Inner- versus Outer-Sphere Ru-Catalyzed Formic Acid

Article pubs.acs.org/Organometallics

Inner- versus Outer-Sphere Ru-Catalyzed Formic Acid Dehydrogenation: A Computational Study Gabriele Manca,† Irene Mellone,†,‡ Federica Bertini,† Maurizio Peruzzini,*,† Luca Rosi,†,‡ Dörthe Mellmann,§ Henrik Junge,§ Matthias Beller,*,§ Andrea Ienco,† and Luca Gonsalvi*,† †

Consiglio Nazionale delle Ricerche, Istituto di Chimica dei Composti Organometallici (CNR-ICCOM), Via Madonna del Piano 10, 50019 Sesto Fiorentino (Firenze), Italy ‡ Università degli Studi di Firenze, Dipartimento di Chimica, Via della Lastruccia 3-13, 50019 Sesto Fiorentino, (Firenze), Italy § Leibniz-Institute for Catalysis (LIKAT) at the University of Rostock, Albert-Einstein Strasse 29a, 18059 Rostock, Germany S Supporting Information *

ABSTRACT: A detailed hybrid density functional theory study was carried out to clarify the mechanism of Ru-catalyzed dehydrogenation of formic acid in the presence of the octahedral complexes [Ru(κ4-NP3)Cl2] (1) and [Ru(κ3-triphos)(MeCN)3](PF6)2 (2·PF6) [NP3 = N(CH2CH2PPh2)3, triphos = MeC(CH2PPh2)3]. It was shown that Ru-hydrido vs Ru-formato species are pivotal to bringing about the efficient release of H2 and CO2 following either a metal-centered (inner-sphere) or a ligand-centered (outer-sphere) pathway, respectively.

■

storage cycle can be envisioned.4 In recent years, many different heterogeneous and homogeneous catalyst systems for the dehydrogenation of formic acid have been studied. In the case of homogeneous catalysts, transition metals, such as Fe, Co, Ir, Rh, and Ru, have been so far the most widely investigated for this reaction, often in association with ancillary ligands such as phosphines. At the current state of the art, interesting results have been reported by Wills,5 Puddephatt,6 and recently by Laurenczy,7 Beller,4i,8 and respective groups, using ruthenium phosphine complexes. Very good results have been also achieved with homogeneous catalysts based on Fe9 and Co10 in the presence of polydentate phosphines, such as P(CH2CH2PPh2)3 and PP3, and NMR spectroscopic studies together with density functional theory (DFT) calculations have been performed to establish the reaction mechanisms.9a In a recent publication,11 we reported on the rutheniumcatalyzed dehydrogenation of formic acid/amines in the presence of facially capping stabilizing ligands, such as the tridentate 1,1,1-tris-(diphenylphosphinomethyl)ethane (triphos) and the tetradentate tris-[2-(diphenylphosphino)ethyl]amine (NP3). In particular, using the octahedral complexes [Ru(κ4NP3)Cl2] (1) and [Ru(κ3-triphos)(MeCN)3](OTf)2 (2), it was shown that good turnover numbers (TONs) could be obtained for the dehydrogenation of formic acid, up to 10 000 using 0.01 mol % of catalysts. The recyclability of 2 was also proven

INTRODUCTION In the quest for alternative energy vectors to fossil fuels, hydrogen is now considered as one of the possible candidates, as it can be successfully combined with today’s mature fuel cell technology.1 However, the use of pressurized gas for stationary and mobile applications is not cost-effective and has inherent safety issues. The development of efficient technologies for hydrogen generation from renewables and its storage in a safe and reversible manner are thus prerequisites for its wide-scale utilization as a fuel.2 Among many methods explored, the use of hydrogen-rich organic chemicals, from which hydrogen can be released reversibly by catalytic dehydrogenation, is receiving increasing interest in the scientific literature.3 Compounds such as methanol and higher alcohols are considered as promising hydrogen storage materials, and the current goal is to decrease the harsh temperature conditions often required to achieve high activities in their dehydrogenation to produce H2 and the related aldehyde. An alternative organic compound that has found increasing interest as a hydrogen storage material is formic acid, a liquid at ambient conditions having 4.4% in weight of hydrogen, that can be safely handled, stored, and transported easily. The dehydrogenation of HCOOH gives gaseous mixtures of H2 and CO2 in the presence of suitable catalysts, which are often endowed with high selectivity, i.e., avoiding the competing dehydration reaction yielding CO (a known poison for fuel cell catalysts) and water. Furthermore, the CO2 obtained as a byproduct can be, in principle, rehydrogenated to HCOOH, so a zero-carbon emission energy © 2013 American Chemical Society

Received: July 31, 2013 Published: November 25, 2013 7053

dx.doi.org/10.1021/om400761t | Organometallics 2013, 32, 7053−7064

Organometallics

Article

Scheme 1. Proposed Catalytic Cycles for HCOOH Dehydrogenation in the Presence of 1 Based on VT NMR Studies11

amount of [Ru(κ4-NP3)(η2-H2)(H)]+ (5). Finally, upon addition of neat formic acid, all the species previously observed, except for a small amount of 5, disappeared to leave a new pattern that we assigned to [Ru(κ4-NP3)(η2-H2)Cl] (6), which may reasonably be obtained by protonation of the hydrido ligand in 4. By leaving the NMR tube at room temperature overnight, the concentration of 6 decreased to regenerate complex 1. On the basis of these results, and by comparison with literature data, two interconnected catalytic cycles A and B centered on 4 were proposed to account for the formation of H2 and CO2 starting from HCOOH (Scheme 1). In the case of 2·PF6, the same experiment described above was carried out in d6-acetone, chosen for solubility reasons. Although hydrogen development was clearly detectable in the 1 H NMR spectra collected during the experiment, no evidence of formation of stable hydrido complexes could be observed. The major species formed upon reaction of 2·PF6 with (NHEt3)(HCO2) gave a pattern compatible with acetonitrile substitution by formate, giving equilibria including [Ru(κ3-triphos)(MeCN)(η 2 -OOCH)] (7) and [Ru(κ 3 -triphos)(η 1 -OOCH)(η 2 OOCH)] (8). Preliminary DFT calculations on a simplified model of 2·PF6 in acetone solution suggested that the reaction with HCO2− to give 7 is favored with a ΔG298 = −9.74 kcal mol−1 and further stabilization (ΔG298 = −5.24 kcal mol−1) is gained by reaction of 7 with HCO2− to give 8 (Scheme 2). These observations left open the question on how hydrogen could be evolved by such formate intermediates, and why the triphos-based and the NP3-based systems behaved in such an apparently different way in NMR experiments. Thus, we decided to extend the preliminary DFT calculations in order to model the two catalytic systems based on 1 and 2·PF6 to better understand the nature, stability, and activation pathways of the intermediates present in the two systems, and the results of such a study are hereby presented.

Scheme 2. Proposed Intermediates Formed by Reaction of 2·PF6 with (NHEt3)(HCO2) Based on NMR Evidence11

using 0.1 mol % of catalyst for eight consecutive runs with minor loss of activity. Preliminary mechanistic variable-temperature (VT) NMR studies on model reactions of formic acid and formate with 1 and the hexafluorophosphate analogue of 2, namely, [Ru(κ3triphos)(MeCN)3](PF6)2 (2·PF6), were reported, showing that different species could be detected as intermediates of the reactions. More in detail, in the case of 1, the reaction with (NHEt3)(HCO2) in CD2Cl2 proceeded at first with the formation of a substitution complex identified as the formato complex [Ru(κ4-NP3)(η1-OOCH)Cl] (3), which evolved into the known stable hydrochloride complex [Ru(κ4-NP3)(H)Cl] (4) by release of CO2. Further addition of formate caused the complete disappearance of 1, a higher concentration of 3 evolving again into 4 and also the comparison of a small 7054


Organometallics

■

Article

COMPUTATIONAL METHODOLOGY

complex 3 observed by NMR was formed in a 1:1 ratio to 1. In principle, two different coordination modes for formate are possible: a chelating mode η2-3m+ (Figure 2a), implying the dissociation of a second chloride ligand, and a monohapto coordination η1-3m, for which two configurational isomers are possible, namely, with the formate substituent trans to N, modeled as η1-3mN (Figure 2b) or trans to phosphorus, η1-3mP (Figure 2c). The main structural features of the three structures are summarized in Table 1. To identify the nature of the most abundant product in solution, the free energy contributions at 298 K (ΔG298, kcal mol−1) to generate the formate derivatives have been calculated. Therefore, despite the small energy differences among the three species 3m (ΔG298 for η2-3m+, η1-3mN, and η1-3mP were calculated as +1.13, −0.81, and −2.73 kcal mol−1, respectively), it can be safely assumed that the most stable species is η1-3mP. The stronger trans influence of phosphorus compared to nitrogen justifies the formation of η1-3mP associated with a significant elongation (0.77 Å) of the Ru−O bond compared to isomer η1-3mN. During the described NMR experiment, upon heating to 273 K, 31P{1H}NMR and 1H NMR measurements showed unequivocally the formation of the hydrido-chloride derivative 4. Again, two different isomers can form, in which, alternatively, the hydride lies opposite to the phosphorus or nitrogen atoms, and these structures were optimized as 4mP and 4mN (Figure 3a,b, respectively). The former resulted in being slightly more stabilized (ΔG298 = −0.71 kcal mol−1). Such a result is not unexpected since the incoming hydride ligand should bind trans to the donor atom with the highest trans influence.18 Although the calculated energy differences are not particularly high, the structural details (see the Supporting Information) match the previously reported NMR data. The Ru−P2 bond (trans to H) in 4mP was ca. 0.1 Å more elongated than that in 4mN. The energies for the reaction pathway leading from η1-3mP to 4mP were then calculated. Since a hydride ligand must be formed in the coordination sphere of ruthenium, a possible mechanism may involve an interaction between the terminal hydrogen of the formate and the metal, followed by CO2 elimination.9a We could rule out an associative mechanism for this step as all efforts in optimizing a hepta-coordinated Ru intermediate were unsuccessful. To obtain the formation of a Ru−H bond, as in the experimentally observed complex 4, a dissociative mechanism involving a penta-coordinated intermediate, derived from prior dissociation of a second chloride ligand, namely, [Ru(κ4-NP3)(η1-OOCH)]+, was thus considered. This step was found to be endoergonic by only 6.46 kcal mol−1, with a long separation between the metal center and the formate hydrogen (4.042 Å). The process resulted in being endothermic (ΔH = +15.2 kcal mol−1), and the entropic contribution (−TΔS = −8.7 kcal mol−1) only partially lowers such an energy, with a net free energy cost of +6.46 kcal mol−1. The dissociation of the chloride ligand might be favored in the real system by the presence of the triethylammonium cation in solution, to form the corresponding salt, poorly soluble in CD2Cl2, and contribute to the driving force for this step. In the calculations, a transition state 3mTS+, shown in Figure 4, was isolated and validated by the presence of an imaginary frequency at −89.9 cm−1, which is indicative of the formation of the Ru−H bond (see the movie in the Supporting Information). This TS exhibits an agostic Ru−H interaction with a distance of 2.154 Å, accompanied by a significant elongation of the formate C−H bond (1.162 Å). The energy barrier for the process was

The model systems reported herein were optimized at the hybrid density functional theory (DFT) level using the B3LYP functional.12 All the DFT calculations were carried out using the Gaussian 09 package.13 The nature of stationary points (minima or transition state) was confirmed for all of the fully optimized structures through vibrational frequencies calculations. The pseudopotential Stuttgart− Dresden14 was used for the Ru center, while the 6-31G basis set with the important addition of the polarization functions (d, p) was used for all atoms including hydrogens. All the structures were optimized in solution using the Conductor-Like Polarizable Continuum Model (CPCM)15 as implemented in the Gaussian 09 package,13 and, in particular, in dichloromethane solution for 1 and in acetone for 2·PF6. The Cartesian coordinates of the optimized structures are reported in the Supporting Information.

■

RESULTS AND DISCUSSION The Mechanism of Formic Acid Dehydrogenation Catalyzed by [Ru(κ4-NP3)Cl2] (1). The computational analysis was started from the structure optimization of 1. In this compound, the Ru atom is located at the center of an octahedron, with four vertices occupied by one nitrogen and three phosphorus atoms of the tetrapodal NP3 ligand and two remaining positions filled by two chloride atoms in a cis position. A model structure 1m (Figure 1) was used in place of

Figure 1. Optimized structure for 1m. Hydrogen atoms on the capping ligand omitted for clarity. Main bond lengths (Å): Ru−N = 2.224, Ru−P2 = 2.275, Ru−Cl1 = 2.607, Ru−Cl2 = 2.510, Ru−P1 = 2.377, and Ru−P3 = 2.400.

1, with methyl substituents on phosphorus atoms instead of the more bulky phenyl ones, in order to significantly decrease the computational time. Steric effects due to phenyl groups could be, in this way, underestimated; however, 1m was compared with an available X-ray structure of 1 found on theCCDC database16 (refcode IFEZUO).17 A good agreement between the calculated structure 1m and the X-ray structure of 1 was observed, with Ru−N distances of 2.224 Å (1m) vs 2.188(7) Å (1); Ru−Cl1 2.607 Å (1m) vs 2.479(2) Å (1); Ru−Cl2 2.510 Å (1m) vs 2.447(2) Å (1); and Ru−P2 2.275 Å (1m) vs 2.254(2) Å (1). In the calculated structure 1m, the Ru−P1 and Ru−P3 distances were ca. 0.11 Å longer than Ru−P2. Interestingly, the different trans influence18 expected for the phosphorus and nitrogen donor atoms shows well from the comparison of Ru−Cl distances in 1m, with Ru−Cl1 (Cl trans to P) ca. 0.1 Å longer than the Ru−Cl2 bond length (Cl trans to N). The first step in the proposed catalytic cycle (Scheme 1) is the substitution of one chloride with a formate ligand. At 233 K, using a Ru/formate 1:1 ratio, the putative formate 7055


Organometallics

Article

Figure 2. Optimized structures for η2-3m+ (a), η1-3mN (b), and η1-3mP (c).

Table 1. Main Calculated Structural Features for η2-3m+, η1-3mN and η1-3mP η2-3m+

η1-3mN

As described above, further addition of formic acid (5 equiv) at 233 K to the reaction mixture in the NMR tube11 yielded as the main product the new complex identified as [Ru(κ4NP3)(η2-H2)Cl] (6), from which hydrogen was released together with regeneration of 1 by increasing the temperature. To establish the pathway responsible for the formation of 6, different possibilities were postulated. First, the formic acid proton has two potential attack sites on complex 4m, namely, the hydride and the chloride ligands. The experimental results showed, however, the exclusive formation of the Ru(η2-H2) derivative, from protonation of the hydrido ligand, which was thus modeled as structure 6m+, where the Ru octahedral geometry features a κ4-NP3 ligand, a chloride, and an η2-H2 nonclassical molecular hydrogen ligand. Also, in this case, two isomers could form, in principle, namely, with η2-H2 trans to N (6mN+, Figure 5a) and trans to P (6mP+, Figure 5b); therefore, both structures were optimized and their relative energies were compared. Structure 6mN+ is estimated to be −3.88 kcal mol−1 more stable than 6mP+, as expected from the better trans effect of P compared to N. Ru−(η2-H2) distances were calculated as 1.702 Å (6mN+) vs 1.778 Å (6mP+), whereas H−H distances of 0.872 and 0.819 Å were obtained, respectively. This bond length difference should involve a blue shift of the associated H−H stretching frequencies. As expected, the corresponding calculated IR frequencies (2745.9 cm−1 for 6mN+ vs 3293.2 cm−1 for 6mP+,) follow this trend and suggest a decreased back-donation from the Ru(dπ) to (H2)σ* orbital. In passing from 4mP to 6mN+, an accessible transition state 4mTS was optimized for the approach of formic acid to 4mP.

η1-3mP

bond lengths (Å) Ru−P2 Ru−P1 Ru−P3 Ru−N Ru−O1 Ru−O2 Ru−Cl C−O CO C−H

2.272 2.375 2.402 2.231 2.309 2.185 1.265 1.272 1.099

2.275 2.383 2.400 2.228 2.142

2.278 2.363 2.395 2.222 2.219

2.604 1.282

2.513 1.283

1.245 1.111

1.247 1.105

estimated as 15.0 kcal mol−1. Such a TS exhibits the Ru−H interaction trans to nitrogen, and all the efforts to isolate a TS with H trans to P failed. The following step would involve the formation of a hexacoordinated complex with a Ru-bound hydride and a slightly interacting bent CO2 ligand. This reaction is favored by −4.62 kcal mol−1 with a significant elongation of the Ru−O1 bond of about 0.1 Å. Elimination of CO2 further stabilizes the system by −14.1 kcal mol−1, to give 4mP by coordinating back the chloride ligand, in agreement with the experimental data. The overall reaction pathway is shown in Scheme 3. 7056


Organometallics

Article

Figure 3. Optimized structures for 4mP (a) and 4mN (b). Main bond lengths (Å) for 4mP: Ru−H = 1.645, Ru−Cl = 2.562, Ru−P2 = 2.379, and Ru−N = 2.222. Main bond lengths (Å) for 4mN: Ru−H = 1.618, Ru−Cl = 2.616, Ru−P2 = 2.258, and Ru−N = 2.350.

a protonation site. A similar TS was calculated for the waterassisted dihydrogen elimination from a Ru(PNN)19a and a series19b of group 6 hydrido complexes. An imaginary frequency at −93.1 cm−1, relative to the shift of the formic acid proton toward the hydride ligand, validates the nature of the TS. The energy barrier for accessing such a transition state was estimated as +7.46 kcal mol−1. The step from 4mTS to 6mP+ gave a calculated energy gain of −0.98 kcal mol−1. Elimination of H2 and coordination of chloride restores the initial complex 1m with a free energy gain of −10.0 kcal mol−1 (Scheme 4). The mechanism described in Scheme 1 (cycle A) was reconsidered, and an additional VT NMR study was carried out (see the Supporting Information for details). Complex 4 was synthesized following literature procedures20 and dissolved in CD2Cl2. The corresponding 31P{1H} NMR spectrum collected at 233 K showed the known AM2 spin system composed by a PM doublet at 41.7 ppm (2JPP = 18 Hz) and a PA triplet at 27.4 ppm. To this solution was added (HCOO)(NHEt3) (1 equiv) under nitrogen at 233 K. The corresponding 31P{1H} NMR spectrum showed the appearance of a new set of signals, at 44.8 ppm (t, 2JPP = 29 Hz, PA) and 23.8 ppm (d, PM), which was attributed to [Ru(κ4-NP3)H(η1-OOCH)] (9) and formed in a 1:2 ratio with 4. Complex 5 characterized by signals at 55.2 ppm (t, 2JPP = 17.4 Hz, PM) and 50.2 ppm (t, PA) formed in traces under these conditions. By increasing the temperature, the ratio between 4 and 9 gradually changed in favor of 9 while the concentration of 5 remained almost unchanged. In detail, at 253 K, the 9:4 ratio was calculated as 2.5:1, reaching the value 5.6:1 after 30 min. At 273 K, the signals due to 9 began slowly to decrease. At 293 K, the 9:4 ratio was 1:2, and after 1 h became 1:4. At this point, the tube was cooled again to 233 K and a second equivalent of (HCOO)(NHEt3) was added. The reactivity previously described was again observed, and again 5 increased only in small quantities. Finally, addition of HCOOH (1 equiv to initial 4) was added at 233 K. This caused the complete conversion of 4 to 9, without significantly affecting the concentration of 5. No other changes were observed by raising the temperature to 293 K. After one night at room temperature, the NMR spectrum showed almost only 4 to have formed again as a major species. Resorting again to DFT calculations, the reaction pathway involving release of H2 and CO2 from 4mP by reaction with formate and formic acid was then modeled. The first step would involve chloride substitution in 4mP by formate anion to generate 9mP, as observed experimentally. This process was

Figure 4. Optimized structure of 3mTS+. Main bond lengths (Å): Ru−H = 2.154, Ru−O1 = 2.209, Ru−P2 = 2.295, and Ru−N = 2.174.

Scheme 3. Free Energy Reaction Pathway Leading from 1m to 4mP

The resulted structure of 4mTS, shown in Figure 6, exhibited a relatively short H1−H2 distance of 1.944 Å, indicative of an interaction between the two ligands. The Cl···H2 separation (2.445 Å) appeared too elongated to account for chloride protonation, confirming the initial hypothesis for the choice of 7057


Organometallics

Article

Figure 5. Optimized structures of 6mN+ (a) and 6mP+ (b). Main bond lengths (Å) for 6mN+: Ru−(η2-H2) = 1.702, H−H = 0.872, Ru−Cl = 2.543, Ru−P2 = 2.304, and Ru−N = 2.236. Main bond lengths (Å) for 6mP+: Ru−(η2-H2) = 1.778, H−H = 0.819, Ru−Cl = 2.491, Ru−P2 = 2.325, and Ru−N = 2.229.

through the transition state 3mTS+ (ΔG = +15.0 kcal mol−1) to form a Ru−H bond and eliminate CO2 (ΔG = −4.6 kcal mol−1). Finally, coordination of chloride (ΔG = −3.9 kcal mol−1) could give back compound 4mP. The free energy reaction pathway associated with this mechanism is shown in Scheme 5. The possible involvement of 5 in cycle A was finally reconsidered. This species was modeled by DFT calculations, and two structural isomers 5mP+ and 5mN+ (Figure 7) emerged, having the hydride ligand either trans to P or trans to N, respectively. Structure 5mP+ was estimated to be more stable by −3.63 kcal mol−1, in view of the more pronounced trans effect of phosphorus, as pointed out by comparison of the optimized Ru−P bonds, with Ru−P2 (trans to hydride) being more elongated than the other two (2.398 vs 2.350 Å). Many different mechanisms generating 5mN+ from 4mP were considered. The most likely should involve the release of CO2 and formation of a Ru−H bond from [Ru(κ4-NP3)(η2-H2)(η1OOCH)] (10mP). However, this pathway provided too high barriers (+35.0 kcal mol−1) to explain the formation of compound 5mN+ even at low temperature (233 K). All the computational efforts to localize a transition state leading to 5mN+ failed. On the basis of NMR and DFT data, it can be concluded that 5 may not be associated with the catalytic cycle of HCOOH dehydrogenation based on the Ru-NP3 system, but rather form as a stable side-product under the reaction conditions applied in the NMR experiments. Consequently, Scheme 1 should be more correctly redrawn as below (Scheme 6). The Mechanism of Formic Acid Dehydrogenation Catalyzed by [Ru(κ3-triphos)(MeCN)3]2+. The facially capping triphosphine 1,1,1,-tris-(diphenylphosphinomethyl)ethane (triphos) has been largely used in coordination chemistry and catalysis due to its strong ability to stabilize transition-metal complexes by its three P donor atoms. Also, in the case of HCOOH dehydrogenation, the Ru(II) complex [Ru(κ3-triphos)(MeCN)3](OTf)2 proved to be an efficient catalyst.11 The computational investigation started by modeling the cationic species [Ru(κ3-triphos)(MeCN)3]2+, where the peripheral phenyl rings were substituted by simpler methyl substituents (triphosMe). The corresponding structure 2m2+ was optimized in acetone solution, the same solvent used for the VT NMR experiments. The structure, shown in Figure 8, although with some slight overestimation due to the pseudopotential usage for Ru,21 could be reasonably compared with that of [Ru(κ3sulphos)(MeCN)3]2+ (see Cambridge Structural Database,16 refcode WIBKOH),22 where the sulphos ligand differs from triphos for the presence of a CH2(C6H4SO3−) moiety replacing

Figure 6. Optimized structure of transition state 4mTS, obtained by interaction of 4mP with a formic acid molecule. Main bond lengths (Å): Ru−H1 = 1.652, Ru−Cl = 2.540, H1−H2 = 1.944, Cl−H2 = 2.445, Ru−P2 = 2.379, and Ru−N = 2.230.

Scheme 4. Reaction Pathway Leading from 4mP to 1m

calculated to be favored by −0.53 kcal mol−1 at 233 K. Then, a protonation step could occur, leading to [Ru(κ4-NP3)(η2-H2)(η1-OOCH)] (10mP). The acidity source in solution is likely to stem from the alkyl ammonium NHEt3+ and/or traces of undissociated HCOOH. The η2-H2 ligand should be trans to P rather than to N, due to better trans influence. Upon the temperature increase, H2 should be easily displaced, giving the ruthenium penta-coordinated intermediate [Ru(κ4-NP3)(η1-OOCH)] (free energy gain of −8.3 kcal mol−1). The free coordination site may be then available for a β-elimination step 7058


Organometallics

Article

Scheme 5. Free Energy Reaction Pathway for the Production of CO2 and H2 from 4mP

Figure 7. Optimized structures of 5mP+ (a) and 5mN+ (b). Main bond lengths (Å) for 5mP+: Ru−H = 1.603, Ru−H2 = 1.672, H−H = 0.920, Ru−P2 = 2.39, and Ru−N = 2.246. Main bond lengths (Å) for 5mN+: Ru−H = 1.609, Ru−H2 = 1.707, H−H = 0.897, Ru−P2 = 2.333, and Ru−N = 2.340.

the bridgehead methyl group. The optimized structure shows three identical Ru−P distances of ca. 2.336 Å (average 2.3118(18) Å in the experimental one), while the Ru−N bond lengths are in the range from 2.115 to 2.120 Å (average 2.099(6) Å in the experimental one), with the Ru-coordinated acetonitrile NC stretching frequency estimated at 2383.3 cm−1. In this system, the three labile solvento-ligands are all potentially prone to substitution with incoming formate, which can, in principle, bind Ru in both η2- and η1-modes. A useful tool to assess the matching between calculations and experimental results is the comparison between the calculated and the experimental IR spectra. In particular, upon sequential substitution of one or more MeCN ligands in 2m2+ with formate, the intensity of the corresponding CN stretching bands was expected to decrease, associated with the appearance of CO carboxylate stretching bands in the range of 1546−1697 cm−1, due to η2- and η1-formate coordination to Ru. Five octahedral structures were, therefore, optimized taking in account the possible substitution degrees and coordination modes, namely, [Ru(κ3triphosMe)(MeCN)(η2-OOCH)]+ (7m+), [Ru(κ3-triphosMe)(η1-OOCH)(η2-OOCH)] (8m), [Ru(κ3-triphosMe)(MeCN)2(η1-OOCH)]+ (11m+), [Ru(κ3-triphosMe)(MeCN)(η1-OCH)2] (12m), and [Ru(κ3-triphosMe)(η1-OOCH)3]− (13m−), shown in Figure 9. Table 2 summarizes the calculated bond lengths associated with these species.

Figure 8. Optimized structure of compound 2m2+. Main bond lengths (Å): Ru−N = 2.117 (average) and Ru−P = 2.336 (average).

Triphos being a symmetrical ligand, no trans influence arising from different donor atoms coordinating the metal center should be considered, as for 1m and derivatives, simplifying the overall modeling. In structure 11m+ (Figure 9c), two Ru-coordinated acetonitrile ligands are maintained and one is replaced by an η1-coordinate formate anion. The formation of compound 11m+ from 2m2+ was calculated to be exoergonic 7059


Organometallics

Article

Scheme 6. Catalytic Cycles for HCOOH Dehydrogenation in the Presence of 1 Based on VT NMR and DFT Studies

Table 2. Main Calculated Structural and Spectroscopic Features for 7m+, 8m, 11m+, 12m, and 13m− 7m+

8m

2.285 2.288 2.322 2.101

2.275 2.276 2.299

11m+

12m

13m−

bond lengths (Å) Ru−P1 Ru−P2 Ru−P3 Ru−N1 Ru−N2 Ru−O1 Ru−O2 Ru−O3 C−O CO C−H

1.100

νCO

1583.2

νC−H

3053.1

νCN

2372.4

2.273 2.272 1.269

2.324 2.330 2.313 2.113 2.109 2.195

2.278 2.272 2.181 1.267−1.282 1.283 1.248 1.246 1.101−1.108 1.108 computed IR stretching frequencies (cm−1) 1588.3 1635.0 1629.7 2949.1 3040.2

by −7.4 kcal mol−1. The second isomer, 7m+ (Figure 9a), showed a coordinated CH3CN molecule and the formate binding the metal in η2-mode. The reaction from 2m2+ to 7m+ was favored by −11.5 kcal mol−1. In the presence of a 3:1 excess of formate, the formation of a neutral compound 8m bearing both η1-OOCH and η2-OOCH (Figure 9b) would also be plausible. The calculated Ru−O distances were in the range between 2.181 Å (for η1-OOCH) and 2.278 Å (for η2-OOCH); thus, two CO stretching frequencies should be present, differing by ca. 50 cm−1 (1630 and 1584 cm−1 for η1-OOCH and η2-OOCH, respectively). The free energy associated with the formation of 8m from 2m2+ was calculated as −16.5 kcal mol−1. The other possible structures

2.302 2.309 2.325 2.105

2.298 2.294 2.293

2.202 2.197

2.204 2.198 2.198 1.277 1.251 1.107

1.279 1.250 1.106 1617.0 1634.2

2952.2

2964.8 2991.0

2377.5

2373.5

1608.2 1618.7 1640.3 2855.4 2959.3 3005.1

obtained by further MeCN substitution, namely, 12m (Figure 9d) and 13m− (Figure 9e), should be less relevant, in view of their higher free energies of formation, at −8.01 and −8.02 kcal mol−1, respectively. The computational results described above confirm the experimental VT NMR tests involving 2·PF6, carried out in acetone-d6 solution at different Ru/formate ratios of 1:1, 1:2, and 1:3, respectively.11 In the first experiment, an initial mixture of 2·PF6 and 8 was observed, to give exclusively 8 upon further addition of formate (second and third experiments) at 233 K. Upon heating and after addition of HCOOH, hydrogen gas was evolved, as shown by the appearance of a singlet at 4.6 ppm in the 1H NMR spectra. 7060


Organometallics

Article

Figure 9. Optimized structures for (a) [Ru(κ3-triphosMe)(MeCN)(η2-OOCH)]+ (7m+), (b) [Ru(κ3-triphosMe)(η1-HCO2)(η2-OOCH)] (8m), (c) [Ru(κ3-triphosMe)(MeCN)2(η1-OOCH)]+ (11m+), (d) [Ru(κ3-triphosMe)(MeCN)(η1-OOCH)2] (12m), and (e) [Ru(κ3-triphosMe)(η2-OOCH)3]− (13m−).

The mechanism involved in hydrogen release from species such as 7m+, 8m, and 11m+ was, therefore, studied. All attempts to obtain a transition state corresponding to an innersphere, metal-centered mechanism similar to the Ru-NP3

system (Scheme 4) failed. It was indeed found that an optimized structure such as [Ru(κ3-triphosMe)(H)(η1-OOCH)2] would lie +17.2 kcal mol−1 higher in energy than the reactants. This result is in line with experimental observations, as stable 7061


Organometallics

Article

metal hydrido species were never observed in VT NMR tests11 starting from 2·PF6. An alternative mechanism was thus modeled encompassing a ligand-centered (outer-sphere) rather than metal-centered (inner-sphere) mechanism.23 This hypothesized mechanism involves as a key step the cleavage of the C−H bond of coordinated formate in 8m, followed by hydrogen abstraction by direct interaction with acidic protons present in solution. In this approach, through the interaction with the ammonium cation, one oxygen of the η2-formato ligand in 8m undergoes protonation, opens the chelate ring changing from η2- to η1-coordination,5 completing the coordination geometry with a

formic acid molecule. Thus, the structure [Ru(κ3-triphosMe)(η1OC(OH)H)(η1-OOCH)2] (14m) was achieved, with a free energy cost of +10.5 kcal mol−1 (Figure 10). From structure 14m, the transition state 14mTS (Figure 11) was calculated, where the acidic proton (H1) of the η1-OC(OH)H group interacts with the hydrogen of the coordinated formate ligand (H1···H2 = 0.871 Å). Such an interaction causes a marked weakening of the O2−H1 and C2−H2 bonds, with distances of 1.407 and 1.582 Å, respectively, and a consequent widening of the O3−C2−O4 angle of about 26°. An energy barrier of +30.8 kcal mol−1 was calculated; however, it is expected to decrease by replacing Me groups with “real” phenyl rings in the simplified calculated model triphosMe. The next

Figure 10. Optimized structure for [Ru(κ3-triphosMe)(η1-OC(OH)H)(η1-OOCH)2] (14m). Main bond lengths (Å): Ru−O1 = 2.183, Ru−O2 = 2.187, Ru−O3 = 2.263, O4−H1 = 0.972, Ru−P1 = 2.285, Ru−P2 = 2.297, and Ru−P3 = 2.300.

Figure 11. Optimized structure for transition state 14mTS. Main bond lengths (Å): Ru−O1 = 2.22, O2−H1 = 1.407, H1−H2 = 0.871, C2− H2 = 1.582, Ru−O3 = 2.317, C2−O3 = 1.201, and C2−O4 = 1.190.

Scheme 7. Free Energy Reaction Pathway Associated with the Release of H2 and CO2 from 8m Following a Proposed OuterSphere, Ligand-Centered Mechanism

7062


Organometallics

Article

Mecklenburg-Vorpommern is gratefully acknowledged. A.I. and G.M. acknowledge the ISCRA-CINECA HP grant “HP10BNL89W” and Centro Ricerca Energia e Ambiente, Colle Val d’Elsa (Supporting Information), for computational resources.

step, accounting for the release of both H2 and CO2 and regeneration of 8m, was found to be exoergonic by −48.1 kcal mol−1 (Scheme 7). In summary, the computational analysis of the mechanism associated with formic acid dehydrogenation in the presence of 2·PF6 showed that, in contrast with the system based on complex 1, an outer-sphere mechanism involving release of H2 and CO2 from the formato ligands without the need for any Ru-hydrido species is plausible and favored by −23.8 kcal mol−1 starting from 2m2+ and having 8m as the pivotal species. Interestingly, recent DFT studies24 on Fe-PP3 catalyzed dehydrogenation of formic acid [PP3 = tris(2-(diphenylphosphino)ethyl)phosphine]9a also explained the reaction pathway by outer-sphere mechanisms.

■

■

CONCLUSIONS In this work, a detailed hybrid density functional theory (DFT) study using the B3LYP functional, with structures optimized in solution using the Conductor-Like Polarizable Continuum Model (CPCM), allowed us to clarify important mechanistic details underlying the Ru-catalyzed dehydrogenation of formic acid and related VT NMR experimental mechanistic studies previously described. In particular, it was possible to propose that different mechanisms can be active, depending on the choice of polydentate phosphines, i.e., 1,1,1-tris-(diphenylphosphinomethyl)ethane (triphos) vs tris-[2-(diphenylphosphino)ethyl]amine (NP3) and the nature and number of ancillary ligands (Cl vs MeCN), making available a different number of vacant coordination sites for activation of catalysts to their active forms. In the specific case at hand, it was shown that the octahedral complex [Ru(κ4-NP3)Cl2] (1) promoted an innersphere activation of formate centered on the corresponding hydrochloride derivative [Ru(κ4-NP3)(H)Cl] (4), whereas complex [Ru(κ3-triphos)(MeCN)3](PF6)2 (2·PF6) promoted the same reaction by a ligand-centered outer-sphere mechanism without the need for Ru-hydrido species to be formed. The implications of these findings may have an importance in the future choice of tailored bifunctional ligands around Ru for efficient catalysts for this process.

■

ASSOCIATED CONTENT

S Supporting Information *

Details of the variable-temperature (VT) NMR experimental procedure, DFT methodology and optimized structures, and movies of transition states (TS). This material is available free of charge via the Internet at http://pubs.acs.org.

■

REFERENCES

(1) (a) Armaroli, N.; Balzani, V. Angew. Chem., Int. Ed. 2007, 46, 52− 66. (b) Lubitz, W.; Tumas, B. Chem. Rev. 2007, 107, 3900−3903. (c) Eberle, U.; Felderhoff, M.; Schüth, F. Angew. Chem., Int. Ed. 2009, 48, 6608−6630. (d) Thomas, M. K. Catal. Today 2007, 120, 389−398. (2) (a) van den Berg, A. W. C.; Arean, C. O. Chem. Commun. 2008, 668−681. (b) Murray, L. J.; Dinca, M.; Long, J. R. Chem. Soc. Rev. 2009, 38, 1294−1314. (c) Graetz, J. Chem. Soc. Rev. 2009, 38, 73−82. (d) Hamilton, C. W.; Baker, R. T.; Staubiz, A.; Manners, I. Chem. Soc. Rev. 2009, 38, 279−293. (3) For recent reviews see, for example: (a) Grasemann, M.; Laurenczy, G. Energy Environ. Sci. 2012, 5, 8171−8181. (b) Enthaler, S.; von Lagermann, J.; Schmidt, T. Energy Environ. Sci. 2010, 3, 1207− 1217. (c) Yadav, M.; Xu, Q. Energy Environ. Sci. 2012, 5, 9698−9725. (4) (a) Jessop, P. G.; Ikariya, T.; Noyori, R. Chem. Rev. 1995, 95, 259−272. (b) Jessop, P. G.; Joó, F.; Tai, C. C. Coord. Chem. Rev. 2004, 248, 2425−2442. (c) Jessop, P. G. In The Handbook of Homogeneous Hydrogenation; de Vries, J. G., Elsevier, C. J., Eds.; Wiley-VCH; Weinheim, Germany, 2007. (d) Leitner, W. Angew. Chem., Int. Ed. 1995, 34, 2207−2221. (e) Himeda, Y. Eur. J. Inorg. Chem. 2007, 3927−3941. (f) Hull, J. F.; Himeda, Y.; Wang, W.-H.; Hashiguchi, B.; Periana, R.; Szalda, D. J.; Muckerman, J. T.; Fujita, E. Nat. Chem. 2012, 4, 383−388. (g) Papp, G.; Csorba, J.; Laurenczy, G.; Joó, F. Angew. Chem., Int. Ed. 2011, 50, 10433−10435. (h) Himeda, Y.; Miyazawa, S.; Hirose, T. ChemSusChem 2011, 4, 487−493. (i) Boddien, A.; Federsel, C.; Sponholz, P.; Mellmann, D.; Jackstell, R.; Junge, H.; Laurenczy, G.; Beller, M. Energy Environ. Sci. 2012, 5, 8907−8911. (5) Morris, D. J.; Clarkson, C. J.; Wills, M. Organometallics 2009, 28, 4133−4140. (6) (a) Gao, Y.; Kuncheria, J.; Yap, G. P. A.; Puddephat, R. J. Chem. Commun. 1998, 2365−2366. (b) Gao, Y.; Kuncheria, J. K.; Jenkins, H. A.; Puddephatt, R. J.; Yap, G. P. A. J. Chem. Soc., Dalton Trans. 2000, 4703−4708. (7) (a) Fellay, C.; Dyson, P. J.; Laurenczy, G. Angew. Chem., Int. Ed. 2008, 47, 1−4. (b) Fellay, C.; Yan, N.; Dyson, P. J.; Laurenczy, G. Chem.Eur. J. 2009, 15, 3752−3760. (8) (a) Loges, B.; Boddien, A.; Junge, H.; Beller, M. Angew. Chem., Int. Ed. 2008, 47, 3962−3965. (b) Boddien, A.; Loges, B.; Junge, H.; Gärtner, F.; Noyes, J. R.; Beller, M. Adv. Synth. Catal. 2009, 351, 2517−2520. (9) (a) Boddien, A.; Mellmann, D.; Gärtner, F.; Jackstell, R.; Junge, H.; Dyson, P. J.; Laurenczy, G.; Ludwig, R.; Beller, M. Science 2011, 333, 1733−1736. (b) Ziebart, C.; Federsel, C.; Anbarasan, P.; Jackstell, R.; Baumann, W.; Spannenberg, A.; Beller, M. J. Am. Chem. Soc. 2012, 134, 20701−20704. (10) Federsel, C.; Ziebart, C.; Jackstell, R.; Baumann, W.; Beller, M. Chem.Eur. J. 2012, 18, 72−75. (11) Mellone, I.; Peruzzini, M.; Rosi, L.; Mellmann, D.; Junge, H.; Beller, M.; Gonsalvi, L. Dalton Trans. 2013, 42, 2495−2501. (12) Becke, A. D. J. Chem. Phys. 1993, 98, 5648−5652. (13) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. A.; Nakatsuji, H.; Caricato, M.; Li, X.; Hratchian, H. P.; Izmaylov, A. F.; Bloino, J.; Zheng, G.; Sonnenberg, J. L.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Vreven, T.; Montgomery, J. J. A.; Peralta, J. E.; Ogliaro, F.; Bearpark, M.; Heyd, J. J.; Brothers, E.; Kudin, K. N.; Staroverov, V. N.; Keith, T.; Kobayashi, R.; Normand, J.; Raghavachari, K.; Rendell, A.; Burant, J. C.; Iyengar, S. S.; Tomasi, J.; Cossi, M.; Rega, N.; Millam, J. M.; Klene, M.; Knox, J. E.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.;

AUTHOR INFORMATION

Corresponding Authors

*Email: [email protected]. *E-mail: [email protected]. *E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was financially supported by CNR, MATTM, and MIUR through projects EFOR, PIRODE, and PRIN 2009, respectively, which are gratefully acknowledged. Thanks are also expressed to ECRF through project Firenze Hydrolab-2. L.G. thanks COST Action CM1205 “CARISMA” for funding travel. Additional support by the BMBF and the state of 7063


Organometallics

Article

Martin, R. L.; Morokuma, K.; Zakrzewski, V. G.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Dapprich, S.; Daniels, A. D.; Farkas, O.; Foresman, J. B.; Ortiz, J. V.; Cioslowski, J.; Fox, D. J. Gaussian 0.9, Revision B.01 ed.; Gaussian, Inc.: Wallingford, CT, 2010. (14) Dolg, M.; Stoll, H.; Preuss, H.; Pitzer, R. M. J. Phys. Chem. 1993, 97, 5852−5859. (15) (a) Barone, V.; Cossi, M. J. Phys. Chem. A 1998, 102, 1995− 2001. (b) Cossi, M.; Rega, N.; Scalmani, G.; Barone, V. J. Comput. Chem. 2003, 24, 669−681. (16) Cambridge Structural Database System, Version 5.32; Cambridge Crystallographic Data Centre: Cambridge, UK, 2011. (17) Anzellotti, A.; Briceno, A.; Delgado, G.; Diaz de Delgado, G.; Fontal, B. Acta Crystallogr., Sect. C: Cryst. Struct. Commun. 2002, 58, m355. (18) (a) Hartley, F. R. Chem. Soc. Rev. 1973, 2, 163−179. (b) Sajith, P. K.; Suresh, C. H. Dalton Trans. 2010, 39, 815−822. (c) Sajith, P. K.; Suresh, C. H. J. Organomet. Chem. 2011, 696, 2086−2092. (d) See, R. F.; Kozina, D. J. Coord. Chem. 2013, 66, 490−500. (19) (a) Sandhya, K. S.; Suresh, C. H. Organometallics 2011, 30, 3888−3891. (b) Sandhya, K. S.; Suresh, C. H. Dalton Trans. 2012, 41, 11018−11025. (20) Chen, X.; Xue, P.; Sung, H. H. Y.; Williams, I. D.; Peruzzini, M.; Bianchini, C.; Jia, G. Organometallics 2005, 24, 4330−4332. (21) (a) Hirva, P.; Haukka, M.; Jakonen, M.; Moreno, M. A. J. Mol. Mod. 2008, 17, 171−181. (b) Lombardi, J. R.; Davis, B. Chem. Rev. 2002, 102, 2431−2460. (c) Barden, C. J.; Rienstra-Kiracofe, J. C.; Schaefer, H. F., III J. Chem. Phys. 2000, 113, 690−700. (d) Yanagisawa, S.; Tsuneda, T.; Hirao, K. J. Chem. Phys. 2000, 112, 545−550. (22) Bianchini, C.; Dal Santo, V.; Meli, A.; Oberhauser, W.; Psaro, R.; Vizza, F. Organometallics 2000, 19, 2433−2444. (23) For an excellent review on inner- and outer-sphere mechanisms in Ru-based catalytic processes, see: Clapham, S. E.; Hadzovic, A.; Morris, R. H. Coord. Chem. Rev. 2004, 248, 2201−2237. (24) Yang, X. Dalton Trans. 2013, 42, 11987−11991. (b) Sanchez-De Armas, R.; Xue, L.; Ahlquist, M. S. G. Chem.Eur. J. 2013, 19, 11869−11873.

7064


Inner- versus Outer-Sphere Ru-Catalyzed Formic Acid

Recommend Documents