Activation of H-H, HO-H, C(sp2)-H, C(sp3)-H, and RO-H Bonds by Transition-Metal Frustrated Lewis Pairs Based on M/N (M = Rh, Ir) Couples.

Reaction of the dimers [(Cp*MCl)2(μ-Cl)2] (Cp* = η5-C5Me5) with Ph2PCH2CH2NC(NH(p-Tolyl))2 (H2L) in the presence of NaSbF6 affords the chlorido complexes [Cp*MCl(κ2N,P-H2L)][SbF6] (M = Rh, 1; Ir, 2). Upon treatment with aqueous NaOH, solutions of 1 and 2 yield the corresponding complexes [Cp*M(κ3N,N',P-HL)][SbF6] (M = Rh, 3; Ir, 4) in which the ligand HL presents a fac κ3N,N',P coordination mode. Treatment of THF solutions of complexes 3 and 4 with hydrogen gas, at room temperature, results in the formation of the metal hydrido-complexes [Cp*MH(κ2N,P-H2L)][SbF6] (M = Rh, 5; Ir, 6) in which the N(p-Tolyl) group has been protonated. Complexes 3 and 4 react with deuterated water in a reversible fashion resulting in the gradual deuteration of the Cp* group. Heating at 383 K THF/H2O solutions of the complexes 3 and 4 affords the orthometalated complexes [Cp*M(κ3C,N,P-H2L-H)][SbF6] [M = Rh, 7; Ir, 8, H2L-H = Ph2PCH2CH2NC(NH(p-Tolyl))(NH(4-C6H3Me))], respectively. At 333 K, complexes 3 and 4 react in THF with methanol, primary alcohols, or 2-propanol giving the metal-hydrido complexes 5 and 6, respectively. The reaction involves the acceptorless dehydrogenation of the alcohols at a relatively low temperature, without the assistance of an external base. The new complexes have been characterized by the usual analytical and spectroscopic methods including the X-ray diffraction determination of the crystal structures of complexes 1-5, 7, and 8. Notably, the chlorido complexes 1 and 2 crystallize both as enantiopure conglomerates and as racemates. Reaction mechanisms are proposed based on stoichiometric reactions, nuclear magnetic resonance studies, and X-ray crystallography as well as density functional theory calculations.

Introduction

In 2006, Stephan’s group reported that the phosphano-borane compound (C6H2Me3)2P(C6F4)B(C6F5)2 reacted reversibly with molecular hydrogen to give the phosphonium-borate species (C6H2Me3)2PH(C6F4)BH(C6F5)2. This reaction demonstrates that compounds of representative elements are capable of activating the dihydrogen molecule breaking the paradigm that hydrogen activation is an exclusive ability of transition-metal compounds.[1] This novel reactivity is based on the cooperative behavior of an acidic (electron acceptor, boron) and a basic (electron donor, phosphorus) component that cannot form dative bonds due to geometry constrains. To highlight this feature, the term “frustrated Lewis pair” (FLP) was coined.[2]

Shortly afterward, the assortment of acidic and basic components was significantly expanded, and it was demonstrated that the resulting FLPs were capable of activating a variety of substrates including imines, olefins, alkynes, organic carbonyl compounds, carbon dioxide, azides, or nitric oxide. Subsequently, the FLP chemistry advanced by incorporating unusual stoichiometric reactions as well as catalytic processes such as hydrogenation (including enantioselective hydrogenation), hydrosilylation, hydroboration, or hydroamination.[3]

In addition, the potential of FLP systems increased considerably with the proposal of Wass’s group to incorporate components based on transition metals in their design, resulting in the so-called transition-metal frustrated Lewis pairs (TMFLPs).[4] The incorporation of transition-metal fragments into FLP systems increases their structural and electronic diversity in such a way that it should allow them to efficiently promote the whole set of elementary reactions characteristic of catalytic processes. In this regard, Wass[5] and Erker’s[6] groups developed extensively the FLP chemistry of Zr/P systems and demonstrated their potential in the activation of small molecules as well as in catalysis. The area was quickly extended to new TMFLPs with various transition metals including bimetallic FLPs.[7] In this context, it should be noted that the reactivity of TMFLP species can be framed in the broader field of metal–ligand cooperation.[8]

A further qualitative leap in the area of FLP systems occurred when it was discovered that some combinations of Lewis acids and Lewis bases exhibited FLP reactivity despite the fact that the formation of the corresponding classical Lewis adduct (CLA) was observed.[9] In this regard, it was established that in order for the system to exhibit FLP behavior it is enough that an equilibrium exists between the CLA form and the dissociated form, that is, that the dissociated form is thermally accessible.[10] To describe this type of system, the concept of “thermally induced frustration” was introduced[11] and the terms “masked”[12] and “dormant”[13] have been used to refer to the involved FLPs.

The activation of the O–H bond of water is one of the steps in the search for efficient catalysts for water splitting on the route to renewable energy generation.[14] Among the strategies employed to this end, metal–ligand cooperative chemistry[14g,15] and FPLs, based on both representative elements[16] and transition-metal components,[7d,7] have been successfully applied. On the other hand, the Cp* ligand forms robust complexes with a large variety of elements of the periodic table and, usually, it is a nonreactive ligand. However, rare examples of cooperative metal–ligand reactivity involving this ligand have been reported. Indeed, hydrogen abstraction from Cp* methyls has been accomplished either by treatment with an external strong base[17] or through an intramolecular pathway by means of a basic ligand.[18] The C–H bond cleavage usually leads to tetramethylfulvene complexes in which the fulvene moiety may display distinct coordination modes.[17b−17d] When this activation was coupled with the activation of the O–D bond of deuterated water, in some instances, a very unusual H/D exchange of the Cp* methyl protons was observed.[17e,18]

The field of metal–ligand cooperation also includes some of the acceptorless alcohol dehydrogenation (AAD) processes catalyzed by metallic compounds. AAD is a dehydrogenative oxidation process with important applications in energy, green chemistry, and organic synthetic methods. Successful cases of AAD include the use of a variety of transition-metal complexes containing chelates, pincers, and related multidentate ligands as catalysts.[19] Some of the ligands possess a basic site able to abstract a proton from the alcohol, and the resulting alkoxide transfers a hydride from the α-CH position to the metal directly or via β-elimination.[19g,20]

With these concerns in mind, in the present article, we report: (i) the preparation and characterization of the masked TMFLP compounds [Cp*M(κ3N,N′,P-HL)][SbF6] (Cp* = η5-C5Me5; H = N,N′-bis(p-Tolyl)-N″-(2-diphenylphosphanoethyl)guanidine; M = Rh, 3; Ir, 4; Chart ); (ii) the reactivity of these complexes with H2 and H2O; (iii) the hydrogen abstraction from Cp* methyls in complexes 3 and 4 that results in an H/D gradual exchange when deuterated reagents were employed; (iv) the orthometalation reaction of one p-Tolyl ring of the phosphano-guanidine ligand, and (v) the acceptorless dehydrogenation of alcohols promoted by 3 and 4.

Chart 1

The Ligand H and the Complexes [Cp*M(κ3N,N′,P-HL)][SbF6]

Part of this work has been previously communicated.[21] Herein, we extend our study to the iridium homologue complex 4. Moreover, the reaction of complexes 3 and 4 with alcohols as well as orthometalation reactions, involving new C(sp3)–H, O–H and C(sp2)–H activations, is also included.

Results and Discussion

Preparation of the Complexes [Cp*MCl(κ2N,P-H2L)][SbF6] (M = Rh, 1; Ir, 2)

Reaction of the dimers [(Cp*MCl)2(μ-Cl)2][22] with the phosphano-guanidine compound H(21) in the presence of NaSbF6 affords the chlorido complexes [Cp*MCl(κ2N,P-H)][SbF6] (M = Rh, 1; Ir, 2; eq ).

Compounds 1 and 2 were characterized by analytical and spectroscopic methods (see the Supporting Information) and by the X-ray determination of their crystal structures. The κ2N,P coordination of the H ligand renders the metal atom a stereogenic center. Consequently, the methylene protons of the phosphano-guanidine ligand are diastereotopic, and in the proton nuclear magnetic resonance (1H NMR) spectrum, they give four distinct resonances. Broad bands in the 3000–3400 cm–1 region of the IR spectra together with 1H NMR singlets at 9.24 and 7.34 ppm (1) and 8.93 and 7.16 ppm (2) are indicative of the presence of two nonequivalent NH groups in the molecule. The 31P{1H} NMR spectrum consists of a doublet centered at 51.37 ppm for the rhodium complex [J(RhP) = 142.4 Hz] and a singlet at 26.52 ppm for the iridium compound, proving the coordination of the phosphorus to the metal (δP free ligand: −21.14 ppm).

Slow evaporation of saturated solutions of 1 and 2 in CH2Cl2/Et2O/n-pentane mixtures gave rise to the simultaneous formation of single crystals of pure enantiomers (conglomerates[23]) and racemates, for both compounds. Enantiopure samples of 1 and 2 slowly racemize in solution. Thus, for example, starting from a dichloromethane solution of pure SRh-1, SRh-1/RRh-1 molar ratios of about 92/8 and 74/26 were measured, by circular dichroism (CD) spectroscopy, after 2 and 18 h at room temperature, respectively.

A view of the cation of both enantiomers of the rhodium complex 1 is depicted in Figure . Views of the rac-1 cations as well as of the cation of the iridium enantiomers RIr-2, SIr-2 and racemate rac-2 are included in the Supporting Information. Figure shows the enantiomorphic relationship of the CD spectra of the two enantiomers of the rhodium complex 1.[24]

Figure 1

View of the cations of the RRh (A) and SRh (B) enantiomers of the rhodium complex [Cp*RhCl(κ2N,P-H)][SbF6] (1). For clarity, hydrogen atoms (except those bonded to nitrogen atoms) have been omitted.

Figure 2

CD spectra of SRh-1 (blue) and RRh-1 (red) in CH2Cl2.

Table S1 (Supporting Information) collects the most relevant structural parameters for the cations of RRh-1, rac-1, RIr-2, SIr-2, and rac-2, comparable structural parameters being observed regardless of the configuration and the nature of the metal center. Hence, only the structural parameters found in the SRh-1 isomer will be discussed. Selected bond lengths and angles of the cation of SRh-1 are summarized in Table . The cation of this complex exhibits “three-legged piano-stool” geometry. An η5-C5Me5 group occupies three fac positions, and the κ2N,P chelating phosphano-guanidine ligand and a chlorine atom complete the coordination sphere of the metal. The absolute configuration of the metal center is S, according to the atom priority sequence η-C5Me5 > Cl > P > N.[25] The structural parameters of the CN3 guanidine moiety deserve some comments. The C–NH(p-Tolyl) bond distances, N(2)–C(25) 1.344(8), N(3)–C(25) 1.369(7) Å, indicate a slight partial double bond character for these bonds,[26] while the N(1)–C(25) bond distance, involving the nitrogen coordinated to the metal atom, is found to be comparatively shorter, 1.310(7) Å, but also longer than typical N=C bond lengths (1.279(8) Å).[26] The sum of the bond angles at the coordinated nitrogen is 359.6(7)° indicating that the C(12)N(1)Rh(1)C(25) fragment is essentially planar. Hydrogen bonds between the N(2)–H(2) proton and the chlorido ligand [N–H = 0.82(7) Å, H···Cl = 2.45(7) Å, N···Cl = 3.203(5) Å, N–H···Cl = 153(7)°] and between the N(3)–H(3) proton and one of the fluorine atoms of the SbF6 anion [N–H = 0.82(8) Å, H···F = 2.13(8) Å, N···F = 2.939(7) Å, N–H···F = 169(7)°] were observed (Figure ).

Table 1

Selected Bonds Lengths (Å) and Angles (°) for Complex S-1

Rh–Cl	2.4199(14)	Cl-Rh–Cta	121.8(1)
Rh–P	2.2916(15)	P–Rh–N(1)	83.31(13)
Rh–N(1)	2.123(5)	P–Rh–Cta	130.4(2)
Rh–Cta	1.8218(1)	N(1)–Rh–Cta	131.2(2)
N(1)–C(25)	1.310(7)	Rh–N(1)–C(12)	118.3(3)
N(2)–C(25)	1.344(8)	Rh–N(1)–C(25)	122.8(4)
N(3)–C(25)	1.369(7)	C(12)–N(1)–C(25)	118.5(5)
Cl–Rh–P	90.54(5)	Σ°N(1)b	359.6(7)
Cl–Rh–N(1)	85.15(14)

Ct represents the centroid of the η5-C5Me5 ligand.

Σ°N(1) is the sum of bond angles around N(1) atom.

Figure 3

H-bond interactions in complex SRh-1. For clarity, only hydrogen atoms of N–H fragments have been depicted.

Preparation of the Complexes [Cp*M(κ3N,N′,P-HL)][SbF6] (M = Rh, 3; Ir, 4)

Solutions of 1 and 2 in 1:1 (v/v) THF/toluene were treated with aqueous NaOH for 1.5 h affording the corresponding complexes [Cp*M(κ3N,N′,P-HL)][SbF6] (M = Rh, 3; Ir, 4) through base-induced elimination of HCl and subsequent coordination of the deprotonated nitrogen (eq ).

The compounds were characterized by analytical and spectroscopic methods (see the Supporting Information) and by the X-ray diffraction determination of their crystal structures. A weak IR band at 3377 and 3362 cm–1 for 3 and 4, respectively, and a broad singlet in the proton NMR spectrum at 7.89 (3) and 8.00 ppm (4) are attributed to the NH functionality. As a consequence of the stereogenicity at the metal, the PCH2CH2N methylene protons are asynchronous and give rise to four resonances at the expected chemical shifts and with the awaited multiplicities (see the Supporting Information). A doublet centered at 48.27 ppm [J(RhP) = 159.0 Hz] and a singlet at 27.75 ppm in the 31P{1H} NMR spectrum are assigned to the phosphorus nucleus of the PPh2 group of the phosphano-guanidine ligand.

The molecular structure of 3 and 4 has been determined by X-ray diffraction means. There is no significant chemical difference to be remarked when comparing the structural parameters of the cations of the two complexes. For more detailed data about the molecular structure of the rhodium complex 3, see ref (21). Figure shows a view of the two crystallographically independent, but chemically equivalent, cations (A and B) found in the asymmetric unit of iridium complex 4. Table collects selected bond lengths and angles of both cations. The molecular structure reveals that the ligand HL presents a fac κ3N,N′,P coordination mode. This type of coordination renders the metal and the central nitrogen atom of the ligand stereogenic. The configuration at metal induces the configuration at nitrogen in such a way that only the RM,SN diastereomer and its SM,RN enantiomer form, both of them being present in the centrosymmetric unit cell of 4·C. In Figure a view of the two independent cations of the RM,SN diastereomer is depicted.

Figure 4

View of the two independent molecules of the cation of complex RM,SN-4. For clarity, only the ipso carbon of the phenyl rings of the PPh2 group is shown, and hydrogen atoms (except the NH proton) have been omitted.

Table 2

Selected Bond Lengths (Å) and Angles (°) for the Two Independent Cations of Complex RM,SN-4

	cation A		cation B
Ir(1)–P(1)	2.2984(12)	Ir(51)–P(51)	2.2843(12)
Ir(1)–N(1)	2.110(4)	Ir(51)–N(51)	2.124(4)
Ir(1)–N(2)	2.121(4)	Ir(51)–N(52)	2.130(4)
Ir(1)–Cta	1.8276(1)	Ir(51)–Cta	1.8338(1)
N(1)–C(25)	1.357(6)	N(51)–C(75)	1.362(6)
N(2)–C(25)	1.329(5)	N(52)–C(75)	1.324(5)
N(3)–C(25)	1.364(6)	N(53)–C(75)	1.361(6)
P(1)–Ir(1)–N(1)	80.02(11)	P(51)–Ir(51)–N(51)	80.31(11)
P(1)–Ir(1)–N(2)	90.53(11)	P(51)–Ir(51)–N(52)	91.49(11)
P(1)–Ir(1)–Cta	134.64(1)	P(51)–Ir(51)–Cta	133.98(1)
N(1)–Ir(1)–N(2)	62.33(14)	N(51)–Ir(51)–N(52)	62.03(14)
N(1)–Ir(1)–Cta	131.10(1)	N(51)–Ir(51)–Cta	133.21(1)
N(2)–Ir(1)–Cta	130.92(1)	N(52)–Ir(51)–Cta	129.55(1)
Ir(1)–N(1)–C(24)	118.3(3)	Ir(51)–N(51)–C(74)	118.5(3)
Ir(1)–N(1)–C(25)	93.9(3)	Ir(51)–N(51)–C(75)	93.8(3)
C(25)–N(1)–C(24)	116.6(4)	C(75)–N(51)–C(74)	114.9(4)
Σ°N(1)b	328.8(6)	Σ°N(51)b	327.2(6)
Ir(1)–N(2)–C(25)	94.3(3)	Ir(51)–N(52)–C(75)	94.7(3)
N(1)–C(25)–N(2)	109.2(4)	N(51)–C(75)–N(52)	109.4(4)

Ct represents the centroid of the η5-C5Me5 ligand.

Σ°N(1) and Σ°N(51) stand for the sum of bond angles around N(1) and N(51) atoms, respectively.

Focusing the discussion on cation A, the fac κ3N,N′,P coordination mode of the HL ligand forces the central N(1) atom to adopt a pyramidal geometry [Σ°N(1) = 328.8(6)°]. This geometry together with the N(1)–C(25) bond length [1.357(6) Å] contrasts with the structural features of the corresponding nitrogen atom in the precursor complex 2 where the H ligand coordinates in a chelate κ2N,P manner (for the corresponding parameters of compound 2, see Table S1, Supporting Information). Remarkably, the bond angles N(1)–Ir(1)–N(2) and N(1)–C(25)–N(2), 62.33(14) and 109.2(4)°, respectively, are far from the hybridization ideal values. All these features will most likely lead to a strong strain within the Ir–N–C–N four-membered metalacycle.

Reaction of Complexes 3 and 4 with Molecular Hydrogen

The structural parameters found in compounds 3 and 4, and in particular the envisaged strain within the four-membered metalacycle M–N–C–N led us to hypothesize that these compounds could behave like masked FLPs: the heterolytic cleavage of one of its M–N bonds could generate a TMFLP in which the metal and the nitrogen would play the role of the acid and basic center, respectively. As reported for compound 3,[21] these assumptions prompted us to try the reaction of complex 4 with molecular hydrogen.

Indeed, treatment of THF solutions of complexes 3 and 4 with hydrogen gas (4 bar), at room temperature, resulted in the formation of the metal hydrido-complexes [Cp*MH(κ2N,P-H)][SbF6] (M = Rh, 5; Ir, 6) in which the N(p-Tolyl) group has been protonated (eq ). Formally, the heterolytic breakage of the molecule of hydrogen gives rise to hydridic M–H and protic N–H bonds. Complete conversion to complex 5 was obtained after 4 h of reaction under the above mentioned conditions. Conversion to the iridium complex 6 was complete after 24 h at 373 K. The reaction is reversible but to achieve appreciable dehydrogenation rates it is necessary to heat THF solutions above 373 K. Indeed, heating at 393 K a solution of the hydride 5 for 30 min, a conversion of 30% to the rhodium compound 3 was observed. After heating at the same temperature a solution of the iridium complex 6 for 2.5 h, a conversion of 50% to the dehydrogenated compound 4 was measured.

A doublet of doublets centered at −10.79 ppm [J(PH) = 38.8 Hz, J(RhH) 22.6 Hz] for complex 5 and a doublet centered at −10.16 ppm [J(PH) = 32.3 Hz] for complex 6 are attributed to the M–H functionality in the cations. The presence of two peaks attributed to NH protons (see the Supporting Information) is indicative of the protonation of the N(p-Tolyl) group. The 31P{1H} NMR spectrum consists of a doublet centered at 61.77 ppm [J(RhP) = 143.9 Hz] for the rhodium complex and a singlet at 27.88 ppm for the iridium one.

The molecular structure of complex 5 corroborates all these features.[21] The compound crystallizes as a racemate in the P2 space group of the monoclinic system with one solvent molecule in the asymmetric unit (5·CD). The RRh enantiomer is depicted in Figure . Selected bond lengths and angles are shown in Table . The phosphano-guanidine ligand displays a κ2N,P coordination mode. A Cp* ligand, formally occupying three coordination sites, and a hydrido ligand [Rh–H = 1.56(5) Å] complete the coordination sphere of the metal. The observed RhH···HN (2) separation, 2.20(7) Å, is shorter than twice the hydrogen Van der Waals radius, 2.4 Å, indicating a significant H···H interaction between the protic NH and hydridic RhH functionalities. The structural parameters of the CN3 guanidino fragment are comparable to those found for the chlorido compound 1, that is, a greater double bond character for the CN bond involving the nitrogen atom coordinated to the metal [N(1)–C(25) 1.309(6) Å] when compared with the remaining CN bonds [N(2)–C(25) 1.359(6) Å, N(3)–C(25) 1.369(6) Å] and a planar geometry at the N(1) atom [Σ°N(1) = 358.9(6)°].

Figure 5

Molecular structure of the cation of complex 5·CD. For clarity, hydrogen atoms (except the Rh–H and N–H protons) have been omitted.

Table 3

Selected Bond Lengths (Å) and Angles (°) for the Cation of Complex 5·CD

Rh–P	2.2419(12)	P–Rh–H	82(2)
Rh–N(1)	2.110(4)	N(1)–Rh–Cta	129.39(1)
Rh–Cta	1.8697(1)	N(1)–Rh–H	88(2)
Rh–H	1.56(5)	Cta–Rh–H	124
N(1)–C(25)	1.309(6)	Rh–N(1)–C(12)	115.9(3)
N(2)–C(25)	1.359(6)	Rh–N(1)–C(25)	124.7(3)
N(3)–C(25)	1.369(6)	C(12)–N(1)–C(25)	118.3(4)
P–Rh–N(1)	82.85(11)	Σ°N(1)b	358.9(6)
P–Rh–Cta	134.74(1)

Ct represents the centroid of the η5-C5Me5 ligand.

Σ°N(1) stands for the sum of bond angles around N(1) atom.

Probably, the structural relaxation within the four-membered Ir–N–C–N metalacycle facilitates the reaction from 3 to 5 as well as from 4 to 6, which, in turn, results in the change in the coordination mode of the phosphano-guanidine ligand from κ3N,N′,P to κ2N,P with the concomitant change of the geometry at the N(1) atom from pyramidal to planar.

Water Activation by Complexes 3 and 4

As previously reported,[21] the rhodium complex 3 reacts with deuterated water in a reversible fashion resulting in the gradual deuteration of the Cp* group. At 293 K, 1H NMR measurements and mass spectrometry analysis show that deuteration of this group is complete after 15 h in [D8]THF/D2O (78%/22%, v/v) solution. Also, deuteration was evidenced by the determination of the crystal structure of 3-d15 by low-temperature single crystal neutron-diffraction experiments.[21] During the deuteration process, only isotopologues of compound 3 at different degrees of deuteration are detected by NMR spectroscopy.

Kinetic measurements indicate that the deuteration process obeys a pseudo-first-order rate law with kobs values from 3.31 × 10–6 to 4.99 × 10–4 s–1, in the 298–333 K temperature range. The formation of 3-d15 from 3 is reversible, and at 313 K, a [D8]THF/H2O (78%/22%, v/v) solution of 3-d15 evolves to 3 with an observed pseudo-first-order rate constant of 3.89 × 10–5 s–1. The measured ratio kH/kD (2.44) indicates that the rate-determining step for the exchange process is the C–H(D) bond cleavage.[21]

Based on density functional theory (DFT) calculations, we previously reported[21] that the H/D exchange relies on the activation of the water O–H bond at I rendering the key hydroxo intermediate II (Scheme ). II ultimately promotes the reversible hydrogen abstraction from Cp* (TS_II) and affords the rhodium(I)-fulvene complex III, which should undergo a reversible H2O/D2O exchange, yielding the progressive hydrogen exchange/deuteration of the Cp* ligand (Scheme ).

Scheme 1

Reaction Sequence for the Hydrogen Exchange at 3

The Cp* ligand of the iridium complex 4 undergoes an H/D exchange process similar to that described for the rhodium analogue 3 but at a much slower rate. Indeed, at 293 K the 1H NMR spectrum of [D8]THF/D2O (78%/22%, v/v) solutions of the iridium complex 4 does not change over time. It is necessary to heat the reaction mixture at 343 K to observe the H/D exchange at an appreciable rate. After 4 days at this temperature, the Cp* ligand is deuterated at about 50%, on average. Apart from the isotopologues of 4 derived from the H/D exchange process, the formation of a new iridium complex, labeled as 8 (vide infra), was detected by NMR spectroscopy. The overlapping of the 1H NMR signals prevents the detailed study of the evolution of both the H/D exchange process and the reaction of formation of complex 8. The complete characterization of 8 will be discussed in the next subsection.

For the sake of comparison, the Gibbs free energy profiles of the hydrogen exchange for both 3 and 4 were calculated at the level wB97XD/def2tzvp//wB97XD/def2svp, using the SMD model for the solvent, at 298 K. Figure shows the calculated intermediates and transition states along with the relative Gibbs free energies.

Figure 6

Gibbs free energy profile (kcal·mol–1) for the hydrogen exchange at 3 (black) and 4 (gray) in the presence of water [wB97XD/def2tzvp//wB97XD/def2svp, in THF (SMD model), 298 K, 1 atm].

DFT calculations indicate that for both 3 and 4 the hydroxo intermediates II and II, respectively, are obtained stepwise by reaction of water with 3 or 4. Actually, the dissociation of the terminal M–N bond of 3 or 4 renders the true FLP complex, namely I and I, which interacts with one water molecule affording I or I. Neither I nor I contains a metal-oxygen bond, rather the incoming water molecule forms an N···H–O bond (Figure ).[27] Subsequent coordination of oxygen to the metal center and the concomitant H–OH bond rupture yield II or II in which a hydrogen bond still exists between the newly formed MOH and NH moieties (Figure ). Once II and II form, hydrogen abstraction from its Cp* ligand gives the η4-tetramethylfulvene ligand in III and III and one weakly bonded water molecule (Figure ). The hydrogen abstraction (II → III; II → III) entails the formal reduction of the metal center from the oxidation state +3 to +1. Accordingly, the metal centers of III and III feature a distorted square planar geometry in which two coordination sites are occupied by fulvene, whereas the phosphorus and nitrogen atoms from the phosphano-guanidine ligand complete the coordination sphere of the metal center (Figure ). Notably, the metal-oxygen distance (III, 3.562; III, 3.616 Å) rules out the existence of a metal-oxygen bond. In addition, the water molecule is weakly bonded to the metal complex by means of the N2–H2···O hydrogen bond and an additional C2···H1–O short contact (Figure ). Finally, the exchange of this weakly bonded water molecule with water (or D2O) solvent molecules should cause the progressive hydrogen exchange (deuteration) of the Cp* ligand. In view of the Gibbs free energy profiles given in Figure , it should be noted that intermediates 3·H2O (or 4·H2O) and II (or II), even if thermally accessible, are less stable than the starting complex 3 (or 4), which nicely agrees with the fact that 3 (or 4) are the only detected species in the course of the H/D exchange, and none of the intermediates has been observed by NMR spectroscopy.

Figure 7

Calculated structures of I, II, III, and TS_II with the numbering scheme adopted. The calculated structures of I, II, III, and TS_II are similar and are not reported for the sake of brevity, the same numbering scheme being adopted. For clarity, most hydrogens are omitted, and only ipso carbon atoms of Tolyl and Phenyl groups are shown. Selected bond lengths/interatomic distances (Å) and angles (°) are: I, N2–H2 1.836, O–H2 0.988, N2–O 2.820, N2–H2–O 173.4, Rh–O 3.651; I, N2–H2 1.842, O–H2 0.986, N2–O 2.826, N2–H2–O 176.0, Ir–O 3.685; II, Rh–O 2.078, O–H2 1.642, N2–H2 1.055, N2–O 2.655, N2–H2–O 159.4; II, Ir–O 2.095, O–H2 1.647, N2–H2 1.051, N2–O 2.649, N2–H2–O 157.5; TS_II, Rh–O 2.219, O–H2 1.787, N2–H2 1.031, N2–O 2.787, N2–H2-O 162.2; C2–H1 1.429, O–H1 1.207, O–H1–C2 158.2; TS_II, Ir–O 2.255, O–H2 1.807, N2–H2 1.209, N2–O 2.798, N2–H2–O 160.4; C2–H1 1.485, O–H1 1.169, O–H1–C2 159.1; III, Rh–O 3.562, O–H2 1.847, N2–H2 1.029, N2–O 2.866, N2–H2–O 170.1; C2–H1 2.267, O–H1 0.972, O–H1–C2 167.8; III, Ir–O 3.616, O–H2 1.849, N2–H2 1.031, N2–O 2.865, N2–H2–O 167.7; C2–H1 2.196, O–H1 0.974, O–H1–C2 170.2.

As for the Gibbs free energy variation along the reaction sequence 3/4 + H2O ⇆ III/III, hydrogen exchange at 4 exhibits a significantly higher activation barrier (ΔGact = +29.9 kcal mol–1) when compared with 3 (ΔGact = +22.8 kcal mol–1), which perfectly fits in with the experimental conditions required for the H/D exchange of 3 and 4, and with the observed degree of deuteration (vide supra).

Finally, as far as the CN3 core is concerned, despite the fact that some degree of delocalization is expected to occur within the three carbon-nitrogen bonds, in the course of the hydrogen exchange a considerable electronic rearrangement takes place and it is reasonably beneficial to the accomplishment of the hydrogen exchange itself. Indeed, the analysis of the calculated carbon-nitrogen bond lengths for complexes 3, 4, I, II, TS_II, and III (M = Rh, Ir, Scheme ) points out that in the course of the hydrogen exchange the C1–N3 bond essentially holds its single bond character, whereas the C1–N1 and C1–N2 bonds switch from single to double and vice versa in the course of the sequence I ⇆ II ⇆ III, the formal protonation of N2 triggering the switch from one electronic distribution to the other.

Scheme 2

Calculated Carbon-Nitrogen Bond Lengths (Å) in 3, 4, I, II, and III (M = Rh, Ir)

Orthometalation Reactions

Heating THF/H2O (4/1, v/v) solutions of 3 or 4 at 383 K affords the orthometalated complexes 7 and 8, respectively (eq ).

Single crystals of 7 and 8 were grown from THF/Et2O (7), CH3OH/Acetone/Et2O/n-pentane (8a, from here on) and CH2Cl2 (8b, from here on) solutions, 8a and 8b featuring different crystal structures. Figure shows the ORTEP plot of the cations [Cp*Ir(κ3C,N,P-H)]+ in 8a and 8b [H = Ph2PCH2CH2NC(NH(p-Tolyl))(NH(4-C6H3Me))], and Table contains selected bond lengths and angles. The rhodium compound 7 exhibits a crystal structure virtually superimposable to 8a, so only 8a will be discussed in detail and selected data of 7 are included in the Supporting Information.

Figure 8

ORTEP plot of the cations of complexes 8a and 8b. Thermal ellipsoids are at 50% probability, and most hydrogen atoms have been omitted for clarity.

Table 4

Selected Bond Lengths (Å) and Angles (°) of 8a and 8b

	8a	8b		8a	8b
Ir–P	2.2621(10)	2.2598(10)	P–Ir–Cta	133.81(3)	135.48(3)
Ir–N(1)	2.081(3)	2.095(3)	N(1)–Ir–C(27)	84.25(14)	82.48(13)
Ir–C(27)	2.065(4)	2.069(4)	N(1)–Ir–Cta	126.22(9)	128.99(9)
Ir–Cta	1.8559(2)	1.8809(2)	C(27)–Ir–Cta	125.66(10)	119.32(10)
N(1)–C(25)	1.312(5)	1.304(5)	Ir–N(1)–C(12)	118.0(2)	118.6(2)
N(2)–C(25)	1.355(5)	1.356(5)	Ir–N(1)–C(25)	122.2(3)	119.9(3)
N(3)–C(25)	1.363(5)	1.374(5)	C(12)–N(1)–C(25)	119.5(3)	120.4(3)
P–Ir–N(1)	77.05(9)	81.44(9)	Σ°N(1)b	359.7(5)	358.9(5)
P–Ir–C(27)	92.47(11)	93.21(11)	N(1)–C(25)–N(3)–C(33)	153.8(4)	–43.0(6)

Ct represents the centroid of the η5-C5Me5 ligand.

Σ°N(1) stands for the sum of bond angles around the N(1) atom.

Both 8a and 8b exhibit a three-legged-piano stool geometry with an η5 coordinated Cp* ligand. The metalated phosphano-guanidine ligand occupies three mutually cis positions at the metal center [8a, P–Ir–N(1) 77.05(9), P–Ir–C(27) 92.47(11), N(1)–Ir–C(27) 84.25(14)°; 8b, P–Ir–N(1) 81.44(9), P–Ir–C(27) 93.21(11), N(1)–Ir–C(27) 82.48(13)°] rendering two fused metalacycles, namely, the five-membered ring Ir–P–C(11)–C(12)–N(1) and the six-membered ring Ir–C(27)–C(26)–N(2)–C(25)–N(1).

Interestingly, the metal center in both 8a and 8b is stereogenic. Nonetheless, as a consequence of the centrosymmetric space group P21/c of 8a and 8b, both enantiomers, namely, SIr-8a/b (shown in Figure ) and RIr-8a/b, are present in the crystal.[28] When dealing with the differences between 8a and 8b, the arrangement of the exocyclic N(3)H(p-Tolyl) moiety with respect to the IrCp*(P)(N1)(C27) core is worth a mention. As a matter of fact, the dihedral angle N(1)–C(25)–N(3)–C(33) is 153.8(4)° in 8a and–43.0(6)° in 8b indicating that the N(3)–C(25) bond adopts a conformation close to s-trans in 8a and close to s-cis in 8b (Figure ). As a final remark, when comparing 8a/8b with 4, reasonably as a consequence of the formation of the less strained six membered ring Ir–C(27)–C(26)–N(2)–C(25)–N(1) instead of the four membered ring Ir–N(1)–C(25)–N(2), the nitrogen atom N(1) exhibits a planar geometry both in 8a [Σ°N(1) = 359.7(5)°] and 8b [Σ°N(1) = 358.9(5)°]. On this ground, N(1) should adopt a sp2 hybridization in 8a and 8b. Accordingly, the N(1)–C(25) bond length is shorter [8a, 1.312(5); 8b, 1.304(5) Å] than the N(2)–C(25) [8a, 1.355(5); 8b, 1.356(5) Å] and N(3)–C(25) bond lengths [8a, 1.363(5); 8b, 1.374(5) Å], suggesting that, despite some degree of delocalization over the CN3 core, the N(1)–C(25) bond should exhibit a higher double bond character when compared with N(2)–C(25) and N(3)–C(25).

No significant differences between the solution NMR spectra of both iridium rotamers 8a and 8b have been found in the 293–233 K temperature range indicating that, under these conditions, rotation around the N(3)–C(25) bond is free. Most probably, crystal packing accounts for the two dispositions encountered in the solid state.

The presence of two 1H peaks attributed to NH protons, at 8.41 and 8.36 ppm for 7, and at 8.43 and 8.34 ppm for 8, is indicative of the protonation of the N(p-Tolyl) group. The orthometalated carbon nucleus gives a doublet of doublets at 141.15 ppm [J(RhC) = 31.7 Hz, J(PC) = 13.8 Hz] for complex 7 and a doublet at 123.61 ppm (J(PC) = 9.1 Hz) for complex 8. The 31P{1H} NMR spectrum consists of a doublet centered at 46.45 ppm [J(RhP) = 153.3 Hz] and one singlet at 19.01 ppm for 8.

The energy profile E vs dihedral angle NCNC (α) for the rotation around the exocyclic C–N bond (Figure ) was calculated for the rhodium complexes 7a/7b showing that the barrier for the rotation is about 10 kcal mol–1. In addition, the Gibbs free energy differences between isomers 7a and 7b as well as between isomers 8a and 8b are small (G-G = +1.4 kcal mol–1, G-G = +0.1 kcal mol–1). On these grounds, both isomers for each metal should be present in solution at room temperature and the interconversions 7a ⇆ 7b and 8a ⇆ 8b should be fast at that temperature, which fits in with the observation of averaged NMR spectra for 7a and 7b as well as for 8a and 8b, and with the isolation of crystals of each isomer using different crystallization mixtures of solvents.

Figure 9

Energy profile (E vs α, wB97XD/def2svp, 298 K) for the rotation around the exocyclic C–N bond.

Reaction of 3 or 4 with Alcohols

At 333 K, complexes 3 and 4 react in THF with methanol, primary alcohols, and 2-propanol cleanly giving the metal-hydrido complexes 5 and 6, respectively (eq ). The reaction involves the dehydrogenation of the alcohols at a relatively low[19,20] temperature and without the assistance of an external base. 1H NMR signals assigned to methyl formate[19h] [δH 8.07, brq; 3.76, brd], acetaldehyde [δH 9.67, q; 2.07, d (J = 2.8 Hz)], propionaldehyde [δH 9.56 t (J = 1.3 Hz)], benzaldehyde (δH 10.02, s), and acetone (δH 2.04, s) were detected after the reaction with methanol, ethanol, n-propanol, benzyl alcohol, and 2-propanol, respectively.

As the reactions with the iridium complex 4 were much slower than those with the rhodium compound 3, kinetic studies were carried out only with 3. Table collects the values of the kinetic constants measured at 333 K (see the Supporting Information). The dehydrogenation rate is greater for methanol and primary alcohols (entries 1–4) than for the secondary alcohol 2-propanol (entry 5).[19h] To obtain information about the mechanism, in independent experiments the reaction was carried out with CH3OD (entry 6) or CD3OH (entry 7). In the reaction with CD3OH, the metal-hydrido region of the 1H NMR spectrum of the resulting product was silent, but when CH3OD was used as a reagent, a Rh–H 1H resonance was observed in the product. Notably, values of 5.21 and 8.29 were obtained for the kobs(CH3OH)/kobs(CH3OD) and kobs(CH3OH)/kobs(CD3OH) ratios, respectively. A detailed kinetic study was not carried out with CD3OD because the reaction rate in this solvent is very low. Indeed, a conversion of only about 4% was measured after 60 h of reaction at 333 K.

Table 5

Kinetic Constant for the Reaction of Complex 3 with Alcohols at 333 Ka

entry	alcohol	106kobs/s–1
1	MeOH	8.12 ± 0.07
2	EtOH	11.6 ± 0.1
3	nPrOH	4.0 ± 0.2
4	BnOH	24.9 ± 0.5
5	iPrOH	1.12 ± 0.04
6	CH3OD	1.56 ± 0.03
7	CD3OH	1.0 ± 0.1

See the SI for experimental details.

As it was observed in the reaction of 3 or 4 with deuterated water, mass and 1H NMR spectra of solutions of compound 3 in alcohols with deuterated hydroxo groups indicate that the progressive deuteration of the methyl groups of the Cp* ring occurs. Kinetic measurements establish that the deuteration process obeys a pseudo-first order rate law. Table collects the values of the kinetic constants measured at 313 K, and for comparative purposes, kobs obtained for D2O[21] was also included. In general, the kobs for the H/D exchange process is greater than that measured for the alcohol dehydrogenation. For example, for CH3OD, the kobs for the Cp* deuteration is 6.39 ± 0.08 × 10–4 s–1 at 323 K (see the Supporting Information), and that for the dehydrogenation process is 1.56 ± 0.03 × 10–6 s–1, at 333 K (entry 6, Table ), that is, the latter is about 400 times lower than the former despite being measured at a temperature 10 K higher. Based on the Eyring plot [ln(kobs/T) vs 1/T] ΔG≠293 of around 24 kcal·mol–1 was calculated in all cases (see the Supporting Information).

Table 6

Kinetic Constants for the H/D Exchange at 313 Ka

entry	R–OD	105kobs/s–1
1	D–OD	9.5 ± 0.2
2	CH3–OD	28.2 ± 0.6
3	CD3–OD	16.9 ± 0.2
4	Et–OD	7.73 ± 0.08
5	iPr–OD	1.92 ± 0.04

See the SI for experimental details.

The mechanism of the reactions of 3 and 4 with methanol was explored by DFT calculations in order to shed light on the deuteration of 3 and 4 in the presence of CH3OD as well as on the formation of 5 and 6, respectively, as a result of the dehydrogenation of methanol. For both reactions, the energy profiles for 3 and 4 were elucidated by means of DFT computational methods at the level wB97XD/def2tzvp//wB97XD/def2svp using the SMD model for the solvent (THF).

As for the H/D exchange, the calculated reaction sequence is reminiscent of that previously discussed for the reaction of 3 or 4 with water (Figure ). As a matter of fact, methanol reacts with 3 or 4 yielding I or I, respectively, in which a N···HO hydrogen bond brings together the dissociated form of 3 or 4, namely, I and I, with a methanol molecule, no metal-oxygen bond being observed (vide infra). In the following, the rupture of the O–H bond affords the methoxo derivatives IV and IV featuring an intramolecular NH···O hydrogen bond between the newly formed NH group and the methoxo ligand. Similar to II and II (Figure ), the Cp* ligand in IV and IV undergoes a hydrogen abstraction yielding the tetramethylfulvene metal(I) complexes V and V, respectively, in which the resulting methanol molecule is still involved in an NH···O hydrogen bond with the NH group. Also, a short OH···CH2fulvene contact is observed between the fulvene ligand and the methanol molecule. Like for III and III, no metal-oxygen bond exists in V and V, and the exchange of the weakly bonded methanol molecule with methanol/methanol-d solvent molecules triggers the hydrogen exchange/deuteration of the Cp* ligand. Similar to the reaction of 3 or 4 with water, the activation barrier for 3 + MeOH ⇆ V (+21.5 kcal·mol–1) is significantly lower than that for 4 + MeOH ⇆ V (+29.0 kcal mol–1), which nicely fits in with the experimental conditions and the outcome of the deuteration reaction with 3 or 4.

Figure 10

Gibbs free energy profile (kcal·mol–1) for the hydrogen exchange at 3 (black) and 4 (gray) in the presence of methanol [wB97XD/def2tzvp//wB97XD/def2svp, in THF (SMD model), 298 K, 1 atm].

As for the dehydrogenation of methanol rendering 5 or 6, DFT calculations suggest that I and I are again key intermediates (Figure ). As a matter of fact, they convert into 5·CH or 6·CH, respectively, via the concerted transition state TS_I or TS_I. Notably, the elimination of CH2O results from the simultaneous migration of one CH hydrogen atom to the metal center and of the OH hydrogen atom to a nitrogen atom of the guanidine moiety (cf. TS_I, Figure ). Accordingly, the carbon–oxygen bond shortens from 1.396 Å (av.) to 1.206 Å (av.) on going from I and I to 5·CH and 6·CH, respectively (Figure ).

Figure 11

Gibbs free energy profile for the reaction 3 (or 4) + MeOH → 5 (or 6) + CH2O [wB97XD/def2tzvp//wB97XD/def2svp, in THF (SMD model), 298 K]. View of the I, 5·CH, and TS_I with the numbering scheme adopted. The calculated structures of I, 6·CH, and TS_I are similar and are not reported for the sake of brevity, the same numbering scheme being adopted. Selected bond lengths/interatomic distances (Å) and angles (°) are I, N2–H2 1.817, O–H2 0.988, N2–O 2.800, N2–H2–O 173.0, Rh–O 3.660, C2–O 1.397; I, N2–H2 1.821, O–H2 0.986, N2–O 2.802, N2–H2–O 172.8, Rh–O 3.673, C2–O 1.396; TS_I, Rh–H1 1.832, H1–C1 1.198, C2–O 1.322, O–H2 1.279, N2–H2 1.206; TS_I, Ir–H1 1.863, H1–C1 1.211, C2–O 1.319, O–H2 1.268, N2–H2 1.216; 5·CH, H3–O 2.447, H1–C2 2.433, C2–O 1.206; 6·CH, H3–O 2.436, H1–C2 2.413, C2–O 1.205.

In this connection, previously reported studies already indicated that mono-[29] and dinuclear[30] iridium complexes as well as ruthenium derivatives[31] are able to perform the acceptorless dehydrogenation of methanol via a concerted transition state taking advantage of the bifunctional character of the metal–ligand platform.

The calculated barriers for the rhodium complex 3 (+22.7 kcal·mol–1 at 298 K, +24.3 kcal·mol–1 at 333 K) and for the iridium complex 4 (+24.5 kcal·mol–1 at 298 K, +26.1 kcal·mol–1 at 333 K) underpin the experimental conditions. It is worth mentioning that according to the Gibbs free energy profile given in Figure the above mentioned overall reactions are slightly endergonic (+2.9 kcal mol–1, M = Rh; +0.6 kcal mol–1, M = Ir). Nonetheless, in this regard, as mentioned before, methyl formate was detected as a product in the reaction of 3 or 4 with methanol, and its formation from formaldehyde was calculated to be exergonic (CH2O → 1/2 HCOOCH3, ΔGr = −9.7 kcal mol–1) which compensate the above mentioned positive ΔGr. On the other hand, when ethanol, 2-propanol, and benzyl alcohol were used, the overall dehydrogenation reaction CHR2OH + 3 (or 4) → CR2O + 5 (or 6) was calculated to be exergonic (ΔGr = from −10.8 to −4.5 kcal·mol–1) in agreement with the observation of acetaldehyde, acetone, and benzaldehyde, respectively, in the reaction mixture.

Conclusions

Compounds 3 and 4 behave like masked TMFLPs. The fac κ3N,N′,P coordination of the phosphano-guanidine ligand forces the central nitrogen atom to adopt an sp3 hybridization thus generating a strong strain within the M–N–C–N four-membered metalacycle. This structural stress makes the “unmasked” TMFLP thermally accessible (eq ). The metal (acidic center) and the iminic nitrogen (basic center) synergistically cooperate in the reversible activation of molecular hydrogen as well as in the activation of the O–H bond of water and alcohols. The resulting nucleophilic M–OH and M–OR fragments are able to reversibly dehydrogenate the Cp* methyl groups giving rise to sequential and complete H/D exchange of the Cp* protons when deuterated D–OD or R–OD solvents were employed. On the other hand, the “unmasked” TMFLP is also able to dehydrogenate alcohols affording metal hydrido derivatives via a concerted transition state involving simultaneously the acidic and the basic sites.

In this respect, the greater reactivity of the aldehyde or ketone products of the alcohol dehydrogenation versus the starting alcohol together with the reversibility of the hydrogenation reaction of complexes 3 and 4 paves the way to the potential application of these complexes to catalyzed reactions of alcohols using borrowing hydrogen methodology. On the other hand, judicious design of tridentate ligands capable of a fac κ3N,N′,P coordination as well as incorporation of d6 ions of precious and nonprecious metals would dramatically expand the applicability of the derived TMFLP species both in small-molecule activation chemistry and in the development of new catalytic processes.

Further work in this area is in progress and will be reported in due course.