Metal-Ligand Role Reversal: Hydride Transfer Catalysis by a Functional Phosphorus Ligand with a Spectator Metal

Abstract

Hydride transfer catalysis is shown to be enabled by the nonspectator reactivity of a transition metal-bound low-symmetry tricoordinate phosphorus ligand. Complex 1•[Ru]⁺, comprising a nontrigonal phosphorus chelate (1, P(N(o-N(2-pyridyl)C₆H₄)₂) and an inert metal fragment ([Ru] = (Me₅C₅)Ru), reacts with NaBH₄ to give a metallohydridophosphorane (1^H•[Ru]) by P–H bond formation. Complex 1^H•[Ru] is revealed to be a potent hydride donor (ΔG°_H–,exp < 41 kcal/mol, ΔG°_H–,calc = 38 ± 2 kcal/mol in MeCN). Taken together, the reactivity of the 1•[Ru]⁺/1^H•[Ru] pair comprise a catalytic couple, enabling catalytic hydrodechlorination in which phosphorus is the sole reactive site of hydride transfer.

Graphical Abstract

The roles that ligands play in modulating the properties of organotransition metal complexes vary, especially as a function of their direct involvement in reactivity. Descriptors spanning a broad spectrum of reactivity from ancillary/spectator/innocent (A),^1–2,3,4 to electronically-coupled/noninnocent (B),^5–6,7,8 and participatory/cooperative (C)^9–10,11 are well-known (Figure 1). On the far extreme of this continuum, a ‘functional’ class of ligands (D) —in which bond making and breaking is localized on the ligand itself in conjunction with a stabilizing ‘spectator metal’—are comparatively less common.^{12–13,14,15,16,17}

Figure 1.

Ligand role in transition metal catalysis as a function of participation in reactivity. L = ligand, M = metal, X,Y= reactive substituents.

Although tricoordinate phosphorus (σ³-P) ligands are largely seen as archetypal spectator ligands within the foregoing continuum, the literature collects sporadic reference to reactions invoking their conversion to higher P-coordinate intermediates.¹⁸ In this vein, we have recently established that nontrigonal σ³-P compounds^{19–20,21,22} exhibit enhanced electrophilicity at P that enables access to stable higher coordinate (σ⁴-P)–M complexes by ‘nonspectator’ reactivity.^{23–24,25,26} Based on these findings, we hypothesized that a new opportunity might be realized for useful catalysis based on a “functional” σ³-P ligand if paired with an inert, coordinatively- and electronically-saturated metal guest. We report here the realization of this vision using a nontrigonal σ³-P-containing chelate in combination with a spectator (Me₅C₅)Ru⁺ fragment to drive hydride transfer reductions of organic acceptors and chloroalkanes without chemical or redox reactions at Ru. Instead, reactivity is solely localized at P. These results: (1) establish new precedent for σ³-P moieties as functional sites for ligand-catalyzed reactions, (2) validate higher-coordinate (σ⁴-P)–M metallophosphoranes as meritorious intermediates in catalysis, and (3) portend new opportunities for catalyst design by a role reversal of the metal-ligand construct, where reactive and functional ligands are tuned by decoration with spectator metal fragments.

Organometallic hydride transfer reactions usually access reactive metal sites for hydride formation,^27,28 but on the basis of the well-known kinetic inertness of coordinatively-saturated Ru(II) complexes, the (Me₅C₅)Ru⁺ fragment was selected as an intentionally passive chelating template for a nontrigonal σ³-P ligand. Specifically, addition of the tridentate nontrigonal σ³-P ligand 1 to [(Me₅C₅)Ru(MeCN)₃](PF₆) provided in high yield the orange N,P,N-chelated complex 1•[Ru]⁺ ([Ru]⁺= (Me₅C₅)Ru⁺) as its hexafluorophosphate salt (Scheme 1). Compared to free 1 (δ 141.7 ppm, Figure 2a), complexation results in a downfield shift of the ³¹P{¹H} NMR resonance for 1•[Ru]⁺ (δ 206.4 ppm, Figure 2b). In the solution phase, ¹H NMR spectra for complex 1•[Ru]⁺ display an apparent mirror symmetry for the two phenylene and two pyridyl moieties, respectively, along with splitting of the Cp* methyl peak as a doublet (⁴J_PH = 3.6 Hz), in line with the expected tridentate coordination of 1 to ruthenium. In the solid state, single-crystal X-ray diffraction confirms the local bonding environment about the coordinatively saturated, 18-electron Ru(II) center (Figure 3, left). As a function of the chelating structure, a short d(Ru–P) bond distance (2.1561(5) Å, Table 1) is found. With respect to the local geometry about phosphorus, coordination to Ru modestly perturbs the local nontrigonal structure, associated with an increase in the angle ∠N₂–P–N₄ to 121.6(5)° (Δ∠N₂–P–N₄ = +12.9° compared to 1). Given the ambient fluxional behavior of a related nontrigonal P(III) triamide by flexion of the ∠N₂–P–N₄ angle,²⁹ this deformation is not expected to be energetically costly.

Scheme 1.

Synthesis of 1•[Ru]⁺ and 1^H•[Ru].

Conditions: (i) [(Me₅C₅)Ru(MeCN)₃][PF₆], CH₂Cl₂, 25 °C, 16 h; 91% yield. (ii) NaBH₄, MeCN, 40 °C, 1 h; 65% yield.

Figure 2.

³¹P NMR Spectra for (a) Compound 1 in C₆D₆. (b) Compound 1•[Ru]⁺ in CD₃CN. (c) Compound 1^H•[Ru] in C₆D₆.

Figure 3.

Structural representations of 1•[Ru]⁺ (left) and 1^H•[Ru] (right) from single crystal X-ray diffraction. Ellipsoids are set at 50% occupancy. All hydrogen atoms (except those attached to P), counter anions, and additional molecules in the asymmetric unit are omitted for clarity. See the SI Section VIII for crystallographic details.

Table 1.

Selected bond distances (Å), angles (deg), and ³¹P NMR chemical shifts (ppm) for 1•[Ru]⁺ and 1^H•[Ru].

metric	1	1•[Ru]+	1H•[Ru]
Ru–P (Å)	--	2.1561(5)	2.229(3)
Ru–N1 (Å)	--	2.106(7)	2.154(9)
P–N3 (Å)	1.7485(8)	1.7071(19)	1.753(9)
P–N2 (Å)	1.7341(8)	1.741(7)	1.882(9)
∠N1–Ru–N5 (°)	--	93.0(5)	103.7(3)
∠N1–Ru–P (°)	--	80.9(2)	76.6(3)
∠N2–P–N4 (°)	108.67(4)	121.6(5)	168.5(4)
31P NMR δ (ppm)	141.7	206.4	19.9

Despite the coordinative and electronic saturation of the ruthenium center in 1•[Ru]⁺, an orange acetonitrile solution of 1•[Ru]⁺ readily reacts with NaBH₄ to give a red suspension within 1 h at 40 °C. ³¹P{¹H} NMR spectroscopy of the isolated red solid revealed the absence of downfield resonances in the vicinity of the starting material, while a single upfield peak (δ 19.9 ppm) that evolves into a doublet in the proton-coupled ³¹P NMR spectrum (J = 469 Hz, Figure 2C) was observed. The corresponding coupling partner is identified in the ¹H NMR spectrum as a doublet (J = 469 Hz) centered at δ 8.38 ppm. Both the chemical shift and coupling magnitude are inconsistent with metal-based hydride.^30,31 Instead, these data indicate the formation of the metallohydridophosphorane complex 1^H•[Ru], containing a higher-coordinate phosphorus moiety resulting from P–H bond formation.

Indeed, X-ray diffraction on crystalline 1^H•[Ru] confirmed the presence of a 5-coordinate phosphorus center of intermediate square pyramidal/trigonal bipyramidal geometry (τ₅ = 0.53)³² bearing a hydride with anti configuration with respect to the Me₅C₅ fragment about the Ru–P vector (Figure 3, right). The addition of the hydride results in significant local geometric changes at P for 1^H•[Ru] as compared to 1•[Ru]⁺. First, the Ru–P bond distance is significantly elongated (d(Ru–P) = 2.229(3) Å, Table 1), in line with previously observed trends in metal-phosphorus bond distances between (σ³-P)–M and (σ⁴-P)–M complexes.^33,34 Second, the angle ∠N₂–P–N₄ for 1^H•[Ru] is increased significantly to 168.5(4)° (Δ∠N₂–P–N₄ = +46.9° compared to 1•[Ru]⁺) to accommodate the hydride substituent. A further consequence of this distortion is a substantial differential in the P–N distances, where d(P–N₂)= 1.882(9) Å but d(P–N₃) = 1.753(9) Å. This effect can be rationalized by reference to the Pimentel–Rundle model of 3c-4e bonding³⁵ for the nearly linear N₂–P–N₄ triad.

DFT models at the M06/def2-TZVP level of theory for 1•[Ru]⁺ and 1^H•[Ru] reproduce well the experimental structures. Bonding analysis within the NBO formalism³⁶ reveals changes to the nature of the Ru–P σ-interactions as a function of hydride binding. A natural localized molecular orbital (NLMO) corresponding to a dative covalent P→Ru σ-interaction for 1•[Ru]⁺ shows modest polarization toward the phosphorus (P 62.0%/Ru 34.5%); the NLMO corresponding to the P–Ru bonding interaction in 1^H•[Ru] indicates an increased degree of covalency (P 55.9%/Ru 41.8%). Further analysis of the electronic models also provide a rationale for the observed reactivity. Inspection of the Kohn-Sham orbitals of 1•[Ru]⁺ (Figure S52) reveals a low-lying unfilled orbital strongly localized on phosphorus and projecting into the concave space within the chelate that serves as the binding site for hydride to form 1^H•[Ru]. Once formed, the P–H bond is described by a NLMO with leading contribution from H (P 47.2%/H 51.6%). Indeed, by natural population analysis (NPA), the P-H hydrogen is assigned more negative natural charge (−0.09) than any other hydrogen in the complex (+0.20), indicating hydridic character.

To test the hydridic reactivity^28,37–38 of the metallophosphorane P–H bond, complex 1^H•[Ru] was first treated with 1,3-dimethylbenzimidazolium (2⁺), a known hydride acceptor (ΔG°_H– = 45 kcal/mol³⁹ for conjugate 2^H, Scheme 2A). Over the course of the reaction, complete consumption of 1^H•[Ru] was observed with concomitant formation of 1•[Ru]⁺ and dihydrobenzimidazole 2^H. Even the reaction of 1^H•[Ru] with 1,3-dimethylpyridinum (3⁺, ΔG°_H– = 41 kcal/mol⁴⁰)—among the weakest tabulated organic hydride acceptors—proceeded cleanly and completely to give 1•[Ru]⁺ and dihydropyridine 3^H (Scheme 2A). The failure to identify a hydride transfer partner that would permit measurement of a dynamic equilibrium establishes 1^H•[Ru] as a potent hydride donor with a limiting thermodynamic hydricity of ΔG°_H– ≤ 41 kcal/mol (Scheme 2B).⁴¹

Scheme 2.

(A) Hydride transfer reactions of 1^H•[Ru] with organic acceptors. (B) Experimental and computed thermodynamic hydricity of 1^H•[Ru].

To more accurately measure the hydricity of 1^H•[Ru], hydride transfer equilibria with transition-metal based acceptors was also explored. However, these reactions were prone to competing side reactions over the course of of the long requisite reaction times (see SI Section V). Similarly, our attempts to measure hydricity via H₂ heterolysis or pK_a-potential methods were hindered by either by a lack of reactivity with H₂ or irreversible reduction features observed in cyclic voltammograms of 1•[Ru]⁺.²⁸ Therefore, to more precisely quantify the hydricity of 1^H•[Ru], hydride transfer reactions between 1^H•[Ru] and 2⁺ or 3⁺ were calculated by DFT at the M06/def2-TZVP level (Scheme 2). The computed hydricity value for 1^H•[Ru] inferred from these calculations (ΔG°_H–,calc = 38 ± 2 kcal/mol) is fully consistent with the experimental observations and marks 1^H•[Ru] as the strongest ‘ligand-centered’ hydride donor yet quantified by almost 20 kcal/mol.^{42–43,44,45,46,47,48} Distortion-interaction analysis provides a partial rationalization for the strongly hydridic nature of 1^H•[Ru] relative to prior ligand-centered hydrides; 31 kcal/mol of strain release is computed for relaxation to 1•[Ru]⁺ following a non-adiabatic hydride transfer from 1^H•[Ru].

The strongly hydridic nature of 1^H•[Ru] enables hydride transfer reactions to other weak electrophiles including chloroalkanes (Figure 4A). Representatively, dissolution of 1^H•[Ru] in CDCl₃ in a Teflon-sealed NMR tube results in a red-to-orange color change within minutes, after which complete consumption of 1^H•[Ru] was observed. Both 1•[Ru]⁺ and CDHCl₂ (δ 5.28 ppm, Figure S41) were identified as the sole products. Similarly, mild heating (50 °C, 16 h) of a solution of 1^H•[Ru] in CD₂Cl₂ led to formation of 1•[Ru]⁺ along with CD₂HCl (δ 2.98 ppm, Figure S43). In both instances, coordination of the displaced Cl⁻ ion to the conjugate hydride acceptor 1•[Ru]⁺ is not observed, in contrast to both transition metal and diazaphospholene hydrodechlorination reactions.^{49–50,51,52} Indeed, dissolution of 1•[Ru]⁺ in a saturated solution of [nBu₄N][Cl] in CD₃CN (~35 equiv, Figures S45–46) showed no evidence of Cl⁻ binding by NMR spectroscopy.

Figure 4.

(A) Stoichiometric hydrodechlorination of chloroalkanes using 1^H•[Ru]. (B) Catalytic hydrodechlorination of CDCl₃ using 1•[Ru]⁺ and [nBu₄][BH₄]. py = pyridine.

The combined reactivities of 1•[Ru]⁺ as a hydride acceptor from borohydride and 1^H•[Ru] as a hydride donor to organic substrates suggested the possibility of employing this complex as a catalyst for hydride transfer (Figure 4B). Whereas the direct reaction of the soluble hydride donor [nBu₄N][BH₄] with bulk CDCl₃ for 48 h at 25 °C proceeds only to minimal conversion (<5% yield, see SI Table S1), the addition of 1 mol% of 1•[Ru]⁺ to a CDCl₃ solution containing [nBu₄][BH₄] (0.27 mM) and pyridine (0.54 mM, as a BH₃ scavenger) gave unambiguous evidence for formation of CDHCl₂ (¹H NMR δ 5.11 ppm, J=1 Hz) with a turnover number of 31 ± 1 (average of two runs). Consistent with the qualitative rates of formation and consumption of 1•[Ru]⁺ and 1^H•[Ru] from stoichiometric studies, in situ ³¹P NMR spectra implicate 1•[Ru]⁺ as the resting state of the catalytic cycle, and imply that future designs that increase the propensity of 1•[Ru]⁺ to accept hydride at P might further enhance the rate of catalysis.

These results demonstrate the capacity for nontrigonal σ³-P ligands to support hydride transfer catalysis via a functional mode of reactivity that transits through a higher-coordinate (σ⁴-P)–M metallohydridophosphorane. In these complexes, the metal is relegated to a peripheral position with respect to bond making and breaking. Indeed, since the reaction chemistry occurs exclusively at P and not Ru, it might be more apt to consider 1•[Ru]⁺ an electrophilic phosphonium ion⁵³ bearing an ancillary (Me₅C₅)Ru substituent. From this point of view, the apparent reversal in roles for a metal-phosphorus complex demonstrated by the present results suggests fresh opportunities for the deliberate design of novel main group catalysts that incorporate spectator metal fragments.