J Mol Model (2014) 20:2250 DOI 10.1007/s00894-014-2250-4
ORIGINAL PAPER
Theoretical mechanistic study of the formic acid decomposition assisted by a Ru(II)-phosphine catalyst Gloria Mazzone & Marta E. Alberto & Emilia Sicilia
Received: 18 February 2014 / Accepted: 10 April 2014 / Published online: 9 May 2014 # Springer-Verlag Berlin Heidelberg 2014
Abstract A density functional theory (DFT) study of formic acid decomposition, catalyzed by a model of the trans-[Ru(TPPTS) 2 (H 2 O) 4 ] 2+ complex, has been performed. A mechanism comprising two competitive catalytic cycles, which have as a common intermediate a monohydride ruthenium complex, has been hypothesized in literature on the basis of high pressure NMR experiments. To explain the observed increase in H2 production rate during the process, it has been suggested by the same authors that the reaction occurs entering the second proposed cycle (Cycle 2), although none of the complexes assumed to be formed have been experimentally observed. To gain more insights into the reaction mechanism, a detailed investigation of both the proposed catalytic cycles has been carried out. To describe the energy profiles, different accurate computational protocols have been employed. Our computations reveal that molecular hydrogen cannot be produced more rapidly following cycle 2, since it requires a larger amount of energy to occur. Moreover, the release of molecular hydrogen has been found to be the step that limits the reaction rate in both cycles, instead of the CO2 dissociation as hypothesized by the authors.
Keywords DFT . Formic acid dehydrogenation . Homogeneous catalysis . Hydrogen production . Ruthenium This paper belongs to Topical Collection QUITEL 2013 Electronic supplementary material The online version of this article (doi:10.1007/s00894-014-2250-4) contains supplementary material, which is available to authorized users. G. Mazzone : M. E. Alberto : E. Sicilia (*) Dipartimento di Chimica e Tecnologie Chimiche, Università della Calabria, 87036 Arcavacata di Rende, CS, Italy e-mail: [email protected]
Introduction As the reserves of fossil fuels on earth are gradually decreasing, for sustainable development, an alternative energy carrier is needed for both environmental and economic reasons. In this field, hydrogen represents one of the more promising solutions as a secondary energy resource [1]. Hydrogen can be produced from a variety of sources, it is non-toxic and, as an energy carrier, it is extremely environmentally benign being water the only non-problematic side product when it is converted into energy. However, hydrogen as such is not freely available on earth and, hence, it cannot be used as a primary energy source and, despite its obvious benefits, an immediate incorporation of hydrogen into the world economy faces a number of challenges. Since, the efficient storage and handling of hydrogen is one of the major obstacles to its use for energy applications, extensive research has been carried out to develop novel materials for hydrogen gas storage and liberation [2–4], although no entirely satisfactory options have been found so far. Among the different hydrogen storage materials, formic acid has merited special attention [5–7]. Formic acid contains 4.4 wt.% and 53 g/L of hydrogen at ambient conditions and produces only gaseous products (H2/CO2), thereby preventing the accumulation of by-products. In addition to hydrogen generation, based on formic acid and carbon dioxide, a sustainable and reversible cycle for energy storage can be conceived by storage of hydrogen in formic acid and release from it. Formic acid decomposition to H2/CO2 is a thermodynamically favored process [8], although highly elevated temperatures are required [9–11]. The use of several homogeneous systems for hydrogen release from HCOOH has been described since the beginning of the twentieth century, being the process catalyzed by a number of metal complexes [12–18], and remarkable results have been reported in recent years [19, 20].
2250, Page 2 of 8
In 2008, Fukuzumi and co-workers investigated the [Rh(Cp*)(bipy)-(H2O)](SO4) system and similar complexes for the hydrogen generation from aqueous formic acid solutions. More recently, this group demonstrated that heteronuclear iridium-ruthenium complexes are highly active catalysts for hydrogen generation in an aqueous solution under ambient conditions [21, 22]. Himeda et al. focused on iridium complexes for hydrogen generation from formic acid/sodium formate [23], whereas Laurenczy’s [24–26] and Beller’s [27, 28] groups independently have demonstrated that hydrogen generation is also possible under relatively mild conditions using ruthenium phosphine complexes. Laurenczy and co-workers have reported the decomposition into hydrogen and carbon dioxide of formic acid/sodium formate (9:1) in an aqueous solution assisted by a water-soluble ruthenium TPPTS (meta-trisulfonated triphenylphosphine) catalyst [24–26]. The noteworthy features of the catalytic process are as follows: the formic acid decomposition takes place under mild conditions, over a large range of pressures, and remarkably no evolution of carbon monoxide is detected in the gas phase. The decomposition of formic acid is slow, unless sodium formate is present as an initiator. An induction period, which can be reduced by pre-treatment with H2, is required to generate the active catalyst. Continuous evolution of hydrogen with constant formic acid addition, and near-complete decomposition of formic acid is achieved. Identification of some intermediates by high pressure NMR experiments allowed proposing a mechanism consisting of two competitive catalytic cycles involving, as a common intermediate, a monohydride ruthenium complex (Scheme 1). Starting from the bisphosphine tetraaqua ruthenium complex I, formate replaces a water ligand to form complex II, observed at room temperature. A second water molecule is released when β-elimination of a hydrogen atom from formate takes place, forming the carbon dioxide hydride complex III, which is the most abundant species observed in solution. Carbon dioxide is replaced by water to give the not observed monohydride complex IV. The catalytic cycle is closed by coordination of formic acid, which replaces a water ligand and protonates the hydride to form the dihydrogen complex V. Molecular hydrogen is subsequently displaced by water to regenerate formate complex II. Monohydride IV is also the starting point for the dihydride mechanism. Complex IV can react with a formate ion, instead of a formic acid molecule, by replacement of a water molecule to give the hydride formate species VI. β-elimination of hydrogen yields the carbon dioxide complex VII, which eliminates carbon dioxide to form VIII. From this complex, hydrogen is eliminated upon addition of a molecule of formic acid. Although none of the complexes hypothesized to be
J Mol Model (2014) 20:2250
I
II
V
Cycle 1
III
IV
VI VII
Cycle 2 VIII
Scheme 1 Reaction mechanisms proposed in refs. [24–26]
formed in the second cycle have been observed, the authors suggest that the reaction mainly occurs via the dihydride mechanism and an explanation is provided for the observed difference between the first and subsequent cycles. Indeed, the rate determining step of the first cycle is assumed to be the carbon dioxide release, whereas the same step occurs more rapidly in the second cycle due to the trans effect of the hydride ligand in complex VII. To gain more insight into the mechanism of dehydrogenation of formic acid by ruthenium complex and to accomplish a full elucidation of the proposed catalytic cycles [24], including the characterization of relevant transition states and short-life intermediates, we have undertaken a density functional theory (DFT) study of formic acid decomposition catalyzed by a simplified model of the trans-[Ru(tppts)2(H2O)4]2+ complex. The outcomes of our computational analysis show that the mechanism suggested by the authors is not able to rationalize some of the experimental evidence and more efforts are
J Mol Model (2014) 20:2250
required in order to reliably describe all the steps involved in the complex process of H2 production.
Computational details All the electronic structure optimizations involved in the decomposition of formic acid assisted by the Ru complex under investigation have been carried out at DFT level by using the hybrid exchange functional by Becke (B3) [32] in combination with the Lee, Yang, and Parr (LYP) correlation functional [33]. Geometries of minima and transition states have been fully optimized using SDD basis on Ru, denoting the small-core Stuttgart–Dresden relativistic effective core potential (ECP) together with its valence basis set [34], and the standard 631G** basis for all the other elements. It is well known that for heavy elements relativistic effects should be taken into account. Nevertheless, for organometallic systems of medium/ large size quasi-relativistic calculations are very costly and, consequently, the use of relativistic effective core potentials turns out to be the most commonly employed alternative. All the stationary points have been characterized as minima or transition states by vibrational analysis, performed within the harmonic approximation, at the same level of theory, ensuring that every transition state had only one imaginary frequency. Harmonic frequencies have been used, without scaling, to calculate zero-point vibrational energy, enthalpic, and entropic corrections. For all transition states, intrinsic reaction coordinate (IRC) calculations [35, 36] have been used to confirm corresponding minima structures. In order to reduce the computational effort required to investigate the whole process, the TPPTS ligands used to increase the solubility of the complex, have been replaced with the less demanding trimethylphosphine ligands. The substitution of such bulky ligands with smaller ones is a generally accepted procedure on the basis of the similarity of the donor/acceptor properties of PMe3 and PAr3 type phosphines [37–39]. Moreover, some test calculations carried out considering triphenylphosphine ligands have demonstrated that no effect, neither steric nor electronic is imputable to these bulky substituents. As a consequence, we do not expect to introduce significant errors in the calculation of relative energies by the use of simplified phosphines. Therefore, the catalyst used in the experimental study [24] has been replaced here by the trans-[Ru(PMe3)2(H2O)4]2+ complex. Refined energies have been obtained in gas phase and in solvent performing single-point calculations on the optimized structures by using the same ECP on Ru and a larger basis set, 6-311+G**, for the rest of the atoms. The aqueous environment has been modeled using the conductor polarized continuum model (CPCM) [40] and the UFF set of radii has been selected to build up the cavity.
Page 3 of 8, 2250
Different computational protocols have been used to evaluate the energetic profiles of the reaction. In the framework of DFT approach, three exchange-correlation functionals have been tested (B3LYP, MPWB1K [46–48], B97-D [41]). The energy values discussed in the text are those obtained with the B97-D functional, which has been successfully employed to study Ru-catalyzed reactions [42, 43], and allows to account for the critical non-covalent interactions characteristic of this system. It is based on a reparameterization of Becke’s ansatz from 1997 [44] and now explicitly includes long-range electron correlations by addition of atom-pair wise dispersion corrections as in the DFT-D general approach of Grimme [41, 45]. A second-order Möller-Plesset perturbation theory (MP2) within the approximation resolution of identity MP2 (RI-MP2) [49] has also been employed and used as reference. Reaction Gibbs free energies in solution, ΔGsol, have been calculated for each process as the sum of two contributions: a gas-phase reaction free energy, ΔGgas, and a solvation reaction free energy term calculated with the continuum approach, ΔGsolv. Gaussian 03 [50] and Turbomole [51] (version 6.3.1) program packages have been used for these calculations.
Results and discussion Figures 1 and 2 show the free energy profiles for the first and the second proposed catalytic cycles, respectively. A schematic drawing of all the involved intermediates and transition states is also included. The two pathways have an initial common part that will be discussed in the next section within the description of the cycle indicated as cycle 1. More detailed information concerning geometric structures of all the intercepted Ru stationary points is given in Figs. S1 and S2 of the Supporting information. Cycle 1: hydride complex formation by β-hydrogen transfer from formate species All the relative energies reported in Fig. 1 are calculated with respect to complex 1, that is the adduct of the trans-[Ru(PMe3)2(H2O)4]2+ and formate reactants. In such an adduct, the formate anion establishes a hydrogen bond network with water ligands. Following the hypothesis proposed by the authors [24], we h a ve f o u n d t h a t t h e c a t a l y s t pr e c ur s o r t r an s -[ Ru(PMe3)2(H2O)4]2+ affords the active catalyst 2, that is the formate complex trans-[Ru(PMe3)2(HCOO)(H2O)3]+, by replacement of a water ligand and overcoming an activation barrier (TS1–2) of 18.2 kcal mol-1. Complex 2 is stabilized by 4.7 kcal mol-1 with respect to the previous one and it is characterized by the coordination of the formate anion through one oxygen atom that lies 2.107 Å from Ru atom.
2250, Page 4 of 8
J Mol Model (2014) 20:2250
ΔGsolv 21.2 18.2 14.8 12.2
10.8 10.0
0.0
4.8
4.4
3.9
-0.5 -2.9 -4.7
-5.2 -10.2 -14.8
-12.8
-15.7
-25.1
TS 2-3 TS 1-2 1
2
TS 3-4
TS 2'-3
TS 2-2' 2'
3
TS 6-7
TS 5-6 4
5
6
TS 7-8 7
TS 8-9 8
9
Fig. 1 Computed relative free energy profile for the H2 production following cycle 1
The next step consists of the β-hydrogen transfer from the metal coordinated formate anion to form complex 3, in which a CO2 molecule is coordinated to the metal center in a η2 fashion. This type of CO2 coordination implies a distortion of the O-C-O angle, which is now equal to 141.9°. The existence of this kind of complex is proposed by the authors [24] on the basis of crystallographic data [29, 30] and DFT calculations [31] on similar systems reported in literature. Accordingly, our calculations reveal that the η2 coordination of the CO2 molecule to the ruthenium center is the most stable one among the CO2 possible coordination modes. This complex can be formed in two different ways: the first one is a concerted mechanism involving the loss of a water ligand and a H atom transfer to the metal center (TS2–3) with the consequent formation of the CO2 molecule coordinated to the metal. The free energy barrier that is necessary to overcome to form intermediate 3 following this way is 25.9 kcal mol-1. The ligands exchange is assisted by an additional water molecule, which stabilizes the structure by forming hydrogen-bonds with both
the formate and the leaving water ligand. Alternatively, the formation of the η2-CO2 complex can take place according to a dissociative mechanism. In the first step, the water ligand leaves a vacant coordination site overcoming an activation barrier of 15.5 kcal mol-1 leading to a penta-coordinated intermediate 2’. Subsequently, the β-hydrogen transfer from the metal-coordinated formate is realized via the TS2’–3 with an energy barrier of only 2.2 kcal mol-1. The vibrational mode of the imaginary frequency associated with the first transition state along this pathway, TS2–2’, clearly shows only the movement of the leaving water ligand at a distance of 3.378 Å, thus forming the complex 2’ in which the metal center maintains its octahedral structure with a vacant coordination site. In the optimized structure of the second transition state, TS2’–3, the formate C-H bond length increases to 1.147 Å and the H-Ru decreases to 2.465 Å. Once complex 3 is formed, the reaction proceeds with CO2 dissociation in favor of a water ligand, in the TS3–4, generating the monohydride complex 4. The ligands interchange takes place
J Mol Model (2014) 20:2250
Page 5 of 8, 2250
ΔGsolv
22.9
16.2 11.8 8.4
11.3
10.7
0.7
3.5
1.7 0.0 -5.5 -4.9
-6.3 -9.0
-11.4
-20.9
TS 11-12
TS 10-11 10
11
TS 14-15
TS 12-13 12
13
TS 15-16 15
14
TS 17-18
TS 16-17 16
17
18
Fig. 2 Computed relative free energy profile for the H2 production following cycle 2
with an activation barrier of 25.0 kcal mol-1. The formed complex 4 can undergo two alternative routes. Cycle 1 can be closed by replacement of a water ligand with a formic acid molecule and regeneration of complex 2. Alternatively, a formate replaces a water molecule and a second catalytic cycle starts (cycle 2). Cycle 2 will be discussed in the next section. Following the reaction path according to cycle 1, the formic acid molecule approaches the monohydride complex 4 establishing a weak interaction between the carbonyl oxygen atom and the water proton in the adduct 5. This complex is 4.7 kcal mol-1 more stable than the previous minimum. Subsequent coordination of formic acid in place of a water ligand (TS5–6) is computed to occur by overcoming a free energy barrier of 9.1 kcal mol-1. In TS5–6 formic acid comes close to the metal and also interacts with the hydride ligand through its proton. The generated complex 6 lies at −15.7 kcal mol-1 with respect to complex 1 and it is characterized by a η1-coordination of formic acid through the carboxylic oxygen atom. In complex 6 the interaction between the hydride ligand and the formic acid proton causes a concomitant increase of the O-H distance. The subsequent H2 formation that leads to the dihydrogen complex 7, is computed to occur by overcoming
of a very small free energy barrier via the transition state TS6– 7. The H2-complex 7, in which molecular hydrogen is coordinated in a η2 mode and the formed formate species remains coordinated to the metal, is stabilized by 25.1 kcal mol-1 with respect to adduct 1. In the next step of the catalytic cycle, the release of molecular hydrogen occurs as a consequence of the bi-coordination of the formate ligand (TS7–8) with a free energy barrier of 29.5 kcal mol-1. In TS7–8, H2 is released and the H-H bond is almost formed (0.750 Å), while the second formate oxygen atom comes close to the ruthenium atom, being 2.801 Å from it. The formed complex 8 can, then, undergo the attack of a water molecule to achieve species 9, with an activation barrier of 7.7 kcal mol-1. The latter complex represents the active catalyst and it is the same species of complex 2 with the exception of the formed hydrogen molecule in the second coordination sphere that stabilizes complex 9 more than complex 2. Restored catalyst is poised to react with the formate moiety and to restart the catalytic cycle. Laurenczy and co-workers, having observed complex III during the reaction (Scheme 1), have suggested that the subsequent dissociation of CO2 generating the not observed mono-hydride complex IV, could be the rate determining step
2250, Page 6 of 8
J Mol Model (2014) 20:2250
[24]. Nevertheless, our calculations (see Fig. 1) do not support this hypothesis, as the step that requires the greatest amount of energy to occur is the release of H2. Indeed, an activation energy barrier of to 29.5 kcal mol-1 has to be overcome. With the aim to provide a rational explanation for the large amount of energy required for the H2 release, several tests have been performed. The first attempt to verify the reliability of our results, has been to use a model catalyst closer to the real one, hence including the bulky triphenylphosphine ligands to calculate the activation energy barrier of the reaction. Our results show that no role is played by these ligands, neither steric nor electronic, having the computed energy barrier very similar to that computed with less demanding ligands (30.9 kcal mol-1 vs 29.5 kcal mol-1). As a consequence, we believe that the use of simplified phosphines do not introduce significant errors in the calculation of relative energies. To further corroborate this result the same test has been performed on the first step of the reaction, obtaining the same energy barrier. (18.3 kcal mol-1 vs 18.2 kcal mol-1). Furthermore, we estimated the energy required for the H2 release removing all the molecules lying in both second- and third-coordination spheres included in the computations of all the intermediates and transition states. The outcomes of these calculations give very similar results concerning the activation barrier, it being 30.2 kcal mol-1. Moreover, we considered the possibility that the release of a hydrogen molecule and coordination of an incoming water molecule to directly form the catalytic active species 9 might take place simultaneously, hence bypassing the formate bicoordination. Unfortunately, all the attempts to find a transition state for such a concerted step failed. Finally, it is noteworthy that all the employed functionals in the framework of DFT approach, and the different level of theory RI-MP2 confirmed the release of H2 as the rate determining step of the whole reaction. Therefore, the high barrier of such step is not likely to be an artefact of the computational protocol used. The energy barriers relative to each step of cycle 1 are collected in Table 1.
Cycle 2: the formate species is added to the hydride complex An alternative catalytic cycle, named cycle 2, has been proposed by the authors in which the monohydride complex 4 reacts with HCOO− instead of HCOOH. Table 1 Computed Gibbs free energy barriers (ΔG, kcal mol-1) in water for each step of cycle 1, at 298.15 K Free energy barriers for the rate determining step are reported in bold
B97D B3LYP MPWB1K RI-MP2
The relative free energy profile (Fig. 2) has been calculated with respect to the reactant 10, that is the hydride complex 4 and the formate species adduct. The initial step of the cycle is the substitution of a water ligand with a formate molecule via TS10–11. In the transition state the water in trans position to the hydride ligand is leaving the complex and the formate species is entering. Formation of the monohydride formate complex 11 is slightly endothermic and it occurs by surmounting an activation barrier of 8.4 kcal mol-1. The formate is now coordinated to the metal by its oxygen atom (O-Ru distance 2.304 Å) in a η1 fashion. The successive step requires an energy barrier of 21.2 kcal mol-1 to be overcome. In the TS11–12 transition state, a water ligand is replaced by the β-hydrogen of the coordinated formate and, at the same time, the CO2 molecule is formed. The dihydride η2-CO2 complex 12 is stabilized by 11.4 kcal mol-1 with respect to the reference adduct 10. The optimized structure shows a O-C-O angle of 134.7°, due to the η2 mode coordination of CO2, whereas the hydride ligand is found at 1.562 Å from the Ru center. The subsequent dissociation of CO2 in favor of a water ligand coordination, via the TS12–13, requires an amount of energy equal to 27.6 kcal mol-1 leading to complex 13 formation. The presence of a hydride ligand in trans to the leaving group does not increase the rate of the CO2 release compared to the same step in cycle 1. The reaction proceeds with the involvement of a formic acid molecule in the second coordination sphere of the metal center, forming complex 14. The presence of a HCOOH molecule, which is already in the right position to replace a water ligand, stabilizes the dihydride intermediate by 15.6 kcal mol-1. Coordination of formic acid through the transition state TS14–15 takes place with a small activation barrier of 5.6 kcal mol-1. In complex 15, which is stabilized by 4.1 kcal mol-1 with respect to the preceding minimum, the formic acid proton interacts with the hydride ligand at a distance of 1.362 Å. The subsequent step of the reaction, that corresponds to the formation of molecular hydrogen coordinated in a η2 mode to the metal center in the 16 complex, takes place with a small energy barrier of only 4.5 kcal mol-1. The involved transition state, TS15–16, is characterized by a six-member ring structure formed by Ru, hydride ligand, and formic acid. Also in cycle 2, the subsequent release of the H2 molecule represents the rate determining step of the whole process. In TS16–17 the H2 molecule leaves the ruthenium coordination site and, at the same time, the other oxygen atom of formate
ΔG 1–2
ΔG 2–2’
ΔG 2’–3
ΔG 3–4
ΔG 5–6
ΔG 6–7
ΔG 7–8
ΔG 8–9
18.2 18.4 16.5 21.1
15.5 19.8 19.4 20.4
2.1 1.3 2.4 4.4
25.0 23.8 19.0 26.9
9.2 12.4 9.5 11.1
0.9 1.1 2.1 1.0
29.5 24.2 24.0 27.4
7.7 13.7 7.5 14.1
J Mol Model (2014) 20:2250 Table 2 Computed Gibbs free energy barriers (ΔG, kcal mol-1) in water for each step of cycle 2, at 298.15 K Free energy barriers for the rate determining step are reported in bold
Page 7 of 8, 2250
B97D B3LYP MPWB1K RI-MP2
ΔG 10–11
ΔG 11–12
ΔG 12–13
ΔG 14–15
ΔG 15–16
ΔG 16–17
ΔG 17–18
8.4 14.0 9.0 12.2
21.2 25.6 26.1 26.8
27.7 25.9 26.3 29.1
5.7 6.7 6.4 7.0
3.5 6.1 5.3 4.4
32.6 28.7 27.5 35.4
7.8 11.1 10.4 9.6
comes at 2.802 Å from the metal. The bi-coordinated formate complex 17 is thus formed overcoming an energy barrier equal to 32.7 kcal mol-1. The last step of the catalytic cycle, required to regenerate the formate hydride complex 18, is the attack of a water molecule with the consequent formate switch from η2 to η1 coordination. The height of the barrier for the TS17–18 corresponding to this rearrangement is computed to be 7.8 kcal mol-1. The restored hydride formate complex 18 corresponds to complex 11 with exception of the formed hydrogen molecule in the second coordination sphere. The catalytic cycle can start again by replacement of a water ligand. In analogy with cycle 1, the same tests to validate the rate determining step of the process have been carried out. The outcomes of the calculations done removing all the extra molecules show that the activation barrier does not change. All the strategies used to intercept a concerted transition state to restore complex 18 do not provide any result. Also in this case both the meta-hybrid functional MPWB1K and RI-MP2 approach, used to compute the free energy barriers for all the steps involved in cycle 2, confirmed the release of H2 as the rate limiting step of the process. The outcomes of these calculations are collected in Table 2. Our computations reveal that molecular hydrogen cannot be produced more rapidly following cycle 2, as hypothesized by the authors to explain the observed difference between the first and subsequent cycles. Indeed, the computed activation barrier is higher than that calculated for the catalytic cycle 1 (32.7 kcal mol-1 versus 29.5 kcal mol-1). Summarizing, the mechanisms proposed and herein accurately investigated leave several open questions to be properly addressed. The hypothesis that CO2 release represents the rate determining step of cycle 1, whereas the same step occurs more rapidly in cycle 2 due to the trans effect of the hydride ligand, is not supported by the results of our computational analysis. On the basis of our results, cycle 2 does not provide an explanation for the difference between the first and the subsequent catalytic cycles.
Conclusions The mechanism of the selective decomposition of formic acid, without CO formation and under relatively mild conditions, assisted by [Ru(H2O)6](tppts)2]2+ as the catalyst for hydrogen
generation has been investigated in detail by using a model complex and applying accurate theoretical approaches. The experimental evidence that the decomposition of formic acid is slow, unless sodium formate is present as an initiator, and an induction period is required to generate the active catalyst, has prompted the authors to propose a mechanism in two cycles. According to their hypothesis, the rate-determining step of both proposed catalytic cycles should be the CO2 release, whose rate has been proposed to be higher in the second cycle due to the trans effect of a hydride ligand. Such hypothesis is not supported by the calculations presented here. First of all, CO2 elimination is not found to be the rate-determining step of both examined catalytic cycles, and the energy barriers for the elimination of molecular hydrogen is the highest of the catalytic cycles. This evidence seems to be a genuine result, since it has been confirmed not only by several tests focused on this key step of the reaction, but also employing different computational approaches to compute the energy barriers. Moreover, our computations reveal that molecular hydrogen cannot be produced more rapidly following cycle 2, as hypothesized by the authors, since it has a higher activation barrier. Given the relevance of the topic, further experimental information about both the characterization of intermediates and reaction rates should be welcome to unequivocally determine the mechanism of the process. Acknowledgments European Union, FP7-PEOPLE-2011-IRSES, project N. 295172, TEMM1P and Università della Calabria are gratefully acknowledged.
References 1. Zuttel A, Borgschulteand A (2008) In: Schlapbach L (ed) Hydrogen as a future energy carrier. Wiley VCH, Weinheim 2. Thomas KM (2009) Dalton Trans:1487–1505 3. Makowski P, Thomas A, Kuhn P, Goettmann F (2009) Energy Environ Sci 2:480–490 4. Demirci UB, Miele P (2009) Energy Environ Sci 2:627–637 5. Fukuzumi S (2008) Eur J Inorg Chem:1351–1362 doi: 10.1002/ ejic.200701369 6. Enthaler S (2008) ChemSusChem 1:801–804 7. Joó F (2008) ChemSusChem 1:805–808 8. Strauss SH, Whitmire KH, Shriver DF (1979) J Organomet Chem 174:C59–C62 9. Hinshelwood CN, Hartley HJ (1923) Chem Soc Trans 123:1333– 1338
2250, Page 8 of 8 10. Wakai C, Yoshida K, Tsujino Y, Matubayashi N, Nakamura M (2004) Chem Lett 33:572–573 11. Blake PG, Davies HH, Jackson GE (1971) J Chem Soc B:1923–1925 doi: 10.1039/J29710001923 12. Coffey RS (1967) Chem Commun 18:923–924 13. Forster D, Beck GR (1971) Chem Commun 994:1072 14. Laine RM, Rinker RG, Ford PC (1977) J Am Chem Soc 99:252–253 15. Yoshida T, Ueda Y, Otsuka S (1978) J Am Chem Soc 100:3941–3942 16. Paonessa RS, Trogler WC (1982) J Am Chem Soc 104:3529–3530 17. King RB, Bhattacharyya NK (1995) Inorg Chim Acta 237:65–69 18. Man ML, Zhou Z, Ng SM, Lau CP (2003) Dalton Trans:3727–3735 19. Johnson CT, Morris DJ, Wills M (2010) Chem Soc Rev 39:81–88 20. Loges B, Boddien A, Gärtner F, Junge H, Beller M (2010) Top Catal 53:902–914 21. Fukuzumi S, Kobayashi T, Suenobu T (2008) Chem Sus Chem 1: 827–834 22. Fukuzumi S, Kobayashi T, Suenobu T (2010) J Am Chem Soc 132: 1496–1497 23. Himeda Y (2009) Green Chem 11:2018–2022 24. Fellay C, Yan N, Dyson PJ, Laurenczy G (2009) Chem Eur J 15: 3752–3760 25. Gan W, Dyson PJ, Laurenczy G (2009) React Kinet Catal Lett 89: 205–213 26. Fellay C, Dyson PJ, Laurenczy G (2008) Angew Chem Int Ed 47: 3966–3968 27. Loges B, Boddien A, Junge H, Beller M (2008) Angew Chem Int Ed 47:3962–3965 28. Boddien A, Loges B, Junge H, Beller M (2008) ChemSusChem 1: 751–758 29. Alvarez R, Carmona E, Marin JM, Poveda ML, Gutiérrez-Puebla E, Monge A (1986) J Am Chem Soc 108:2286–2294 30. Bristow GS, Hitchcock PB, Lappert MF (1981) J Chem Soc Chem Commun:1145–1146 31. Ohnishi Y-Y, Matsunaga T, Nakao Y, Sato H, Sakaki S (2005) J Am Chem Soc 127:4021–4032 32. Becke AD (1993) J Chem Phys 98:5648–5652 33. Lee C, Yang W, Parr RG (1988) Phys Rev B Condens Matter Mater Phys 37:785–789 34. Andrae D, Häussermann U, Dolg M, Stoll H, Preuss H (1990) Theor Chim Acta 77:123–141
J Mol Model (2014) 20:2250 35. Fukui KJ (1970) Phys Chem 74:4161–4163 36. Gonzalez C, Schlegel HB (1989) J Chem Phys 90:2154–2161 37. Kovács G, Schubert G, Joó F, Pápai I (2005) Organometallics 24: 3059–3065 38. Donoso D, Mendizabal F (2011) Theor Chem Accounts 129:381– 387 39. Crabtree RH (2001) The organometallic chemistry of the transition metals. Wiley, New York 40. Cossi M, Rega N, Scalmani G, Barone V (2003) J Comput Chem 24: 669–681 41. Grimme S (2006) J Comput Chem 27:1787–1799 42. Piacenza M, Hyla-Kryspin I, Grimme S (2007) J Comput Chem 28: 2275–2285 43. Sieffert N, Bühl M (2010) J Am Chem Soc 132:8056–8070 44. Becke AD (1997) J Chem Phys 107:8554–8560 45. Grimme S (2004) J Comput Chem 25:1463–1473 46. Zhao Y, Truhlar DG (2004) J Phys Chem A 108:6908–6918 47. Sousa SF, Fernandes PA, Ramos MJ (2007) J Phys Chem A 111: 10439–10452 48. Bras NF, Moura-Tamames SA, Fernandes PA, Ramos MJ (2008) J Comput Chem 29:2565–2574 49. Möller C, Plesset MS (1934) Phys Rev 46:618–622 50. Gaussian 03, Revision C.02, Frisch MJ, Trucks GW, Schlegel HB, Scuseria GE, Robb MA, Cheeseman JR, Montgomery Jr. JA, Vreven T, Kudin KN, Burant JC, Millam JM, Iyengar SS, Tomasi J, Barone V, Mennucci B, Cossi M, Scalmani G, Rega N, Petersson GA, Nakatsuji H, Hada M, Ehara M, Toyota K, Fukuda R, Hasegawa J, Ishida M, Nakajima T, Honda Y, Kitao O, Nakai H, Klene M, Li X, Knox JE, Hratchian HP, Cross JB, Bakken V, Adamo C, Jaramillo J, Gomperts R, Stratmann RE, Yazyev O, Austin AJ, Cammi R, Pomelli C, Ochterski JW, Ayala PY, Morokuma K, Voth GA, Salvador P, Dannenberg JJ, Zakrzewski VG, Dapprich S, Daniels AD, Strain MC, Farkas O, Malick DK, Rabuck AD, Raghavachari K, Foresman JB, Ortiz JV, Cui Q, Baboul AG, Clifford S, Cioslowski J, Stefanov BB, Liu G, Liashenko A, Piskorz P, Komaromi I, Martin RL, Fox DJ, Keith T, Al-Laham MA, Peng CY, Nanayakkara A, Challacombe M, Gill PMW, Johnson B, Chen W, Wong MW, Gonzalez C, Pople JA (2004) Gaussian Inc, Wallingford, CT 51. Ahlrichs R, Bär R, Häser M, Horn H, Kölmel C (1989) Chem Phys Lett 162:165–169
ORIGINAL PAPER
Theoretical mechanistic study of the formic acid decomposition assisted by a Ru(II)-phosphine catalyst Gloria Mazzone & Marta E. Alberto & Emilia Sicilia
Received: 18 February 2014 / Accepted: 10 April 2014 / Published online: 9 May 2014 # Springer-Verlag Berlin Heidelberg 2014
Abstract A density functional theory (DFT) study of formic acid decomposition, catalyzed by a model of the trans-[Ru(TPPTS) 2 (H 2 O) 4 ] 2+ complex, has been performed. A mechanism comprising two competitive catalytic cycles, which have as a common intermediate a monohydride ruthenium complex, has been hypothesized in literature on the basis of high pressure NMR experiments. To explain the observed increase in H2 production rate during the process, it has been suggested by the same authors that the reaction occurs entering the second proposed cycle (Cycle 2), although none of the complexes assumed to be formed have been experimentally observed. To gain more insights into the reaction mechanism, a detailed investigation of both the proposed catalytic cycles has been carried out. To describe the energy profiles, different accurate computational protocols have been employed. Our computations reveal that molecular hydrogen cannot be produced more rapidly following cycle 2, since it requires a larger amount of energy to occur. Moreover, the release of molecular hydrogen has been found to be the step that limits the reaction rate in both cycles, instead of the CO2 dissociation as hypothesized by the authors.
Keywords DFT . Formic acid dehydrogenation . Homogeneous catalysis . Hydrogen production . Ruthenium This paper belongs to Topical Collection QUITEL 2013 Electronic supplementary material The online version of this article (doi:10.1007/s00894-014-2250-4) contains supplementary material, which is available to authorized users. G. Mazzone : M. E. Alberto : E. Sicilia (*) Dipartimento di Chimica e Tecnologie Chimiche, Università della Calabria, 87036 Arcavacata di Rende, CS, Italy e-mail: [email protected]
Introduction As the reserves of fossil fuels on earth are gradually decreasing, for sustainable development, an alternative energy carrier is needed for both environmental and economic reasons. In this field, hydrogen represents one of the more promising solutions as a secondary energy resource [1]. Hydrogen can be produced from a variety of sources, it is non-toxic and, as an energy carrier, it is extremely environmentally benign being water the only non-problematic side product when it is converted into energy. However, hydrogen as such is not freely available on earth and, hence, it cannot be used as a primary energy source and, despite its obvious benefits, an immediate incorporation of hydrogen into the world economy faces a number of challenges. Since, the efficient storage and handling of hydrogen is one of the major obstacles to its use for energy applications, extensive research has been carried out to develop novel materials for hydrogen gas storage and liberation [2–4], although no entirely satisfactory options have been found so far. Among the different hydrogen storage materials, formic acid has merited special attention [5–7]. Formic acid contains 4.4 wt.% and 53 g/L of hydrogen at ambient conditions and produces only gaseous products (H2/CO2), thereby preventing the accumulation of by-products. In addition to hydrogen generation, based on formic acid and carbon dioxide, a sustainable and reversible cycle for energy storage can be conceived by storage of hydrogen in formic acid and release from it. Formic acid decomposition to H2/CO2 is a thermodynamically favored process [8], although highly elevated temperatures are required [9–11]. The use of several homogeneous systems for hydrogen release from HCOOH has been described since the beginning of the twentieth century, being the process catalyzed by a number of metal complexes [12–18], and remarkable results have been reported in recent years [19, 20].
2250, Page 2 of 8
In 2008, Fukuzumi and co-workers investigated the [Rh(Cp*)(bipy)-(H2O)](SO4) system and similar complexes for the hydrogen generation from aqueous formic acid solutions. More recently, this group demonstrated that heteronuclear iridium-ruthenium complexes are highly active catalysts for hydrogen generation in an aqueous solution under ambient conditions [21, 22]. Himeda et al. focused on iridium complexes for hydrogen generation from formic acid/sodium formate [23], whereas Laurenczy’s [24–26] and Beller’s [27, 28] groups independently have demonstrated that hydrogen generation is also possible under relatively mild conditions using ruthenium phosphine complexes. Laurenczy and co-workers have reported the decomposition into hydrogen and carbon dioxide of formic acid/sodium formate (9:1) in an aqueous solution assisted by a water-soluble ruthenium TPPTS (meta-trisulfonated triphenylphosphine) catalyst [24–26]. The noteworthy features of the catalytic process are as follows: the formic acid decomposition takes place under mild conditions, over a large range of pressures, and remarkably no evolution of carbon monoxide is detected in the gas phase. The decomposition of formic acid is slow, unless sodium formate is present as an initiator. An induction period, which can be reduced by pre-treatment with H2, is required to generate the active catalyst. Continuous evolution of hydrogen with constant formic acid addition, and near-complete decomposition of formic acid is achieved. Identification of some intermediates by high pressure NMR experiments allowed proposing a mechanism consisting of two competitive catalytic cycles involving, as a common intermediate, a monohydride ruthenium complex (Scheme 1). Starting from the bisphosphine tetraaqua ruthenium complex I, formate replaces a water ligand to form complex II, observed at room temperature. A second water molecule is released when β-elimination of a hydrogen atom from formate takes place, forming the carbon dioxide hydride complex III, which is the most abundant species observed in solution. Carbon dioxide is replaced by water to give the not observed monohydride complex IV. The catalytic cycle is closed by coordination of formic acid, which replaces a water ligand and protonates the hydride to form the dihydrogen complex V. Molecular hydrogen is subsequently displaced by water to regenerate formate complex II. Monohydride IV is also the starting point for the dihydride mechanism. Complex IV can react with a formate ion, instead of a formic acid molecule, by replacement of a water molecule to give the hydride formate species VI. β-elimination of hydrogen yields the carbon dioxide complex VII, which eliminates carbon dioxide to form VIII. From this complex, hydrogen is eliminated upon addition of a molecule of formic acid. Although none of the complexes hypothesized to be
J Mol Model (2014) 20:2250
I
II
V
Cycle 1
III
IV
VI VII
Cycle 2 VIII
Scheme 1 Reaction mechanisms proposed in refs. [24–26]
formed in the second cycle have been observed, the authors suggest that the reaction mainly occurs via the dihydride mechanism and an explanation is provided for the observed difference between the first and subsequent cycles. Indeed, the rate determining step of the first cycle is assumed to be the carbon dioxide release, whereas the same step occurs more rapidly in the second cycle due to the trans effect of the hydride ligand in complex VII. To gain more insight into the mechanism of dehydrogenation of formic acid by ruthenium complex and to accomplish a full elucidation of the proposed catalytic cycles [24], including the characterization of relevant transition states and short-life intermediates, we have undertaken a density functional theory (DFT) study of formic acid decomposition catalyzed by a simplified model of the trans-[Ru(tppts)2(H2O)4]2+ complex. The outcomes of our computational analysis show that the mechanism suggested by the authors is not able to rationalize some of the experimental evidence and more efforts are
J Mol Model (2014) 20:2250
required in order to reliably describe all the steps involved in the complex process of H2 production.
Computational details All the electronic structure optimizations involved in the decomposition of formic acid assisted by the Ru complex under investigation have been carried out at DFT level by using the hybrid exchange functional by Becke (B3) [32] in combination with the Lee, Yang, and Parr (LYP) correlation functional [33]. Geometries of minima and transition states have been fully optimized using SDD basis on Ru, denoting the small-core Stuttgart–Dresden relativistic effective core potential (ECP) together with its valence basis set [34], and the standard 631G** basis for all the other elements. It is well known that for heavy elements relativistic effects should be taken into account. Nevertheless, for organometallic systems of medium/ large size quasi-relativistic calculations are very costly and, consequently, the use of relativistic effective core potentials turns out to be the most commonly employed alternative. All the stationary points have been characterized as minima or transition states by vibrational analysis, performed within the harmonic approximation, at the same level of theory, ensuring that every transition state had only one imaginary frequency. Harmonic frequencies have been used, without scaling, to calculate zero-point vibrational energy, enthalpic, and entropic corrections. For all transition states, intrinsic reaction coordinate (IRC) calculations [35, 36] have been used to confirm corresponding minima structures. In order to reduce the computational effort required to investigate the whole process, the TPPTS ligands used to increase the solubility of the complex, have been replaced with the less demanding trimethylphosphine ligands. The substitution of such bulky ligands with smaller ones is a generally accepted procedure on the basis of the similarity of the donor/acceptor properties of PMe3 and PAr3 type phosphines [37–39]. Moreover, some test calculations carried out considering triphenylphosphine ligands have demonstrated that no effect, neither steric nor electronic is imputable to these bulky substituents. As a consequence, we do not expect to introduce significant errors in the calculation of relative energies by the use of simplified phosphines. Therefore, the catalyst used in the experimental study [24] has been replaced here by the trans-[Ru(PMe3)2(H2O)4]2+ complex. Refined energies have been obtained in gas phase and in solvent performing single-point calculations on the optimized structures by using the same ECP on Ru and a larger basis set, 6-311+G**, for the rest of the atoms. The aqueous environment has been modeled using the conductor polarized continuum model (CPCM) [40] and the UFF set of radii has been selected to build up the cavity.
Page 3 of 8, 2250
Different computational protocols have been used to evaluate the energetic profiles of the reaction. In the framework of DFT approach, three exchange-correlation functionals have been tested (B3LYP, MPWB1K [46–48], B97-D [41]). The energy values discussed in the text are those obtained with the B97-D functional, which has been successfully employed to study Ru-catalyzed reactions [42, 43], and allows to account for the critical non-covalent interactions characteristic of this system. It is based on a reparameterization of Becke’s ansatz from 1997 [44] and now explicitly includes long-range electron correlations by addition of atom-pair wise dispersion corrections as in the DFT-D general approach of Grimme [41, 45]. A second-order Möller-Plesset perturbation theory (MP2) within the approximation resolution of identity MP2 (RI-MP2) [49] has also been employed and used as reference. Reaction Gibbs free energies in solution, ΔGsol, have been calculated for each process as the sum of two contributions: a gas-phase reaction free energy, ΔGgas, and a solvation reaction free energy term calculated with the continuum approach, ΔGsolv. Gaussian 03 [50] and Turbomole [51] (version 6.3.1) program packages have been used for these calculations.
Results and discussion Figures 1 and 2 show the free energy profiles for the first and the second proposed catalytic cycles, respectively. A schematic drawing of all the involved intermediates and transition states is also included. The two pathways have an initial common part that will be discussed in the next section within the description of the cycle indicated as cycle 1. More detailed information concerning geometric structures of all the intercepted Ru stationary points is given in Figs. S1 and S2 of the Supporting information. Cycle 1: hydride complex formation by β-hydrogen transfer from formate species All the relative energies reported in Fig. 1 are calculated with respect to complex 1, that is the adduct of the trans-[Ru(PMe3)2(H2O)4]2+ and formate reactants. In such an adduct, the formate anion establishes a hydrogen bond network with water ligands. Following the hypothesis proposed by the authors [24], we h a ve f o u n d t h a t t h e c a t a l y s t pr e c ur s o r t r an s -[ Ru(PMe3)2(H2O)4]2+ affords the active catalyst 2, that is the formate complex trans-[Ru(PMe3)2(HCOO)(H2O)3]+, by replacement of a water ligand and overcoming an activation barrier (TS1–2) of 18.2 kcal mol-1. Complex 2 is stabilized by 4.7 kcal mol-1 with respect to the previous one and it is characterized by the coordination of the formate anion through one oxygen atom that lies 2.107 Å from Ru atom.
2250, Page 4 of 8
J Mol Model (2014) 20:2250
ΔGsolv 21.2 18.2 14.8 12.2
10.8 10.0
0.0
4.8
4.4
3.9
-0.5 -2.9 -4.7
-5.2 -10.2 -14.8
-12.8
-15.7
-25.1
TS 2-3 TS 1-2 1
2
TS 3-4
TS 2'-3
TS 2-2' 2'
3
TS 6-7
TS 5-6 4
5
6
TS 7-8 7
TS 8-9 8
9
Fig. 1 Computed relative free energy profile for the H2 production following cycle 1
The next step consists of the β-hydrogen transfer from the metal coordinated formate anion to form complex 3, in which a CO2 molecule is coordinated to the metal center in a η2 fashion. This type of CO2 coordination implies a distortion of the O-C-O angle, which is now equal to 141.9°. The existence of this kind of complex is proposed by the authors [24] on the basis of crystallographic data [29, 30] and DFT calculations [31] on similar systems reported in literature. Accordingly, our calculations reveal that the η2 coordination of the CO2 molecule to the ruthenium center is the most stable one among the CO2 possible coordination modes. This complex can be formed in two different ways: the first one is a concerted mechanism involving the loss of a water ligand and a H atom transfer to the metal center (TS2–3) with the consequent formation of the CO2 molecule coordinated to the metal. The free energy barrier that is necessary to overcome to form intermediate 3 following this way is 25.9 kcal mol-1. The ligands exchange is assisted by an additional water molecule, which stabilizes the structure by forming hydrogen-bonds with both
the formate and the leaving water ligand. Alternatively, the formation of the η2-CO2 complex can take place according to a dissociative mechanism. In the first step, the water ligand leaves a vacant coordination site overcoming an activation barrier of 15.5 kcal mol-1 leading to a penta-coordinated intermediate 2’. Subsequently, the β-hydrogen transfer from the metal-coordinated formate is realized via the TS2’–3 with an energy barrier of only 2.2 kcal mol-1. The vibrational mode of the imaginary frequency associated with the first transition state along this pathway, TS2–2’, clearly shows only the movement of the leaving water ligand at a distance of 3.378 Å, thus forming the complex 2’ in which the metal center maintains its octahedral structure with a vacant coordination site. In the optimized structure of the second transition state, TS2’–3, the formate C-H bond length increases to 1.147 Å and the H-Ru decreases to 2.465 Å. Once complex 3 is formed, the reaction proceeds with CO2 dissociation in favor of a water ligand, in the TS3–4, generating the monohydride complex 4. The ligands interchange takes place
J Mol Model (2014) 20:2250
Page 5 of 8, 2250
ΔGsolv
22.9
16.2 11.8 8.4
11.3
10.7
0.7
3.5
1.7 0.0 -5.5 -4.9
-6.3 -9.0
-11.4
-20.9
TS 11-12
TS 10-11 10
11
TS 14-15
TS 12-13 12
13
TS 15-16 15
14
TS 17-18
TS 16-17 16
17
18
Fig. 2 Computed relative free energy profile for the H2 production following cycle 2
with an activation barrier of 25.0 kcal mol-1. The formed complex 4 can undergo two alternative routes. Cycle 1 can be closed by replacement of a water ligand with a formic acid molecule and regeneration of complex 2. Alternatively, a formate replaces a water molecule and a second catalytic cycle starts (cycle 2). Cycle 2 will be discussed in the next section. Following the reaction path according to cycle 1, the formic acid molecule approaches the monohydride complex 4 establishing a weak interaction between the carbonyl oxygen atom and the water proton in the adduct 5. This complex is 4.7 kcal mol-1 more stable than the previous minimum. Subsequent coordination of formic acid in place of a water ligand (TS5–6) is computed to occur by overcoming a free energy barrier of 9.1 kcal mol-1. In TS5–6 formic acid comes close to the metal and also interacts with the hydride ligand through its proton. The generated complex 6 lies at −15.7 kcal mol-1 with respect to complex 1 and it is characterized by a η1-coordination of formic acid through the carboxylic oxygen atom. In complex 6 the interaction between the hydride ligand and the formic acid proton causes a concomitant increase of the O-H distance. The subsequent H2 formation that leads to the dihydrogen complex 7, is computed to occur by overcoming
of a very small free energy barrier via the transition state TS6– 7. The H2-complex 7, in which molecular hydrogen is coordinated in a η2 mode and the formed formate species remains coordinated to the metal, is stabilized by 25.1 kcal mol-1 with respect to adduct 1. In the next step of the catalytic cycle, the release of molecular hydrogen occurs as a consequence of the bi-coordination of the formate ligand (TS7–8) with a free energy barrier of 29.5 kcal mol-1. In TS7–8, H2 is released and the H-H bond is almost formed (0.750 Å), while the second formate oxygen atom comes close to the ruthenium atom, being 2.801 Å from it. The formed complex 8 can, then, undergo the attack of a water molecule to achieve species 9, with an activation barrier of 7.7 kcal mol-1. The latter complex represents the active catalyst and it is the same species of complex 2 with the exception of the formed hydrogen molecule in the second coordination sphere that stabilizes complex 9 more than complex 2. Restored catalyst is poised to react with the formate moiety and to restart the catalytic cycle. Laurenczy and co-workers, having observed complex III during the reaction (Scheme 1), have suggested that the subsequent dissociation of CO2 generating the not observed mono-hydride complex IV, could be the rate determining step
2250, Page 6 of 8
J Mol Model (2014) 20:2250
[24]. Nevertheless, our calculations (see Fig. 1) do not support this hypothesis, as the step that requires the greatest amount of energy to occur is the release of H2. Indeed, an activation energy barrier of to 29.5 kcal mol-1 has to be overcome. With the aim to provide a rational explanation for the large amount of energy required for the H2 release, several tests have been performed. The first attempt to verify the reliability of our results, has been to use a model catalyst closer to the real one, hence including the bulky triphenylphosphine ligands to calculate the activation energy barrier of the reaction. Our results show that no role is played by these ligands, neither steric nor electronic, having the computed energy barrier very similar to that computed with less demanding ligands (30.9 kcal mol-1 vs 29.5 kcal mol-1). As a consequence, we believe that the use of simplified phosphines do not introduce significant errors in the calculation of relative energies. To further corroborate this result the same test has been performed on the first step of the reaction, obtaining the same energy barrier. (18.3 kcal mol-1 vs 18.2 kcal mol-1). Furthermore, we estimated the energy required for the H2 release removing all the molecules lying in both second- and third-coordination spheres included in the computations of all the intermediates and transition states. The outcomes of these calculations give very similar results concerning the activation barrier, it being 30.2 kcal mol-1. Moreover, we considered the possibility that the release of a hydrogen molecule and coordination of an incoming water molecule to directly form the catalytic active species 9 might take place simultaneously, hence bypassing the formate bicoordination. Unfortunately, all the attempts to find a transition state for such a concerted step failed. Finally, it is noteworthy that all the employed functionals in the framework of DFT approach, and the different level of theory RI-MP2 confirmed the release of H2 as the rate determining step of the whole reaction. Therefore, the high barrier of such step is not likely to be an artefact of the computational protocol used. The energy barriers relative to each step of cycle 1 are collected in Table 1.
Cycle 2: the formate species is added to the hydride complex An alternative catalytic cycle, named cycle 2, has been proposed by the authors in which the monohydride complex 4 reacts with HCOO− instead of HCOOH. Table 1 Computed Gibbs free energy barriers (ΔG, kcal mol-1) in water for each step of cycle 1, at 298.15 K Free energy barriers for the rate determining step are reported in bold
B97D B3LYP MPWB1K RI-MP2
The relative free energy profile (Fig. 2) has been calculated with respect to the reactant 10, that is the hydride complex 4 and the formate species adduct. The initial step of the cycle is the substitution of a water ligand with a formate molecule via TS10–11. In the transition state the water in trans position to the hydride ligand is leaving the complex and the formate species is entering. Formation of the monohydride formate complex 11 is slightly endothermic and it occurs by surmounting an activation barrier of 8.4 kcal mol-1. The formate is now coordinated to the metal by its oxygen atom (O-Ru distance 2.304 Å) in a η1 fashion. The successive step requires an energy barrier of 21.2 kcal mol-1 to be overcome. In the TS11–12 transition state, a water ligand is replaced by the β-hydrogen of the coordinated formate and, at the same time, the CO2 molecule is formed. The dihydride η2-CO2 complex 12 is stabilized by 11.4 kcal mol-1 with respect to the reference adduct 10. The optimized structure shows a O-C-O angle of 134.7°, due to the η2 mode coordination of CO2, whereas the hydride ligand is found at 1.562 Å from the Ru center. The subsequent dissociation of CO2 in favor of a water ligand coordination, via the TS12–13, requires an amount of energy equal to 27.6 kcal mol-1 leading to complex 13 formation. The presence of a hydride ligand in trans to the leaving group does not increase the rate of the CO2 release compared to the same step in cycle 1. The reaction proceeds with the involvement of a formic acid molecule in the second coordination sphere of the metal center, forming complex 14. The presence of a HCOOH molecule, which is already in the right position to replace a water ligand, stabilizes the dihydride intermediate by 15.6 kcal mol-1. Coordination of formic acid through the transition state TS14–15 takes place with a small activation barrier of 5.6 kcal mol-1. In complex 15, which is stabilized by 4.1 kcal mol-1 with respect to the preceding minimum, the formic acid proton interacts with the hydride ligand at a distance of 1.362 Å. The subsequent step of the reaction, that corresponds to the formation of molecular hydrogen coordinated in a η2 mode to the metal center in the 16 complex, takes place with a small energy barrier of only 4.5 kcal mol-1. The involved transition state, TS15–16, is characterized by a six-member ring structure formed by Ru, hydride ligand, and formic acid. Also in cycle 2, the subsequent release of the H2 molecule represents the rate determining step of the whole process. In TS16–17 the H2 molecule leaves the ruthenium coordination site and, at the same time, the other oxygen atom of formate
ΔG 1–2
ΔG 2–2’
ΔG 2’–3
ΔG 3–4
ΔG 5–6
ΔG 6–7
ΔG 7–8
ΔG 8–9
18.2 18.4 16.5 21.1
15.5 19.8 19.4 20.4
2.1 1.3 2.4 4.4
25.0 23.8 19.0 26.9
9.2 12.4 9.5 11.1
0.9 1.1 2.1 1.0
29.5 24.2 24.0 27.4
7.7 13.7 7.5 14.1
J Mol Model (2014) 20:2250 Table 2 Computed Gibbs free energy barriers (ΔG, kcal mol-1) in water for each step of cycle 2, at 298.15 K Free energy barriers for the rate determining step are reported in bold
Page 7 of 8, 2250
B97D B3LYP MPWB1K RI-MP2
ΔG 10–11
ΔG 11–12
ΔG 12–13
ΔG 14–15
ΔG 15–16
ΔG 16–17
ΔG 17–18
8.4 14.0 9.0 12.2
21.2 25.6 26.1 26.8
27.7 25.9 26.3 29.1
5.7 6.7 6.4 7.0
3.5 6.1 5.3 4.4
32.6 28.7 27.5 35.4
7.8 11.1 10.4 9.6
comes at 2.802 Å from the metal. The bi-coordinated formate complex 17 is thus formed overcoming an energy barrier equal to 32.7 kcal mol-1. The last step of the catalytic cycle, required to regenerate the formate hydride complex 18, is the attack of a water molecule with the consequent formate switch from η2 to η1 coordination. The height of the barrier for the TS17–18 corresponding to this rearrangement is computed to be 7.8 kcal mol-1. The restored hydride formate complex 18 corresponds to complex 11 with exception of the formed hydrogen molecule in the second coordination sphere. The catalytic cycle can start again by replacement of a water ligand. In analogy with cycle 1, the same tests to validate the rate determining step of the process have been carried out. The outcomes of the calculations done removing all the extra molecules show that the activation barrier does not change. All the strategies used to intercept a concerted transition state to restore complex 18 do not provide any result. Also in this case both the meta-hybrid functional MPWB1K and RI-MP2 approach, used to compute the free energy barriers for all the steps involved in cycle 2, confirmed the release of H2 as the rate limiting step of the process. The outcomes of these calculations are collected in Table 2. Our computations reveal that molecular hydrogen cannot be produced more rapidly following cycle 2, as hypothesized by the authors to explain the observed difference between the first and subsequent cycles. Indeed, the computed activation barrier is higher than that calculated for the catalytic cycle 1 (32.7 kcal mol-1 versus 29.5 kcal mol-1). Summarizing, the mechanisms proposed and herein accurately investigated leave several open questions to be properly addressed. The hypothesis that CO2 release represents the rate determining step of cycle 1, whereas the same step occurs more rapidly in cycle 2 due to the trans effect of the hydride ligand, is not supported by the results of our computational analysis. On the basis of our results, cycle 2 does not provide an explanation for the difference between the first and the subsequent catalytic cycles.
Conclusions The mechanism of the selective decomposition of formic acid, without CO formation and under relatively mild conditions, assisted by [Ru(H2O)6](tppts)2]2+ as the catalyst for hydrogen
generation has been investigated in detail by using a model complex and applying accurate theoretical approaches. The experimental evidence that the decomposition of formic acid is slow, unless sodium formate is present as an initiator, and an induction period is required to generate the active catalyst, has prompted the authors to propose a mechanism in two cycles. According to their hypothesis, the rate-determining step of both proposed catalytic cycles should be the CO2 release, whose rate has been proposed to be higher in the second cycle due to the trans effect of a hydride ligand. Such hypothesis is not supported by the calculations presented here. First of all, CO2 elimination is not found to be the rate-determining step of both examined catalytic cycles, and the energy barriers for the elimination of molecular hydrogen is the highest of the catalytic cycles. This evidence seems to be a genuine result, since it has been confirmed not only by several tests focused on this key step of the reaction, but also employing different computational approaches to compute the energy barriers. Moreover, our computations reveal that molecular hydrogen cannot be produced more rapidly following cycle 2, as hypothesized by the authors, since it has a higher activation barrier. Given the relevance of the topic, further experimental information about both the characterization of intermediates and reaction rates should be welcome to unequivocally determine the mechanism of the process. Acknowledgments European Union, FP7-PEOPLE-2011-IRSES, project N. 295172, TEMM1P and Università della Calabria are gratefully acknowledged.
References 1. Zuttel A, Borgschulteand A (2008) In: Schlapbach L (ed) Hydrogen as a future energy carrier. Wiley VCH, Weinheim 2. Thomas KM (2009) Dalton Trans:1487–1505 3. Makowski P, Thomas A, Kuhn P, Goettmann F (2009) Energy Environ Sci 2:480–490 4. Demirci UB, Miele P (2009) Energy Environ Sci 2:627–637 5. Fukuzumi S (2008) Eur J Inorg Chem:1351–1362 doi: 10.1002/ ejic.200701369 6. Enthaler S (2008) ChemSusChem 1:801–804 7. Joó F (2008) ChemSusChem 1:805–808 8. Strauss SH, Whitmire KH, Shriver DF (1979) J Organomet Chem 174:C59–C62 9. Hinshelwood CN, Hartley HJ (1923) Chem Soc Trans 123:1333– 1338
2250, Page 8 of 8 10. Wakai C, Yoshida K, Tsujino Y, Matubayashi N, Nakamura M (2004) Chem Lett 33:572–573 11. Blake PG, Davies HH, Jackson GE (1971) J Chem Soc B:1923–1925 doi: 10.1039/J29710001923 12. Coffey RS (1967) Chem Commun 18:923–924 13. Forster D, Beck GR (1971) Chem Commun 994:1072 14. Laine RM, Rinker RG, Ford PC (1977) J Am Chem Soc 99:252–253 15. Yoshida T, Ueda Y, Otsuka S (1978) J Am Chem Soc 100:3941–3942 16. Paonessa RS, Trogler WC (1982) J Am Chem Soc 104:3529–3530 17. King RB, Bhattacharyya NK (1995) Inorg Chim Acta 237:65–69 18. Man ML, Zhou Z, Ng SM, Lau CP (2003) Dalton Trans:3727–3735 19. Johnson CT, Morris DJ, Wills M (2010) Chem Soc Rev 39:81–88 20. Loges B, Boddien A, Gärtner F, Junge H, Beller M (2010) Top Catal 53:902–914 21. Fukuzumi S, Kobayashi T, Suenobu T (2008) Chem Sus Chem 1: 827–834 22. Fukuzumi S, Kobayashi T, Suenobu T (2010) J Am Chem Soc 132: 1496–1497 23. Himeda Y (2009) Green Chem 11:2018–2022 24. Fellay C, Yan N, Dyson PJ, Laurenczy G (2009) Chem Eur J 15: 3752–3760 25. Gan W, Dyson PJ, Laurenczy G (2009) React Kinet Catal Lett 89: 205–213 26. Fellay C, Dyson PJ, Laurenczy G (2008) Angew Chem Int Ed 47: 3966–3968 27. Loges B, Boddien A, Junge H, Beller M (2008) Angew Chem Int Ed 47:3962–3965 28. Boddien A, Loges B, Junge H, Beller M (2008) ChemSusChem 1: 751–758 29. Alvarez R, Carmona E, Marin JM, Poveda ML, Gutiérrez-Puebla E, Monge A (1986) J Am Chem Soc 108:2286–2294 30. Bristow GS, Hitchcock PB, Lappert MF (1981) J Chem Soc Chem Commun:1145–1146 31. Ohnishi Y-Y, Matsunaga T, Nakao Y, Sato H, Sakaki S (2005) J Am Chem Soc 127:4021–4032 32. Becke AD (1993) J Chem Phys 98:5648–5652 33. Lee C, Yang W, Parr RG (1988) Phys Rev B Condens Matter Mater Phys 37:785–789 34. Andrae D, Häussermann U, Dolg M, Stoll H, Preuss H (1990) Theor Chim Acta 77:123–141
J Mol Model (2014) 20:2250 35. Fukui KJ (1970) Phys Chem 74:4161–4163 36. Gonzalez C, Schlegel HB (1989) J Chem Phys 90:2154–2161 37. Kovács G, Schubert G, Joó F, Pápai I (2005) Organometallics 24: 3059–3065 38. Donoso D, Mendizabal F (2011) Theor Chem Accounts 129:381– 387 39. Crabtree RH (2001) The organometallic chemistry of the transition metals. Wiley, New York 40. Cossi M, Rega N, Scalmani G, Barone V (2003) J Comput Chem 24: 669–681 41. Grimme S (2006) J Comput Chem 27:1787–1799 42. Piacenza M, Hyla-Kryspin I, Grimme S (2007) J Comput Chem 28: 2275–2285 43. Sieffert N, Bühl M (2010) J Am Chem Soc 132:8056–8070 44. Becke AD (1997) J Chem Phys 107:8554–8560 45. Grimme S (2004) J Comput Chem 25:1463–1473 46. Zhao Y, Truhlar DG (2004) J Phys Chem A 108:6908–6918 47. Sousa SF, Fernandes PA, Ramos MJ (2007) J Phys Chem A 111: 10439–10452 48. Bras NF, Moura-Tamames SA, Fernandes PA, Ramos MJ (2008) J Comput Chem 29:2565–2574 49. Möller C, Plesset MS (1934) Phys Rev 46:618–622 50. Gaussian 03, Revision C.02, Frisch MJ, Trucks GW, Schlegel HB, Scuseria GE, Robb MA, Cheeseman JR, Montgomery Jr. JA, Vreven T, Kudin KN, Burant JC, Millam JM, Iyengar SS, Tomasi J, Barone V, Mennucci B, Cossi M, Scalmani G, Rega N, Petersson GA, Nakatsuji H, Hada M, Ehara M, Toyota K, Fukuda R, Hasegawa J, Ishida M, Nakajima T, Honda Y, Kitao O, Nakai H, Klene M, Li X, Knox JE, Hratchian HP, Cross JB, Bakken V, Adamo C, Jaramillo J, Gomperts R, Stratmann RE, Yazyev O, Austin AJ, Cammi R, Pomelli C, Ochterski JW, Ayala PY, Morokuma K, Voth GA, Salvador P, Dannenberg JJ, Zakrzewski VG, Dapprich S, Daniels AD, Strain MC, Farkas O, Malick DK, Rabuck AD, Raghavachari K, Foresman JB, Ortiz JV, Cui Q, Baboul AG, Clifford S, Cioslowski J, Stefanov BB, Liu G, Liashenko A, Piskorz P, Komaromi I, Martin RL, Fox DJ, Keith T, Al-Laham MA, Peng CY, Nanayakkara A, Challacombe M, Gill PMW, Johnson B, Chen W, Wong MW, Gonzalez C, Pople JA (2004) Gaussian Inc, Wallingford, CT 51. Ahlrichs R, Bär R, Häser M, Horn H, Kölmel C (1989) Chem Phys Lett 162:165–169
