.2 (d, CH₂, 23 Hz), 31.2 (CH₃) (The signals of Ph-ipso-C and CNCtBu have not been detected.); ³¹P{¹H} NMR (CD₂Cl₂): δ (ppm) 40.1.

Compound 5b ([PhP(μ-pyO)₄Ru(CNtBu)], C₃₁H₃₀N₅O₄PRu). To a dispersion of compound 3b (0.113 g, 0.193 mmol) in toluene (7 mL), which was stirred at room temperature, CNtBu (0.0184 g, 0.22 mmol) was added. Thereafter, the dispersion was stirred at 90 °C for 2.5 h and then stored at room temperature overnight, whereupon the yellow solid was filtered off, washed with toluene (2 mL) and dried in vacuum. Yield: 0.068 g (0.102 mmol, 53%). (Some crystals suitable for X-ray diffraction analysis formed upon gas phase diffusion of n-pentane into the combined filtrate and washings.) Elemental analysis for C₃₁H₃₀N₅O₄PRu (668.64 g·mol⁻¹): C, 55.77%; H, 4.56%; N, 10.46%; found C, 55.67%; H, 4.52%; N, 10.47%. ¹H NMR (CD₂Cl₂): δ (ppm) 8.61 (dd, 4H, pyO-6, 6.0 Hz, 2.0 Hz), 8.45 (m, 2H, Ph-ortho), 7.35-7.45 (m, 3H, Ph-meta/para), 7.23 (ddd, 4H, pyO-4, 8.4 Hz, 6.9 Hz, 2.0 Hz), 6.51 (dd, 4H, pyO-3, 8.4 Hz, 0.9 Hz), 6.32 (ddd, 4H, pyO-5, 6.9 Hz, 6.0 Hz, 1.4 Hz), 1.90 (s, 9H, CH₃); ¹³C{¹H} NMR (CD₂Cl₂): δ (ppm) 165.7 (d, pyO-2, 5.4 Hz), 152.4 (d, Ph-ipso, 225 Hz), 152.3 (d, pyO-6, 3.9 Hz), 138.0 (pyO-4), 134.3 (d, Ph-ortho, 14 Hz), 127.8 (d, Ph-para, 4 Hz), 126.1 (d, Ph-meta, 17 Hz), 113.1 (pyO-5), 111.1 (d, pyO-3, 3 Hz), 57.7 (CMe₃), 31.7 (CH₃); ³¹P{¹H} NMR (CD₂Cl₂): δ (ppm) 28.0.

4. Conclusions

In this study, we have shown that 2-picolyl (pic) moieties may be employed as bridging entities between P and Ru atoms in complexes, in which a hemilabile Ru-bound ligand (in our case a chelating 2-pyridyloxy group, pyO) can undergo chelate opening and take advantage of stabilization of the dangling ligator function by binding to the adjacent P atom. Whereas the pic groups (in the starting material PhP(pic)₂ and in the Ru complexes resulting therefrom) imply the advantage of a more robust building block with respect to lowered hydrolytic sensitivity compared with related pyO-based systems (PhP(pyO)₂ and in the Ru complexes resulting therefrom), they lower the Lewis acidity of the P atom. This affects both the nature of the P–Ru bond (which features significantly lowered Ru→P π-back-bonding contributions) and the tendency for binding of the dangling ligand arm to the P atom. Thus, in PhP(pic)₂-based systems (3a, 4a, 5a), the latter is noticeably less pronounced than in the related PhP(pyO)₂-based Ru complexes (3b, 4b, 5b). This fosters reactions back toward formation of the Ru(κ²-pyO)-chelate (with release of monodentate ligands in equilibrium, such as Ru-bound isonitrile) or even Ru(κ²-pyO)-chelate opening with formation of Ru(κ-O-pyO)-complexes, which feature a dangling pyO nitrogen atom. The dangling N atom of the latter may cause unwanted side reactions (e.g., reaction of 4a and dichloromethane or traces of HCl contained therein with formation of 6a).

In summary, further exploration of related kinds of coordinative switches within Ru-P-systems may benefit from electronegative substituents at the P atom. In general, the subject matter of ligand migration from Ru to an adjacent P atom is worth exploring further. In a recent study, Tanushi and Radosevich showed the migration of an Ru-bound hydride to a special phosphane ligand III (which also bears pyridine anchors as Ru-binding site) with the formation of complex IV (Scheme 3) [49]. This hints at the greater potential of such systems for stabilizing monodentate ligands with an Ru-bound phosphane P atom.

From an academic point of view, the herein presented compound 5b represents a rare example of a monometallic phosphane complex with hexacoordination of both the transition metal and the phosphorus atom.