Alternative Conceptual Approach to the Design of Bifunctional Catalysts: An Osmium Germylene System for the Dehydrogenation of Formic Acid

Abstract

The reaction of the hexahydride OsH₆(PⁱPr₃)₂ with a P,Ge,P-germylene-diphosphine affords an osmium tetrahydride derivative bearing a Ge,P-chelate, which arises from the hydrogenolysis of a P–C(sp³) bond. This Os(IV)–Ge(II) compound is a pioneering example of a bifunctional catalyst based on the coordination of a σ-donor acid, which is active in the dehydrogenation of formic acid to H₂ and CO₂. The kinetics of the dehydrogenation, the characterization of the resting state of the catalysis, and DFT calculations point out that the hydrogen formation (the fast stage) exclusively occurs on the coordination sphere of the basic metal center, whereas both the metal center and the σ-donor Lewis acid cooperatively participate in the CO₂ release (the rate-determining step). During the process, the formate group pivots around the germanium to approach its hydrogen atom to the osmium center, which allows its transfer to the metal and the CO₂ release.

Short abstract

An alternative class of bifunctional catalysts can be assembled by coordination of σ-donor Lewis acids to platinum-group-metal basic fragments. In contrast to what happens with the previously reported bifunctional catalysts, this design allows enhancing the basicity of the base and the acidity of the acid. According to this, a bifunctional catalyst for the dehydrogenation of formic acid, based on an osmium(IV)-germylene cooperative system, has been prepared and the mechanism of the catalysis established.

Introduction

The direct participation of both the metal center and a ligand of the catalyst during a transition-metal-catalyzed organic transformation is a successfully employed strategy in the current homogeneous organometallic catalysis. The metal–ligand cooperation is based on the distribution of roles among the main actors of the catalyst. In most of the cases, the metal acts as a Lewis acid whereas the ligand is a Lewis base (a in Chart 1).¹ Such a principle is the support of most relevant catalysts of this class, including Shvo’s cyclopentadienone metal complexes,² Fujita’s iridium pyridonate,³ Noyori-Ikariya’s ruthenium amide,⁴ or Milstein’s PNN-metal systems.⁵ Nevertheless, the design of bifunctional catalysts to operate with exchanged metal–ligand roles has awakening great interest in the last few years.⁶ Such catalysts are formed by a basic metal and an acidic σ-acceptor Z-type ligand where the bonding σ-orbital is empty (b in Chart 1).⁷ Unfortunately, the number of Z-type ligands is notably limited⁸ and as a consequence the reactions employing catalysts of this class are still very scarce. Metal Lewis bases mainly involve 3d metals, with iron,⁹ cobalt,¹⁰ and nickel¹¹ in prominent positions, in addition to rhodium,¹² palladium,¹³ platinum,¹⁴ and gold.¹⁵ The Lewis acid site of the ligands is boron^{9,10a−10c,11b−11d,13a,15b,15d} in the great majority of the cases and to a much lesser extent aluminum,^11a15a gallium,^11e11f indium,¹² silicon,^13b and antimony.^7b,14,15c Catalytic reactions include oligomerization of primary silanes,^11a sila-Negishi couplings,^13b hydrosilylation of aldehydes,^10c11c olefin and alkyne hydrogenations,^{9,11b,11d,11f} CO₂ hydrogenation to formate,^11e amine–borane dehydrogenation,^10a,10b cycloisomerization of propargylamides,^15a hydroaminations,^15c,15d C–H amination,¹² enyne cyclization,^14,15b and dehalogenation of aryl chlorides.^13a

Chart 1. Bifunctional-Type Catalysts.

The electron density flow of the metal–ligand interaction in both classes of bifunctional catalysts goes from the Lewis base to the Lewis acid, as usual. However, that works against the nature of both, since it decreases the basicity of the former and the acidity of the latter. We now address an alternative approach (c in Chart 1): the assembly of a bifunctional catalyst carrying such a flow in favor of both actors, enhancing the basicity of the base and the acidity of the acid. The challenge needed a basic center able of accepting electron density from an acidic center capable of donating it. We reasoned that such a situation would be possible by means of the use of a basic metal center, which should be in a high oxidation state, in contrast to those employed in catalysts involving a M→Z interaction. The high oxidation state should favor the acceptance of electron density from a σ-donor orbital and prevent significant back-bonding donation to an empty π-orbital. Metal fragments with these characteristics are the polyhydrides of the platinum-group metals.¹⁶ Among them, the osmium polyhydrides occupy a predominant position due to their proven ability to catalyze interesting organic synthesis reactions,¹⁷ in addition to the dehydrogenation of liquid organic hydrogen carriers¹⁸ and boranes.¹⁹ Promising acidic ligands capable of donating σ-electron density to a platinum-group metal in a high oxidation state are the heavier tetrylenes,²⁰ heavier counterparts of Fisher-type carbenes. They prefer to situate the two free electrons in an orbital with marked s character, leaving a vacant orbital with high p character at an orthogonal plane. The metal–tetrylene bonding interaction is significantly weaker than the metal–carbene interaction, and its strength decreases on going down group 14. Thus, germylenes are particularly interesting due to the center position of germanium in the group, which grants a reasonable σ-donor ability and a low π-acceptor capacity to them.²¹ The latter is evidenced by the marked tendency of these ligands to undergo inter- and intramolecular addition of 2e^– donor groups.²²

Formic acid is a polar molecule that is able of interacting with a polar M–L bond. Furthermore, because it is an attractive member of the family denoted as liquid organic hydrogen carriers, due to its low toxicity and flammability and reasonable gravimetric and volumetric hydrogen content (4.4 wt % and 53.4 g/L, respectively),²³ its dehydrogenation to H₂ and CO₂ is of great interest. The reaction is catalyzed by both homogeneous and heterogeneous systems, including bifunctional catalysts based on the traditional acidic-metal/basic-ligand approach.²⁴ The catalysis is divided into two main stages: H₂ formation and CO₂ formation.²⁵ A mechanistic feature of the traditional bifunctional catalysts is the participation of both the metal and the ligand in the H₂ formation stage, whereas the CO₂ formation exclusively takes place on the metal coordination sphere.²⁶ Thus, the dehydrogenation of formic acid might be an ideal reaction to be used as a proof of concept validation for our hypothesis and also to analyze the mechanistic differences between the traditional bifunctional catalysis and that now designed.

This paper reports the preparation of a bifunctional catalyst for the dehydrogenation of formic acid, based on an osmium(IV) germylene cooperative system, which works in a manner different from that of the traditional catalysts of this class and represents a new conceptual approach to the design of bifunctional catalysts.

Results and Discussion

Preparation of the Catalyst

In the search for a catalyst with a robust skeleton, which provides a strong base–acid interaction to the system, we introduced the germylene acid between two phosphine groups, with the initial idea of giving it the role of the central moiety of a neutral P,Ge,P-pincer ligand. With this aim, we chose 1,3-bis(di-tert-butylphosphanylmethyl)-1,3-dihydro-2λ²-benzo[d][1,3,2]diazagermole, because it had previously shown a notable versatility in platinum-group-metal chemistry.²⁷ With this molecule in hand, we selected the d² hexahydride OsH₆(PⁱPr₃)₂ (1) as the metal precursor, as it had proven to be efficient for the preparation of the tetrahydride OsH₄{κ³-P,O,P-[xant(PⁱPr₂)₂]} by the replacement of the triisopropylphosphines and a hydrogen molecule with the neutral P,O,P-pincer ligand 9,9-dimethyl-4,5-bis(diisopropylphosphino)xanthene.^17e

Treatment of toluene solutions of 1 with 1.1 equiv of the P,Ge,P-pincer, at 110 °C, for 18 h led to the coordination of the germylene moiety to the osmium center, as expected. However, the substitution of only one triisopropylphosphine ligand was observed, while the hydrogenolysis of the CH₂–P^tBu₂ bond of one of the pincer side arms took place. As a result, di-tert-butylphosphine and the tetrahydride 2, bearing one of the initial triisopropylphosphines and a Ge,P-chelating germylene-phosphine ligand, were formed (Scheme 1). The elimination of di-tert-butylphosphine from the original P,Ge,P-pincer is notable. In spite of it being argued as the reason for the degradation of some hydrogenation catalysts²⁸ and of being known for the rupture of P–C bonds of some quaternary phosphonium salts under catalytic conditions,²⁹ the hydrogenolysis of P–C bonds is a slightly common reaction. It should be furthermore mentioned that, although complex 1 has shown a notable ability to activate σ-bonds,³⁰ including C–C,³¹ C–N,³² and C–O³³ among others, it had never participated in the rupture of a C–P bond.

Scheme 1. Formation of 2.

Complex 2 was isolated as a pale yellow solid in 48% yield and characterized by X-ray diffraction analysis. Figure 1 gives a view of the structure. The coordination polyhedron around the osmium atom can be rationalized as a pentagonal bipyramid with the phosphine P atoms at the apical positions (P(1)–Os–P(2) = 169.88(2)°), whereas the hydride ligands and the germylene group define the base of the polyhedron. The Os–Ge distance, 2.3593(3) Å, compares well with those reported for other osmium germylene derivatives.³⁴ A DFT-optimized structure confirmed the classical character of the hydride ligands, since it reveals a separation between them of longer than 1.834 Å. In toluene, these ligands are involved in thermally activated position exchange processes. Thus, the ¹H NMR spectrum at room temperature shows only one high-field resonance for the two groups of inequivalent hydrides, which appears at −10.45 ppm as a doublet of doublets, with both H–P coupling constants of 12.4 Hz. Lowering the temperature of the sample produces a broadening of the resonance. However, decoalescence is not reached even at 183 K. In the same temperature range, the ³¹P{¹H} NMR spectrum contains two doublets at 99.8 and 55.5 ppm. In agreement with the mutually trans disposition of the phosphorus nuclei, the P–P coupling constant is 224.5 Hz.

Figure 1.

Molecular diagram of complex 2 (50% probability ellipsoids). Hydrogen atoms (except hydride ligands) are omitted for clarity. Selected bond lengths (Å) and angles (deg): Os–Ge = 2.3593(3), P(1)–Os–P(2) = 169.88(2), P(1)–Os–Ge = 81.038(16), P(2)–Os–Ge = 108.668(17), N(2)–Ge–N(1) = 86.49(9), N(1)–Ge–Os = 161.01(7), N(2)–Ge–Os = 110.75(6).

Complex 2 is not exactly the compound initially designed by us but is very similar. The main difference is the lack of a link between the germylene group and the monodentate phosphine. Because complex 2 still fulfills the planned features regarding the Os–Ge interaction, we decided to pursue the initial program.

Reactions of 2 with Benzoic and Acetic Acids

The OsH₄ moiety is certainly the basic part of 2, whereas the acid center is located at the germanium atom. In agreement with this, complex 2 reacted with carboxylic acids in toluene at room temperature to give the trihydrides 3 and 4 (Scheme 2). These compounds result from the protonation of the basic osmium moiety and the neutralization of the Lewis acidity of the germylene by an arm of the corresponding carboxylate anion. The protonation generates a dihydrogen ligand that is released, occupying the generated coordination vacancy of the other arm of the carboxylate groups.

Scheme 2. Reactions of 2 with Benzoic and Acetic Acids.

Complexes 3 and 4 were isolated as pale yellow solids in 56% and 48% yields, respectively. Their formation was confirmed by a structural X-ray diffraction analysis of 3 (Figure 2). The addition of the anion to the germanium atom gives rise to a P,Ge,O-tridentate ligand, which coordinates in a fac fashion to the metal center, with bite angles P(1)–Os–Ge, P(1)–Os–O(1), and Ge–Os–O(1) of 81.22(4), 97.11(13), and 79.87(12)°. The resulting coordination polyhedron around the osmium atom is the typical pentagonal-bipyramidal characteristic of a osmium(IV) polyhydride species with the phosphorus atoms in apical positions (P(1)–Os–P(2) = 170.07(6)°), whereas the hydride ligands and the germanium and oxygen atoms of the tridentate group lie at the base. The Os–Ge bond length of 2.4204(7) Å is about 0.06 Å longer than in 2.The DFT-optimized structure revealed that the OsH₃ unit forms a hydride-compressed dihydride system with H(01) and H(02) separated by 1.540 Å and H(02) and H(03) by 1.752 Å. As is usual in this class of polyhydrides, the hydride ligands are involved in thermally activated positional exchange processes in solution. Thus, the ¹H NMR spectra of both species, in toluene-d₈, at room temperature display a broad resonance at about −11.6 ppm for the inequivalent hydride ligands. In contrast to 2, this resonance decoalesces between 243 and 233 K, to afford a broad triplet around at −4 ppm and an AB spin system close to −15.5 ppm. The H–H coupling constant values of the latter, >124 Hz, suggest that the hydrogen nuclei of the compressed dihydride undergo quantum exchange coupling^16,32b,35 in addition to the thermally activated positional exchange process. In agreement with 2, the ³¹P{¹H} NMR spectra show two doublets at about 84 and 48 ppm with P–P coupling constants of around 236 Hz.

Figure 2.

Molecular diagram of complex 3 (50% probability ellipsoids). Hydrogen atoms (except hydride ligands) are omitted for clarity. Selected bond lengths (Å) and angles (deg): Os–Ge = 2.4204(7), Os–O(1) = 2.154(5), Ge–O(2) = 2.042(4), O(1)–C(1) = 1.267(8), O(2)–C(1) = 1.272(8), P(1)–Os–P(2) = 170.07(6), P(1)–Os–Ge = 81.22(4), P(2)–Os–Ge = 108.45(5), P(1)–Os–O(1) = 97.11(13), Ge–Os–O(1) = 79.87(12), N(2)–Ge–N(1) = 86.7(2), N(2)–Ge–O(2) = 100.3(2), N(1)–Ge–O(2) = 98.9(2), N(1)–Ge–Os = 108.98(15), N(2)–Ge–Os = 155.50(17), O(2)–Ge–Os = 95.75(12).

Reaction of 2 with Fomic Acid

Once it was known that complex 2 activates the O–H bond of usual carboxylic acids, such as benzoic and acetic acids, and established the nature of the resulting compounds, we investigated the behavior of formic acid. The addition of 1.0 equiv of the latter to toluene solutions of the tetrahydride, at room temperature, rapidly produced the release of 1.0 equiv of H₂ and the quantitative formation of 5, the formate counterpart of 3 and 4. Complex 5 was fully characterized by ¹H, ³¹P{¹H}, and ¹³C{¹H} NMR spectroscopy (see the Supporting Information). In agreement with 3 and 4, the ¹H spectrum of 5 contains a broad resonance centered at −11.73 ppm, which is split into a broad signal at −4.16 ppm and an AB spin system at −15.66 ppm (²J_H–H > 153.0 Hz) at temperatures lower than 233 K, whereas the ³¹P{¹H} spectrum displays two doublets at 84.1 and 48.3 ppm with a P–P coupling constant of 234.6 Hz. In solution, complex 5 is unstable. It releases CO₂ to regenerate 2 (Scheme 3).

Scheme 3. Stoichiometric Cycle for the Dehydrogenation of Formic Acid to H₂ and CO₂ Promoted by 2.

The formation of 5 and its decomposition constitutes a cycle for the stoichiometric dehydrogenation of formic acid into H₂ and CO₂, which is kinetically controlled by the CO₂ release, as illustrated Scheme 3. To gain information about the activation parameters of the process, we followed the transformation of 5 into 2 by ³¹P{¹H} NMR spectroscopy, between 303 and 338 K, as a function of time. Figure 3 shows the spectra of the reaction at 323 K. The consumption of 5 is an exponential function of time, which fits a first-order process, according to the expression

1

where [5]₀ is the initial concentration of 5 and [5] is the concentration at time t. Table 1 gathers the values obtained for k₁ in the studied temperature range. The activation parameters calculated through an Eyring analysis (Figure 4) are ΔH₁^⧧ = 21 ± 2 kcal mol^–1 and ΔS₁^⧧ = −3 ± 5 cal K^–1 mol^–1. These values yield an activation energy ΔG₁^⧧ at 298 K of 22 ± 3 kcal mol^–1.

Figure 3.

³¹P{¹H} spectra (161.98 MHz, in toluene-d₈) for the decarboxylation of 5 at 323 K.

Table 1. Rate Constants for the Transformation of 5 into 2.

temp (K)	rate constant kst (min–1)
303	0.001
313	0.002
318	0.007
328	0.017
338	0.045

Figure 4.

Eyring plot (blue: transformation of 5 into 2; red, catalytic dehydrogenation of formic acid).

Dehydrogenation of Formic Acid Catalyzed by 2: Kinetics and Mechanism

As expected from Scheme 3, complex 2 catalyzes the dehydrogenation of formic acid to H₂ and CO₂. The catalysis was carried out in toluene, in a closed reactor under constant-volume conditions, between 333 and 373 K. To see whether the cycle of Scheme 3 is furthermore catalytic, the kinetics of the catalysis was also studied under pseudo-first-order conditions. The partial pressure of the generated hydrogen (P_H2; atm) was determined according to eq 2, where P_T and P_CO2 are the total pressure and CO₂ partial pressure, respectively.

2

Table 2 collects initial rates obtained from the gas evolution experiments by graphing the expression shown in eq 3, where V is the volume of the reactor (L), R is the molar gas constant, T is the temperature (K), and V_sol is the volume of the catalytic solution (L). Figure 5 exemplifies the reactions completed at 373 K, with a concentration of formic acid of 0.26 M.

3

Table 2. Kinetic Data for the Dehydrogenation of Formic Acid Catalyzed by 2.

T (K)	[cat.] (102 M)	[HCOOH] (M)	d[H2]/dt (102 M min–1)	k (min–1)
373	0.26	0.26	0.241	0.912
373	0.53	0.26	0.589	1.110
373	0.79	0.26	0.830	1.045
373	1.06	0.26	0.895	0.845
373	1.32	0.26	1.277	0.965
373	0.53	0.40	0.589	1.112
373	0.53	0.53	0.465	0.877
373	0.53	0.66	0.459	0.866
373	0.53	0.79	0.477	0.899
363	0.53	0.26	0.306	0.577
353	0.53	0.26	0.159	0.300
343	0.53	0.26	0.100	0.189
333	0.53	0.26	0.088	0.167

Figure 5.

Plot of P_H2 vs time (T = 373 K, [HCOOH]₀ = 0.26 M, 10²[2] = 0.26 M (pale blue asterisks), 0.53 M (purple circles), 0.79 M (green triangles), 1.06 M (blue diamonds), 1.32 M (red squares)).

A general rate law for the dehydrogenation of formic acid catalyzed by 2 is given by eq 4:

4

The rate dependence on the formic acid concentration was studied at 373 K, for a concentration of 2 of 5.3 × 10^–3 M, with variable initial concentrations of carboxylic acid ([HCOOH]₀) between 0.26 and 0.79 M. Under these conditions, the dehydrogenation rate is independent of [HCOOH]₀ (Figure 6), indicating that a = 0 in eq 4. The catalyst dependence was also investigated at 373 K, for [HCOOH]₀ = 0.26 M, increasing the concentration of 2 from 2.6 × 10^–3 to 1.32 × 10^–2 M. In contrast to Figure 6, the plot of ln(d[H₂]/dt) vs ln[2] gives a straight line of slope 1.01 (Figure 7), in accordance with a first-order dependence: i.e., b = 1 in eq 4. Therefore, the rate law for the dehydrogenation of formic acid catalyzed by 2 is

5

Figure 6.

Plot of d[H₂]/dt vs [HCOOH]₀ (T = 373 K, [2] = 0.53 × 10^–2 M).

Figure 7.

Plot of ln(d[H₂]/dt) vs ln [2] (T = 373 K, [HCOOH]₀ = 0.26 M).

A plot of d[H₂]/dt vs [2] (Figure 8) yields for k₅ a value of 0.96 ± 0.10 min^–1 at 298 K. The corresponding Eyring analysis (Figure 4, red points) affords values of ΔH₅^⧧ = 19 ± 1 kcal mol^–1, ΔS₅^⧧ = −7 ± 3 cal K^–1 mol^–1, and ΔG₅^⧧ = 21 ± 2 kcal mol^–1 at 298 K. To our delight, they compare well with those obtained for the stoichiometric release of CO₂ from 5. As it should be, since both Eyring analyses fit the same straight line, which leads to common activation parameters for the catalysis and the stoichiometric release of CO₂ from 5 of ΔH^⧧ = 23 ± 1 kcal mol^–1, ΔS^⧧ = 4 ± 3 cal K^–1 mol^–1, and ΔG^⧧ = 21 ± 2 kcal mol^–1 at 298 K. Figure 4 gives overwhelming evidence supporting the catalytic nature of the cycle shown in Scheme 3, which was furthermore reinforced by the ³¹P{¹H} NMR spectra of catalytic solutions quenched at half of the dehydrogenation. They reveal that complex 5 is the only osmium species present in a detectable concentration while formic acid is not completely consumed. The value of ΔG^⧧ at 298 K lies in the central part of the range previously reported for this parameter in the dehydrogenation of formic acid catalyzed by other homogeneous systems (17–26 kcal mol^–1).^18b,26c,36

Figure 8.

Plot of d[H₂]/dt vs [2] (T = 373 K, [HCOOH]₀ = 0.26 M).

The dehydrogenation of formic acid catalyzed by 2 shows significant differences with regard to those promoted by traditional bifunctional catalysts, from a mechanistic point of view. In contrast to the traditional systems, the hydrogen formation (the fast stage) exclusively occurs on the coordination sphere of the basic metal center, whereas the CO₂ release (the catalytic rate-determining step) is a cooperative process between the metal and the σ-donor Lewis acid. To gain information about the intimate details of the cooperation, we carried out DFT calculations (wB97XD/SDD/cc-pVDZ) with regard to the CO₂ formation. The ΔG values were calculated in toluene at 298.15 K and 1 atm. Figure 9 shows the calculated profiles, whereas Scheme 4 collects the optimized intermediates.

Figure 9.

Energy profiles for the transformation of 5 into 2.

Scheme 4. DFT-Calculated Intermediates for Paths a (Blue) and b (Red).

The stoichiometric decomposition of 5 into 2 and CO₂ is endergonic by 5.4 kcal mol^–1, where the gas elimination acts as a driving force of the reaction. Under catalytic conditions, the absorbed energy is compensated by the exothermic hydrogen formation. The process can be started in two alternative ways (paths a and b), which involve the rupture of the Ge–O bond to form intermediate 6t_a (path a) or the rupture of the Os–O bond to give intermediate 6t_b (path b)_. According to path a, once intermediate 6t_a is formed, the monodentate formate ligand pivots around the Os–O bond to situate its hydrogen on the Ge atom. In this context, it should be noted that a β-hydrogen elimination of the formate hydrogen atom is not possible, due to the saturated character of the metal center. The resulting species 7t_a evolves to 8t by means of the rupture of the formate C–H bond. The subsequent release from the coordinated CO₂ molecule leads to the germyl intermediate 9t, which undergoes α-hydrogen elimination to regenerate 2. This route must overcome an activation energy of 38.0 kcal mol^–1, which is out of the range from the ΔG^⧧ values experimentally obtained. Thus, path a must be rejected as a reasonable proposal. The unsaturated intermediate 6t_b, formed according to path b, saturates the metal coordination sphere by means of an agostic interaction with a methyl C–H bond of the triisopropylphosphine. Its formation takes place with an activation energy of 24.9 kcal mol^–1. This value is the highest of the profile associated with path b and compares well with the experimental value of ΔG^⧧, supporting path b as the only one possible. The formate group, which is now κ¹-coordinated to the Ge atom, pivots around the latter to approach its hydrogen atom to the osmium center. As a result, the formate hydrogen atom displaces the methyl C–H bond from the osmium to afford 7t_b, which finally loses CO₂ to yield 2. The NBO charges at TS_7tb-2 strongly support the basic character of the metal center and the acidic nature of the Ge atom (Chart 2).

Chart 2. NBO Charges at TS_7tb-2.

CO₂ formation usually occurs by β-hydrogen elimination on unsaturated M{κ¹-OC(O)H} intermediates³⁷ or hydrogen abstraction on saturated species of the same type. In the second case, the metal center of the catalyst slips from the coordinated oxygen into the hydrogen atom before releasing the CO₂ molecule.^18a,36b,38 The movement described herein is an alternative manner of implementing the CO₂ release.

Concluding Remarks

An alternative class of bifunctional catalysts can be assembled by coordination of σ-donor Lewis acids to platinum-group-metal basic fragments. In contrast to what happens with the previously reported bifunctional catalysts, this design allows enhancing the basicity of the base and the acidity of the acid. As a proof of this, we have prepared a bifunctional catalyst, for the dehydrogenation of formic acid to H₂ and CO₂, coordinating a bidentate phosphine-tetrylene ligand to an osmium(IV) tetrahydride metal fragment. Furthermore, we have discovered during the preparation of this complex that polyhydrides with a high number of hydride ligands, such as the osmium hexahydride OsH₆(PⁱPr₃)₂, are able to perform the hydrogenolysis of P–C(sp³) bonds.

The kinetics of the formic acid dehydrogenation, the characterization of the resting state of the catalysis along with the establishment of the activation parameters of its evolution, and DFT calculations regarding the rate-determining step prove that the prepared Os(IV)–Ge(II) catalyst works in a manner different from that of the traditional bifunctional catalysts, based on an acid metal fragment and a cooperative basic ligand. In contrast to the traditional systems, the hydrogen formation (the fast stage) exclusively occurs on the coordination sphere of the basic metal center, whereas the CO₂ release (the rate-determining step) is a cooperative processes involving the metal and the σ-donor Lewis acid. The process taking place during the cooperation is also an alternative manner of implementing the CO₂ release.

Experimental Section

General Information

General experimental details, X-ray analysis, spectroscopic and general instrumental techniques, and computational information are given in the Supporting Information. Chemical shifts are expressed in ppm, whereas the coupling constants J and N (N = J_H–P + J_H–P′ for ¹H and N = J_C–P + J_C–P′ for ¹³C{¹H)) are given in Hz. The complex OsH₆(PⁱPr₃)₂³⁹ and 1,3-bis(di-tert-butylphosphanylmethyl)-1,3-dihydro-2λ²-benzo[d][1,3,2]diazagermole^27a were prepared as previously described.

Preparation of 2

The complex OsH₆(PⁱPr₃)₂ (332 mg, 0.643 mmol) and 1,3-bis((di-tert-butylphosphanyl)methyl)-1,3-dihydro-2λ²-benzo[d][1,3,2]diazagermole (350 mg, 0.707 mmol) were dissolved in 10 mL of toluene, in a Schlenk with a Teflon stopcock, and heated at 110 °C for 18 h. The orange solution was transferred to a Schlenk and evaporated to dryness. The crude product was washed with pentane (3 × 2 mL), at −78 °C, to afford a pale yellow solid: yield 213 mg (48%). Crystals suitable for X-ray diffraction analysis were obtained from a concentrated solution of the solid in pentane at −30 °C in the glovebox. Anal. Calcd for C₂₅H₅₂GeN₂OsP₂: C, 42.56; H, 7.43; N, 3.97. Found: C, 42.84; H, 7.47; N, 4.22. HRMS (electrospray, m/z): calcd for C₂₅H₄₉GeN₂OsP₂ [M – 3H]⁺, 705.2192; found, 705.2213. IR (cm^–1): ν(OsH) 2064 (w). ¹H NMR (300 MHz, benzene-d₆, 298 K): δ 7.18–7.02 (m, 2H, Ph), 6.98 (d, ³J_H–H = 7.6, 1H, Ph), 6.76 (d, ³J_H–H = 7.6, 1H, Ph), 3.24 (s, 3H, NCH₃), 2.96 (d, ²J_H–P= 4.1, 2H, NCH₂), 1.72 (sept, ³J_H–H = 6.9, 3H, CH(CH₃)₂), 1.25 (d, ³J_H–P = 12.3, 18H, ^tBu), 1.18 (dd, ³J_H–P = 13.1, ³J_H–H = 6.9, 18H, PCH(CH₃)₂), −10.45 (dd, ²J_H–P = ²J_H’–P′ = 12.4, 4H, OsH₄). ¹H NMR (400 MHz, toluene-d₈, 193 K, high-field region): δ −10.40 (br dd, ²J_H–P = ²J_H′–P′ = 12.8 Hz, 4 H, OsH₄). T₁(min) (ms, 400 MHz, toluene-d₈, 223 K): 220 ± 22 (−10.45 ppm). ³¹P{¹H} NMR (121 MHz, benzene-d₆, 298 K): δ 99.8 (d, ²J_P–P = 224.5), 55.5 (d, ²J_P–P = 224.5). ¹³C{¹H}-APT NMR plus HSQC and HMBC (75 MHz, benzene-d₆, 298 K): δ 144.2 (s, C_q-Ph), 140.3 (d, ³J_C–P = 12.1, C_q-Ph), 119.0 (s, Ph), 117.5 (s, Ph), 110.0 (s, Ph), 108.3.0 (s, Ph), 35.1 (dd, ¹J_C–P = 19.3, ³J_C–P = 1.7, C_q-^tBu), 32.5 (s, NCH₃), 31.8 (d, ¹J_C–P = 25.5, NCH₂P), 30.2 (d, ²J_C–P = 4.0, CH₃-^tBu), 29.4 (dd, ¹J_C–P = 24.5, ³J_C–P = 2.1, CH(CH₃)₂), 21.06 (s, CH(CH₃)₂).

Reaction of 2 with Benzoic Acid

Compound 2 (100 mg, 0.142 mmol) and benzoic acid (19.0 mg, 0.156 mmol) in 5 mL of toluene were stirred for 16 h at room temperature. After evaporation to dryness, 3 mL of pentane was added to form a pale yellow solid of complex 3, which was washed with pentane (3 × 2 mL): yield 66 mg (56%). Crystals suitable for an X-ray diffraction analysis were obtained from a concentrated solution of the solid in pentane at −30 °C in the glovebox. Anal. Calcd for C₃₂H₅₆GeN₂O₂OsP₂: C, 46.55; H, 6.84; N, 3.39. Found: C, 46.31; H, 6.56; N, 3.78. HRMS (electrospray, m/z): calcd for C₃₂H₅₄GeN₂O₂OsP₂ [M – 2H]⁺, 826.2482; found, 826.2478. IR (cm^–1): ν(OsH) 1995 (m); ν(C–O) 1530 (s); 1488 (s). ¹H NMR (300 MHz, toluene-d₈, 298 K): δ 7.92 (dd, ³J_H–H= 8.0, ⁴J_H–H= 1.8, 2H, PhCOO), 7.16 (m, 1H, Ph), 7.10 (m, 1H, Ph), 7.03 (m, 3H, Ph), 6.85 (m, 2H, Ph), 3.53 (s, 3H, NCH₃), 3.24 (dd, ²J_H–P= 12.6, ²J_H–H= 1.8, 1H, NCH₂), 2.53 (dd, ²J_H–P= 12.4, ²J_H–H= 1.8, 1H, NCH₂), 1.78 (m, 3H, PCH(CH₃)₂), 1.22 (dd, ²J_H–P = 13.0, ³J_H–H = 7.1, 9H, PCH(CH₃)₂), 1.06 (dd, ²J_H–P = 13.0, ³J_H–H = 7.1, 9H, PCH(CH₃)₂), 1.12 (d, ³J_H–P = 12.8, 9H, P^tBu₂), 1.01 (d, ³J_H–P = 11.9, 9H, P^tBu₂), −11.55 (br, 3H, OsH₃). ¹H NMR (300 MHz, toluene-d₈, 183 K, high-field region): δ −3.64 (br s, 1H, OsH), −15.12 (AB system, Δν = 1080 Hz, J_A–B = 162 Hz, 2 H, OsH₂). T₁(min) (ms, Os–H, 300 MHz, toluene-d₈, 223 K): 64 ± 6 (−16.69 ppm); the T₁(min) value of the resonance at −4.08 ppm could not be calculated due to its broadness. ³¹P{¹H} NMR (121 MHz, toluene-d₈, 298 K): δ 84.6 (d, ²J_P–P = 236.5), 49.2 (d, ²J_P–P = 236.5). ¹³C{¹H}-APT NMR plus HSQC and HMBC (75 MHz, toluene-d₈, 298 K): δ 176.2 (s, C_q-COO), 145.3 (s, C_q-Ph), 140.6 (d, ³J_C–P = 12.8, C_q-Ph), 140.5 (s, C_q-Ph), 132.1 (s, Ph), 130.5 (s, Ph), 128.3 (s, Ph), 117.5 (s, Ph), 116.5 (s, Ph), 108.5 (s, Ph), 107.4 (s, Ph), 37.1 (d, ¹J_C–P = 22.4, C_q-P^tBu₂), 36.1 (dd, ¹J_C–P = 18.2, ³J_C–P = 2.7, C_q-P^tBu₂), 32.8 (s, NCH₃), 31.2 (d, ¹J_C–P = 26.4, NCH₂), 29.4 (dd, ²J_C–P = 19.3, ⁴J_C–P = 3.3, CH₃-^tBu), 26.7 (dd, ¹J_C–P = 23.6, ³J_C–P = 1.9, CH(CH₃)₂), 20.5 (s, CH(CH₃)₂), 20.0 (s, CH(CH₃)₂).

Reaction of 2 with Acetic Acid

A solution of complex 2 (100 mg, 0.142 mmol) in 5 mL of toluene was treated with glacial acetic acid (9.4 mg, 0.156 mmol, 9 μL). The mixture was stirred for 4 h at room temperature. After evaporation of the solvent, the residue was extracted with pentane (3 × 1 mL). The liquors were evaporated to dryness to afford a pale yellow solid of complex 4: yield 52 mg (48%). Anal. Calcd for C₂₇H₅₄GeN₂O₂OsP₂: C, 42.47; H, 7.13; N, 3.67. Found: C, 42.73; H, 7.39; N, 4.04. HRMS (electrospray, m/z): calcd for C₂₇H₅₂GeN₂O₂OsP₂ [M – 2H]⁺, 764.2325; found, 764.2314. IR (cm^–1): ν(OsH) 2001 (m); ν(C–O) 1546 (s), 1488 (s). ¹H NMR (300 MHz, benzene-d₆, 298 K): δ 7.09 (m, 2H, Ph), 6.89 (m, 2H, Ph), 3.44 (s, 3H, NCH₃), 3.26 (dd, ²J_H–P= 12.7, ²J_H–H= 1.5, 1H, NCH₂), 2.62 (dd, ²J_H–P= 12.3, ²J_H–H= 1.5, 1H, NCH₂), 1.81 (m, 6H, CH₃COO + PCH(CH₃)₂), 1.25–0.95 (m, 36H, PCH(CH₃)₂ + P^tBu₂), −11.65 (br, 3H, OsH₃). ¹H NMR (300 MHz, toluene-d₈, 183 K, high-field region): δ −3.83 (br s, 1H, OsH), −15.37 (AB system, Δν = 1083 Hz, J_A–B = 124.5 Hz, 2 H, OsH₂). T₁(min) (ms, Os–H, 300 MHz, toluene-d₈, 218 K): 41 ± 4 (−17.17 ppm); the T₁(min) value of the resonance at −4.04 ppm could not be calculated due to the broadness of it. ³¹P{¹H} NMR (121 MHz, benzene-d₆, 298 K): δ 84.96 (d, ²J_P–P = 237.5), 48.32 (d, ²J_P–P = 237.5). ¹³C{¹H}-APT NMR plus HSQC and HMBC (75 MHz, benzene-d₆, 298 K): δ 181.6 (s, C_q-COO), 145.5 (s, C_q-Ph), 140.7 (d, ³J_C–P = 12.8, C_q-Ph), 117.4 (s, Ph), 116.6 (s, Ph), 108.6 (s, Ph), 107.4 (s, Ph), 37.2 (d, ¹J_C–P = 22.7, C_q-P^tBu₂), 36.3 (d, ¹J_C–P = 17.7, C_q-P^tBu₂), 32.7 (s, NCH₃), 31.4 (d, ¹J_C–P = 26.5 Hz, NCH₂), 29.6 (dd, ²J_C–P = 25.4, ⁴J_C–P = 3.4, CH₃-^tBu), 27.0 (d, ¹J_C–P = 23.4, CH(CH₃)₂), 20.3 (d, ²J_C–P = 17.1, CH(CH₃)₂), 19.8 (s, CH₃COO).

Reaction of 2 with Formic Acid: Characterization of 5

Compound 2 (10 mg, 0.0142 mmol) and formic acid (0.7 mg, 0.0156 mmol, 0.6 μL) were dissolved in 0.5 mL of toluene-d₈ in a Young-NMR tube. ¹H NMR (300 MHz, toluene-d₈, 298 K): δ 7.01 (m, 2H, Ph), 6.81 (m, 2H, Ph), 6.63 (s, 1H, HCOO), 3.43 (s, 3H, NCH₃), 3.21 (dd, ²J_H–P= 12.1, ²J_H–H= 1.7, 1H, NCH₂), 2.49 (dd, ²J_H–P= 12.3, ²J_H–H= 1.7, 1H, NCH₂), 1.74 (m, 3H, PCH(CH₃)₂), 1.18 (dd, ³J_H–P = 12.6, ³J_H–H = 6.9, 9H, PCH(CH₃)₂), 1.12 (d, ³J_H–P = 12.7, 9H, P^tBu), 1.05 (dd, ³J_H–P = 12.8, ³J_H–H = 6.9, 9H, PCH(CH₃)₂), 0.96 (d, ³J_H–P = 12.1, 9H, P^tBu), −11.73 (br, 3H, OsH₃). ¹H NMR (300 MHz, toluene-d₈, 183 K, high-field region): δ −4.16 (br s, 1H, OsH), −15.66 (AB system, Δν = 1175 Hz, J_A–B = 153.0 Hz, 2 H, OsH₂). T₁(min) (ms, Os–H, 300 MHz, toluene-d₈, 223 K): 64 ± 6 (−17.28 ppm); the T₁(min) value of the resonance at −4.32 ppm could not be calculated due to its broadness. ³¹P{¹H} NMR (121 MHz, toluene-d₈, 298 K): δ 84.1 (d, ²J_P–P = 234.6), 48.3 (d, ²J_P–P = 234.6). ¹³C{¹H}-APT NMR plus HSQC and HMBC (75 MHz, toluene-d₈, 298 K): δ 173.2 (s, HCOO), 145.5 (s, C_q-Ph), 140.7 (d, ³J_C–P = 12.9, C_q-Ph), 117.9 (s, Ph), 117.0 (s, Ph), 108.9 (s, Ph), 107.8 (s, Ph), 37.6 (d, ¹J_C–P = 22.7, C_q-^tBu), 36.8 (dd, ¹J_C–P = 18.3, ³J_C–P = 2.8, C_q-^tBu), 33.0 (s, NCH₃), 31.4 (d, ¹J_C–P = 26.5, NCH₂), 29.8 (dd, ³J_C–P = 12.4, ⁵J_C–P = 3.3, CH₃-^tBu), 27.0 (dd, ¹J_C–P = 23.7, ³J_C–P = 1.9, CH(CH₃)₂), 20.4 (s, CH(CH₃)₂).

General Procedure for the Formic Acid Dehydrogenation Studies

The progress of the reaction was monitored using a reactor equipped with a pressured transducer (Man on the Moon series X103 kit; https://www.manonthemoontech.com/x102-gas-evolution.html). The total volume of the reactor was 18.8 mL. The procedure employed was as follows: under an argon atmosphere, a solution of 2 in 1 mL of toluene was incorporated into the reactor, which was closed and placed in an oil bath at the desired temperature. Then, the pressure was monitored until a stable value was reached and the reactor tared. Subsequently the formic acid was injected through a septum cap. This moment was considered the initial time of the catalysis. The reaction was followed by measuring the total pressure as a function of the time.

NMR Spectroscopic Studies of the Decarboxylation of Complex 5

The decarboxylation of complex 5 was monitored by ³¹P{¹H} NMR spectroscopy (162.0 MHz). NMR spectra were acquired in a temperature range from 318 to 338 K at different times. A 0.5 mL portion of a 0.0142 M solution of 2 in toluene-d₈ was placed in a Young NMR tube with a capillary containing a solution of PPh₃ as an internal standard. The addition of 10 μL of a 0.7 M solution of HCOOH afforded the instantaneous formation of 5. The ³¹P{¹H} NMR parameters were modified as follows for the integration of the signal: pulse program (zgig30), d₁ ≥ 5T₁ (d₁ = 33 s). Prior to the study, the relaxation times of 2, 5, and PPh₃ were determined by varying the delay times.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.inorgchem.1c02893.

• General information, structural analysis of complexes 2 and 5, computational data, and NMR spectra (PDF)

• Cartesian coordinates of computed complexes (XYZ)

Accession Codes

CCDC 2096535–2096536 contain the supplementary crystallographic data for this paper. These data can be obtained free of charge via www.ccdc.cam.ac.uk/data_request/cif, or by emailing data_request@ccdc.cam.ac.uk, or by contacting The Cambridge Crystallographic Data Centre, 12 Union Road, Cambridge CB2 1EZ, UK; fax: +44 1223 336033.

The authors declare no competing financial interest.