David
Balcells
a,
Odile
Eisenstein
ab,
Mats
Tilset
*ac and
Ainara
Nova
*a
aCentre for Theoretical and Computational Chemistry (CTCC) Department of Chemistry, University of Oslo, P.O. Box 1033, Blindern, N-0315 Oslo, Norway. E-mail: mats.tilset@kjemi.uio.no; ainara.nova@kjemi.uio.no
bInstitut Charles Gerhardt, CNRS UMR 5253, Université de Montpellier, F-34095 Montpellier, France
cDepartment of Chemistry, University of Oslo, P.O. Box 1033, Blindern, N-0315 Oslo, Norway
First published on 16th February 2016
The contribution of AuIII species to catalysis is still debated due to the limited number of characterized intermediates with this oxidation state. In particular, the coordination of alkenes and alkynes to AuIII followed by insertion into AuIII–X bonds has been suggested but rarely proven experimentally. Here, these reactions are explored by means of DFT and CCSD(T) calculations considering [AuX3(L)] and [AuX2(L)2]+ complexes. In these complexes, L = ethylene and acetylene have been chosen as substrates of high interest and representative of any unsaturated organic substrate, whereas X is Cl, Me or H, as found in metal salts and as model for intermediates involved in catalysis. Isoelectronic PtII complexes are also considered for comparison. Ethylene coordination occurs preferentially perpendicular for all X except H, whereas for acetylene, coordination takes place in-plane for all X except Cl. These coordination isomers can represent either minima (intermediates) or saddle points (transition states) on the potential energy surface, depending on X. NBO analysis shows how this variety of structures results from the combination of electronic (M–L donation and back-donation) and steric (cis L–X repulsion) effects. With the sole exception of [AuMe2(ethylene)2]+, rotation of the unsaturated ligand and insertion into a cis Au–X bond involve low to moderate energy barriers, ΔG‡ = 2.5 to 23.5 kcal mol−1, and are thermodynamically feasible, ΔG = 4.3 to −47.2 kcal mol−1. The paucity of experimental observations for such reactions should thus be caused by other factors, like the participation of the intermediates and products in competitive side reactions including the reductive elimination of XCHnCHnX (n = 1 or 2).
The X-ray structures of PtII–alkene and alkyne complexes show that the η2-coordinated ligand is perpendicular to the PtX3 plane,6,11 although rotation of the organic moiety takes place at room temperature.12 The nature of the Pt–alkene bond in Zeise's complex was first described by using the Dewar–Chatt–Duncanson model in 1953.13 Afterwards, computational methods were used to analyse the rotational barrier of L (ethylene and acetylene) in [PtCl3L]− complexes.14,15 With AuIII, NMR data on the η2-alkene complexes known to date are consistent with a perpendicular orientation.9 The same preference is observed in the crystal structure of [AuMe2(cod)]+, although the two CC double bonds of the cod ligand are not exactly parallel (dihedral angle between alkenes planes of 13.7°) and are probably influenced by the cyclic nature of the cod ligand.8 With alkynes, the preferred coordination mode with AuIII can only be inferred computationally. For instance, a DFT study by Pernpointner and Hashmi showed perpendicular orientation of propyne in the coordination to the [PtCl2(H2O)] and AuCl3 moieties.16
The orientation of the alkene in AuIII complexes is important because it can alter the nature of the reactions taking place in the coordination sphere of the metal centre. For instance, the perpendicular coordination of ethylene in [(tpy)Au(ethylene)(OCOCF3)]+ (tpy = 2-p-tolylpyridine) triggers the intermolecular addition of a free CF3CO2− anion to the coordinated ethylene anti to the metal, whereas the in-plane coordination triggers the unexpected intramolecular insertion of the alkene ligand into the AuIII–OCOCF3 bond (Fig. 2a).17 Although in this particular case the intermolecular addition is preferred experimentally and computationally, the other pathway cannot be fully excluded in other systems as shown by the insertion of norbornene into AuIII–Me bonds18 and of allene derivatives into AuIII–H bonds in complexes supported by pincer ligands (Fig. 2b and c respectively).19 Insertion of ethylene into AuIII–H has also been proposed computationally as an elementary step in the experimentally observed catalytic hydrogenation of ethylene.20 In addition, the insertion of alkynes and allenes into AuIII–Si bonds has been proved experimentally and studied computationally.21 To the best of our knowledge, these are the only cases for which an insertion reaction has been established in AuIII species. The scenario is similar with AuI, in which case few insertion reactions have been reported.22 In line with this, computational studies have shown that insertion of alkenes into AuI–H bonds has prohibitively high energy barriers.23 This is in contrast to the demonstrated ability of alkenes to insert into PtII–R (R = H, hydrocarbyl) bonds.24,25
Fig. 2 (a) Coordination modes of ethylene for the inter- and intramolecular reaction of OCOCF3− anion with AuIII coordinated ethylene. Insertion reactions of alkene into (b) AuIII–Me18 and (c) AuIII–H19 bonds. |
In this work, we explore with computational methods (DFT/CCSD(T)) the interaction of AuIII with double and triple CC bonds. We analyse the π complexes and the insertion reactions into AuIII–X bonds. The energetics of these systems will shed light on how AuIII catalysts form intermediates and initiate reactions in the presence of unsaturated organic molecules, information which is not available experimentally. For this purpose, we have used ethylene and acetylene due to their high interest as chemical feedstocks and their ability to represent any unsaturated organic substrate. The [LAuIIIX3] and [L2AuIIIX2]+ complexes (X = Cl, Me, and H; L = ethylene and acetylene) have been studied. X = Cl is considered because [AuCl3] is the most frequently used AuIII catalyst1,2 and [AuCl3L] is analogous to the well-known PtII Zeise salt7 (Fig. 1); X = Me because AuIII–alkyl species can be catalytic intermediates,26 and X = H because AuIII hydrides are involved in hydrogenation reactions.20 The [Au(ethylene)2(Me)2]+ complex is used to understand the geometry and electronic structure of [AuMe2(cod)]+, which is to date the only AuIII–alkene complex characterized by X-ray diffraction.8 In addition, isoelectronic PtII complexes have been used as further reference because of the larger number of isolated related complexes. The insight given by this study can contribute significantly to the isolation of AuIII–alkyne complexes, which is still unprecedented, and the design of new AuIII-catalyzed reactions in which product formation or catalyst deactivation involves insertion steps.
Fig. 3 Structure labelling (A–E) for the rotation and insertion of L = ethylene (en) and acetylene (yn) in [AuIIIX3L] complexes (X = Cl, Me, H). |
The free energy profiles connecting structures A–E are shown in Fig. 4. For ethylene and X = Cl, the profile depends dramatically on the metal centre. With PtII, the in-plane coordinated ethylene, Pt-B-en-Cl is not an energy minimum but rather a bifurcation point associated with the rotation of ethylene in Pt-A-en-Cl, and is connected with the transition state (TS) for the ethylene insertion into a cis-Pt–Cl bond, Pt-C-en-Cl.37 In contrast, for AuIII, both Au-A-en-Cl and Au-B-en-Cl are energy minima connected by a low-energy rotation transition state (not shown in Fig. 4 for clarity; see ESI†).38 The same scenario applies for X = Me with AuIII. Nonetheless, the nature of X also influences strongly the Gibbs free energy profile, because with X = H, Au-A-en-H, which is less stable than Au-B-en-H, is no longer a minimum but the TS for ethylene rotation. The insertion of ethylene into a cis-Pt/Au–X bond involves transition state C and product D, a T-shaped 3-coordinate species in which the two remaining X ligands are cis to each other. The cis–trans isomerization transition state was found to be at lower energy than transition state C (X = Cl, see ESI†) and in the case of X = Me product E is obtained directly from C.39 Therefore, this barrier was not studied further.
The energy patterns for the rotation and insertion of ethylene are also significantly different for [AuCl3(ethylene)] and [PtCl3(ethylene)]− (Fig. 4).40 With AuIII, the energy differences between the perpendicular (Au-A-en-X) and the in-plane (Au-B-en-X) isomers are much lower than with PtII. The A → B isomerization is endoergic for X = Cl (8.0 kcal mol−1) and Me (1.2 kcal mol−1) showing a preference for the perpendicular isomer. In contrast, with X = H, A becomes the TS for rotation with a low barrier of 7.2 kcal mol−1 above B. The way electronic and steric effects combine to lead to these differences will be discussed below (vide infra). The rotation energy barriers with X = Cl and Me (not shown in Fig. 4 for clarity) are also low, i.e. 8.8 and 2.8 kcal mol−1 respectively. This indicates that facile or essentially free Au–alkene rotation should be expected at room temperature for AuX3 alkene complexes, at least in the absence of steric hindrance inhibiting the rotation in substituted alkenes.
The TS energies for insertion into the M–X bond (C-en-X) are higher and follow a different trend when compared to rotation. While the stability of the in-plane isomer preceding the insertion TS decreases in the order H > Me > Cl, the insertion barrier decreases in the order Me > Cl > H. For TS C, in contrast to B, the capacity of X to migrate from the metal to the alkene should play a relevant role. This should be a more facile process with the p lone pairs of X = Cl and the spherical electron distribution at H, than with the directionally highly localized sp3 bonding electron pair of X = Me. The direct insertion product is Au-D-en-X for X = Cl and H, whereas X = Me yields Au-E-en-Me. This results from the difference in the trans influence of X (lowest for Cl, highest for H) and the additional non-covalent Au⋯Cl (2.439 Å) and Au⋯H interactions (1.939 Å) that stabilize D and may become repulsive with X = Me.41 Besides the low energy barriers, insertion is thermodynamically favourable by 4.8 and 6.8 kcal mol−1 with X = Cl and Me. With X = H, Au-B-en-H is the thermodynamic sink, however the moderate free energy difference between B and E, i.e. ΔG‡ = 8.1 and ΔG = 4.3 kcal mol−1 suggests an equilibrium between both species. In contrast, with PtII the insertion has a prohibitive energy barrier of 41.8 kcal mol−1 and is endoergic by 26.5 kcal mol−1. The differences in insertion ability between AuIII and PtII–ethylene suggest that this reaction could be involved in decomposition processes of AuIII–alkene complexes, which are not observed with PtII, as showed by the high stability of Zeise's salt and analogous [PtCl3(alkene)]− complexes.7,11
The energy trends obtained for ethylene rotation and insertion in A–E are also found for acetylene but the energies of all stationary structures are consistently shifted to lower energies relative to A. This is especially pronounced for platinum, where the insertion barrier is lowered from 41.8 to 27.2 kcal mol−1. The latter barrier agrees with the observation of slow acetylene insertion into the Pt–Cl bond in [PtCl2(N^N)] (N^N = 2,9-dimethyl-1,10-phenanthroline).42 For gold, the in-plane coordination of acetylene is preferred not only for X = H, but also for X = Me (Au-B-yn), with energy barriers for rotation (A-yn) of 9.7 and 2.5 kcal mol−1, respectively. For X = Cl, the rotation barrier of acetylene is also very low (3.5 kcal mol−1), with the in-plane coordination structure, Au-B-yn-Cl, being the TS. Taking as a reference point the lowest energy isomer of M–ethylene and acetylene, i.e.A or B, respectively, the insertion barriers are significantly lower for acetylene (19.1, 9.9 and 7.5 kcal mol−1 for X = Me, Cl, and H, respectively) than for ethylene (23.5, 16.3 and 8.1 kcal mol−1 for X = Me, Cl, and H, respectively). The lower energies of C, D, and E with a partially or fully broken π-bond in the case of the alkyne relative to alkene are due to several factors including (i) the lower energy loss for breaking one π bond in a triple bond compared to in a double one, (ii) the weaker coordination of acetylene compared to ethylene (as seen in the isodesmic reactions, eqn (1) and (2), and (iii) the stronger M–C(sp2) bond in the M–vinyl insertion product from acetylene, compared to the M–C(sp3) bond in the M–alkyl insertion product from ethylene.43 In addition, for X = H, the reaction is further thermodynamically driven by the reductive elimination of H2 yielding AuI-yn-H2 (Fig. 4).
These results clearly show that ethylene or acetylene insertion into Au–X bonds (X = Cl, Me, and H) is not encumbered by prohibitive energy barriers or endoergic steps (with the exception of the ethylene insertion into Au–H). These processes may be even faster in polar solvents, as suggested by the ca. 1–2 kcal mol−1 decrease in the energy barriers for coordination, rotation and insertion of ethylene in Au-Cl-en when considering benzene and TFE as solvents (Fig. S2†). Nonetheless, this reaction cannot occur if generation of either A and even more B has a high energy cost that needs to be added to the energetics of the insertion process itself. This may be indeed the case considering the weak coordination of an unsaturated ligand to AuIII relative to PtII, as shown for ethylene by the thermochemistry of eqn (3). This is primarily caused by the reduced back-donation from Au (vide infra). There are other factors that may reduce the lifetime of the Au-π adduct and preclude possible insertion processes. For instance, addition of an external nucleophilic molecule (solvent or substrate) to L may be energetically facile.17 In addition to the transient nature of A or B, the insertion product, E, may be scavenged by protonation9 or reductive elimination44 processes, which can be promoted by low energy barriers and tunneling effects.45 Indeed, at AuH3, the insertion of acetylene into one Au–H bond proceeds with concomitant reductive elimination of H2 (Fig. 4). With ethylene, the reductive elimination of XCH2CH2X calculated from the trans insertion products (E) has an energy barrier of 7.2 and 23.3 kcal mol−1 for X = Me and Cl, respectively (eqn (4)). The corresponding barriers for the elimination yielding XCHCHX from the acetylene insertion products are even lower, 0.3 and 11.8 kcal mol−1 for X = Me and Cl, respectively (eqn (5)). The poor stability of AuIII alkene and alkyne complexes, as inferred from these thermodynamic (A/B stability) and kinetic (reductive elimination from E) data, accounts for the limited success attained so far in the isolation and characterization of such species as well as their immediate insertion products.
(1) |
(2) |
(3) |
(4) |
(5) |
In the following discussion of ethylene rotation, the rotation angle θ is defined as the angle between the molecular plane of Au and the plane defined by Au and the two C atoms of each ethylene. Thus, for θ = 90° and 0°, the ethylene is perpendicular and in-plane, respectively (see Fig. 5). The optimized geometry of [AuX2(ethylene)2]+ corresponds to neither of the structures in Fig. 5, since θ = 77° yielding a species labelled Au-F′-en-Me (Fig. 6). A similar deviation from the idealized perpendicular structure was also seen in [AuMe2(cod)]+, with θ = 85°. Surprisingly, the deviation away from the perpendicular structure is even more pronounced in Au-F′-en-Me than in the crystal structure of [AuMe2(cod)]+. Thus, it appears that the deviation from θ = 90° is not only imposed by the constrained cyclic nature of the cod ligand, but rather reflects an intrinsic feature of the AuIII–alkene bond (vide infra).
In [AuMe2(ethylene)2]+, rotation of one ethylene ligand has a very low energy barrier of 3.9 kcal mol−1 (Au-G-en-Me, Fig. 6), and should take place even at low temperatures. In contrast, insertion of ethylene into the Au–Me bond has a prohibitively high energy barrier of almost 40 kcal mol−1 (Au-J-en-Me). The dramatic influence of the ligand trans to the Au–Me bond that undergoes insertion becomes apparent when this barrier is compared to that associated with Au-C-en-Me (23.5 kcal mol−1, Fig. 4). In line with these data, no insertion products are observed with [AuMe2(cod)]+.8
In order to determine the influence of the X group in [AuX2(ethylene)2]+ and [AuX2(acetylene)2]+ on the orientation of the unsaturated ligands from the ideal structures shown in Fig. 5, geometries were optimized for X = Cl and H (Table 1). The related Pt complexes were also considered. With X = Cl, all geometries are almost ideal with two perpendicular oriented double or triple bonds (F), whereas for X = Me, all geometries deviate significantly (F′). In contrast, X = H shows a distinct scenario, with several different structures, including a AuI dihydrogen complex, [Au(H2)(ethylene)2]+ (AuI-en2-H2), I′, and G. This structural diversity suggests that the weak bonding of alkenes and alkynes to AuIII allows for multiple structural distortions likely dominated by dispersion forces, including weak Hδ+⋯Hδ−, CH⋯π and CH⋯Cl interactions.
The trend observed in the SE () for Au-A-en-X, Cl > Me > H, can be ascribed to the increasing trans influence of these ligands (Cl < Me < H), which raises the energy of the empty orbital in the same order. This effect causes ethylene to coordinate more strongly in Au-A-en-Cl than in Au-A-en-Me and Au-A-en-H (eqn (6) and (7)). In addition to the trans influence of Xt there is a cis influence of Xc, specifically in the geometry of B, which involves donation of electron density from an occupied orbital localized on the ligand (the σAu–Xc-bond) to . This donation decreases in the order H > Me > Cl as shown by the green bars in Fig. 8a.
(6) |
(7) |
The different trend observed in the donation in A (Au–Cl > Au–Me > Au–H) and B (Au–H > Au–Cl > Au–Me) can be understood by considering the steric repulsion between the occupied orbitals of Xc and ethylene by using natural steric analysis.47 The steric exchange energies between ethylene and Xc (Fig. 8b)48 clearly show a higher steric hindrance in B than in A, which follows the trend H < Me < Cl. In B, the highest exchange energy with X = Cl is mainly due to the interaction between a lone pair on Cl(p) with the CC(π) and the CH bonds of ethylene. This steric hindrance is also reflected in the deviation from 90° of the square planar geometry for Clt–Au–Clc angle (87.4°) and the Au–ethylene elongation by 0.07 Å when going from A to B (Table 2). In contrast, with X = H and Me, the Au–ethylene distances are shorter in B than in A by 0.04 and 0.13 Å, respectively, in agreement with the higher stabilization energies in the former.
The structural preferences for [MX2(ethylene)2]n+ and [MX2(acetylene)2]n+ (M = Au, n = 1; M = Pt, n = 0; X = Cl, Me, and H) shown in Table 1 can also be explained by a similar combination of electronic and steric interactions shown in Fig. 7 and 8. In particular, the higher cis influence of hydride (Fig. 8) together with the small size of this atom may account for the geometrical diversity obtained for X = H (AuI-H2-en, Au-I′-yn, Pt-G-en, and Pt-I′-yn) compared to X = Cl and Me (F for all X = Cl and F′ for all X = Me). In the latter complexes, both donation and back-donation are involved.8 Reduction of the former interaction when moving from Au-F-en-Cl to Au-F′-en-Me (from 130.4 to 70.1 kcal mol−1) due to the highest trans influence of Me is consistent with the longer Au–ethylene distances in Au-F′-en-Me (2.46 Å) compared to Au-F-en-Cl (2.31 Å). The weak interaction of ethylene in Au-E′-en-Me probably explains that distortion from the ideal θ angle, to increase the dispersion forces between the two coordinated ethylenes,49 does not imply a loss of energy.
The preference for the perpendicular coordination in [PtCl3(ethylene)]−, [AuX3(ethylene)], [PtX2(ethylene)2], and [AuX2(ethylene)2]+ for X = Me and Cl obtained in this study (Fig. 4 and Table 1) agree with the observation of this coordination mode in crystal structures of η2-coordinated alkene in mononuclear PtII complexes,6,11 with the crystal structure of [AuMe2(cod)]+8 and with the NMR characterization of the AuIII-norbornene pincer complex9 shown in Scheme 1. According to the present study the two main factors explaining the low success in characterizing η2-alkene complexes of Au compared to Pt are: (1) the weakest AuIII–alkene interaction (eqn (3)), and (2) the lowest energy barrier for inserting the alkene into the Au–X bond in AuX3L type systems (Fig. 4), which reduce the lifetime of the η2-alkene complexes. The same factors account for the unsuccessful characterization of AuIII-η2-alkyne complexes in which the M–L interaction is even weaker and the insertion energy barrier lower than with alkene. Chelating ligands with larger donor character and steric bulk should thus stabilize these intermediates and facilitate their characterization, due to their ability to strengthen the Au–L interaction and prevent L insertion into Au–X by disfavouring or excluding in-plane coordination. Our X-ray characterized [AuMe2(cod)]+ complex does indeed fall into this category.
Concerning AuIII, the energy barriers calculated for the insertion of ethylene in [AuX3(ethylene)] (X = Cl, Me, and H) are 23.5, 16.3 and 8.1 kcal mol−1, respectively. These energies suggest that the insertion process should be feasible at room temperature and may be competitive to intermolecular addition processes. In addition to the nature of the X group, for X = Me, the energy barrier for the insertion process depends on the ligand trans to the Me group that migrates (Scheme 1, blue). Hence, when the trans ligand is Me, i.e. [AuMe3(ethylene)], the energy barrier is 23.5 kcal mol−1, whereas when this ligand is ethylene, i.e. [AuMe2(ethylene)2]+, the energy barrier is 39.5 kcal mol−1. This energy difference may be due to the higher trans effect of Me compared to ethylene (Scheme 1, both blue), which weakens the trans Au–Me bond (Scheme 1, red). The lower donation in Au-F′-en-Me (70 kcal mol−1 from the NBO analysis) compared to the donation in Au-A-en-Me (329 kcal mol−1)51 is also consistent with the difficult migration of the Me group (Scheme 1, red) to ethylene in the former system.
In addition to the trans effect, other factors may determine the preference for an intermolecular nucleophilic addition rather than an insertion process, e.g. the ability of some metal-free ligands to act as external nucleophiles. This would be the case for alcohols, amines, and anions such as CF3CO2−,2,17 but not for other anions like alkyl− and H−. In the latter case, insertion should be preferred as it has been proposed for the reaction of norbornene with [(P–C)AuMe2] complex (P–C = cyclometalated diisopropylnaphthylphosphine (Fig. 2b).18 In this reaction, the cyclometalated system is activated by methyl anion abstraction with B(C6F5)3 prior to norbornene coordination and insertion. To our knowledge, this is the only example in which a product derived from alkene insertion into a AuIII–Me bond has been isolated and characterized. A possible reason for this paucity of precedents is the short-lifetime expected for the insertion products due to fast reductive elimination in tricoodinated AuIII species, as it has been shown in this study (eqn (4) and (5)). This decomposition process is inhibited in the reaction shown in Fig. 2b because of the chelating nature of the ancillary P–C ligand.
Footnote |
† Electronic supplementary information (ESI) available: Further computational details available include: (i) DFT functional benchmarking, (ii) complete energy profiles for [AuX3(ethylene)] with X = Cl and Me, (iii) NBO analysis with L = acetylene and (iv) optimized coordinates and energies. See DOI: 10.1039/c5dt05014f |
This journal is © The Royal Society of Chemistry 2016 |