F. Mark
Chadwick
a,
Alasdair I.
McKay
a,
Antonio J.
Martinez-Martinez
a,
Nicholas H.
Rees
a,
Tobias
Krämer
b,
Stuart A.
Macgregor
*b and
Andrew S.
Weller
*a
aDepartment of Chemistry, Chemistry Research Laboratories, University of Oxford, OX1 3TA, UK. E-mail: andrew.weller@chem.ox.ac.uk
bInstitute of Chemical Sciences, Heriot Watt University, Edinburgh, EH14 4AS, UK. E-mail: S.A.Macgregor@hw.ac.uk
First published on 6th July 2017
Single-crystal to single-crystal solid/gas reactivity and catalysis starting from the precursor sigma-alkane complex [Rh(Cy2PCH2CH2PCy2)(η2η2-NBA)][BArF4] (NBA = norbornane; ArF = 3,5-(CF3)2C6H3) is reported. By adding ethene, propene and 1-butene to this precursor in solid/gas reactions the resulting alkene complexes [Rh(Cy2PCH2CH2PCy2)(alkene)x][BArF4] are formed. The ethene (x = 2) complex, [Rh(Cy2PCH2CH2PCy2)(ethene)2][BArF4]-Oct, has been characterized in the solid-state (single-crystal X-ray diffraction) and by solution and solid-state NMR spectroscopy. Rapid, low temperature recrystallization using solution methods results in a different crystalline modification, [Rh(Cy2PCH2CH2PCy2)(ethene)2][BArF4]-Hex, that has a hexagonal microporous structure (P6322). The propene complex (x = 1) [Rh(Cy2PCH2CH2PCy2)(propene)][BArF4] is characterized as having a π-bound alkene with a supporting γ-agostic Rh⋯H3C interaction at low temperature by single-crystal X-ray diffraction, variable temperature solution and solid-state NMR spectroscopy, as well as periodic density functional theory (DFT) calculations. A fluxional process occurs in both the solid-state and solution that is proposed to proceed via a tautomeric allyl-hydride. Gas/solid catalytic isomerization of d3-propene, H2CCHCD3, using [Rh(Cy2PCH2CH2PCy2)(η2η2-NBA)][BArF4] scrambles the D-label into all possible positions of the propene, as shown by isotopic perturbation of equilibrium measurements for the agostic interaction. Periodic DFT calculations show a low barrier to H/D exchange (10.9 kcal mol−1, PBE-D3 level), and GIPAW chemical shift calculations guide the assignment of the experimental data. When synthesized using solution routes a bis-propene complex, [Rh(Cy2PCH2CH2PCy2)(propene)2][BArF4], is formed. [Rh(Cy2PCH2CH2PCy2)(butene)][BArF4] (x = 1) is characterized as having 2-butene bound as the cis-isomer and a single Rh⋯H3C agostic interaction. In the solid-state two low-energy fluxional processes are proposed. The first is a simple libration of the 2-butene that exchanges the agostic interaction, and the second is a butene isomerization process that proceeds via an allyl-hydride intermediate with a low computed barrier of 14.5 kcal mol−1. [Rh(Cy2PCH2CH2PCy2)(η2η2-NBA)][BArF4] and the polymorphs of [Rh(Cy2PCH2CH2PCy2)(ethene)2][BArF4] are shown to be effective in solid-state molecular organometallic catalysis (SMOM-Cat) for the isomerization of 1-butene to a mixture of cis- and trans-2-butene at 298 K and 1 atm, and studies suggest that catalysis is likely dominated by surface-active species. [Rh(Cy2PCH2CH2PCy2)(η2η2-NBA)][BArF4] is also shown to catalyze the transfer dehydrogenation of butane to 2-butene at 298 K using ethene as the sacrificial acceptor.
The dehydrogenation of light alkanes such as butane and pentane, and their subsequent isomerization is particularly interesting, as while these alkanes are unsuitable as transportation fuels or feedstock chemicals, their corresponding alkenes have myriad uses.32,33,35 The discovery of abundant sources of light alkanes in shale and offshore gas fields provides additional motivation to study their conversion into fuels and commodity chemicals.36 As light alkanes are gaseous at, or close to, room temperature and pressure the opportunity for solid/gas catalytic processes under these conditions is presented. Such conditions are also attractive due to the physical separation of catalyst and substrates/products that they offer, as well as opportunities to reduce thermally-induced catalyst decomposition processes.
Although heterogeneous solid/gas systems for alkane dehydrogenation and alkene isomerization are well known,29,37–40 they often require high temperatures for their operation which lead to reductions in selectivity as well as catalyst deactivation through processes such as coking. As is often the case41,42 well-defined supported or molecular systems can offer lower barriers, albeit still having to overcome the endergonic dehydrogenation reaction when run without a sacrificial acceptor.43,44 As far as we are aware there are only a handful of examples of purely molecular, i.e. not supported, solid-phase catalysts for alkane dehydrogenation or alkene isomerization. The Ir-pincer catalysts, such as Ir(PCPiPr)(C2H4) [PCPiPr = κ3-C6H3-2,6-(CH2PiPr2)2], recently reported by Goldman and co-workers, promote the transfer dehydrogenation, and subsequent double-bond isomerization, of butane, pentane and octane using acceptors such as tert-butylethene, ethene or propene.45,46 These operate at temperatures of 200 °C or above in sealed-tube conditions in which all the alkane is expected to be in the gas phase, and can actually outperform homogeneous systems in terms of activity. Experimental evidence points towards a presumed molecular species as the active catalyst, although the precise details have not been disclosed. Siedle & Newmark reported the room temperature solid/gas isomerization of simple alkenes using iridium or rhodium phosphine cations partnered with Keggin-type trianions, such as [Ir(H)2(PPh3)2]3[PW12O40],47–49 however the precise molecular structure of the catalyst was not determined.
We have recently reported the synthesis, using single-crystal to single-crystal solid/gas techniques,41,50–53 of well-defined sigma-alkane complexes,54,55 typified by [Rh(R2PCH2CH2PR2)(η2η2-NBA)][BArF4] [R = iBu, Cyp, Cy; NBA = norbornane; ArF = 3,5-(CF3)2C6H3]; Scheme 2a.56–59 The key to these complexes' relative stability in the solid-state is the arrangement of [BArF4]− anions that provide a well-defined cavity (i.e. they are “crystalline molecular flasks”,60,61Scheme 2b) that results in very small changes in unit cell volumes and retention of crystallinity during the transformations of the organometallic cation. This allows for the characterization of products directly by single-crystal X-ray crystallography and solid-state NMR spectroscopy (SSNMR). These complexes, some of which are stable at room temperature (e.g. R = Cy, [1-NBA][BArF4]), allow for the reaction chemistry of sigma-alkane complexes to be probed using solid/gas experimental techniques, for example C–H activation processes.62 Of relevance to this paper is the use of the alkane as a labile ligand that can be readily displaced in solid/gas reactivity and catalysis. We have recently reported that [Rh(iBu2PCH2CH2PiBu2)(η2η2-NBA)][BArF4] reacts with ethene to form [Rh(iBu2PCH2CH2PiBu2)(ethene)2][BArF4], that will catalyze ethene hydrogenation using solid/gas techniques, and also briefly commented on 1-butene isomerization.63
Scheme 2 (a) Synthesis of sigma-alkane complex [1-NBA][BArF4] by solid/gas reactivity; (b) Oh arrangement of [BArF4] anions in the solid-state that encapsulate cation (ArF groups removed). |
This [Rh(iBu2PCH2CH2PiBu2)(η2η2-NBA)][BArF4] system can suffer from loss of crystallinity in substitution reactions in the solid-state, as well as thermal instability to form the [BArF4]− coordinated zwitterion that is a poor catalyst. By contrast [1-NBA][BArF4], with its more rigid cyclohexyl groups, is stable as a crystalline solid for months at 298 K under an Ar-atmosphere, although on dissolution – even at very low temperature in CDFCl2 – the zwitterion [Rh(Cy2PCH2CH2PCy2)(η6-3,5-(CF3)2C6H3)BArF3], [1-BArF4], is immediately formed reflecting the weak binding of the alkane ligand (ca. 80 kJ mol−1 or less).54–56 This weak binding, albeit stabilized in the solid-state, suggests that [1-NBA][BArF4] may provide the ideal platform for studying solid/gas reactivity and catalysis in exceptionally well-defined molecular systems, providing a highly reactive {Rh(bis-phosphine)}+ fragment with cis vacant (or at least very weakly stabilized) sites, Scheme 3. We report here that this is the case, and show that the alkane ligand in [1-NBA][BArF4] can be substituted for ethene, propene and 1-butene to give well-defined alkene complexes, some of which can only be prepared using such solid/gas routes. For propene and butene complexes rapid double-bond isomerization processes occur in the solid-state, which have been probed using variable temperature solid-state NMR spectroscopy, D-labelling studies and periodic DFT calculations. These exceptionally well-defined crystalline systems, which we term solid-state molecular organometallic catalysts (SMOM-Cat), are also active precatalysts for the solid/gas isomerization of 1-butene, demonstrating structure/activity relationships between the extended molecular structure and the measured catalytic activity. They also catalyze the transfer dehydrogenation/isomerization of butane to 2-butene.
Scheme 3 Generation of active cis-latent sites by solid/gas reactivity by displacement of a weakly bound alkane ligand. |
Scheme 4 Synthesis of [1-(ethene)2][BArF4] as octahedral (“Oct”, C2/c) and hexagonal (“Hex”, P6322) polymorphs. |
Solution 1H and 31P{1H} NMR spectroscopy (CD2Cl2, Ar atmosphere, 193 K) of a freshly dissolved sample prepared in the solid-state were also fully consistent with formulation as a bis-ethene complex. In particular, at 193 K a sharp doublet at δ 73.6 [J(RhP) = 145 Hz] was observed, while in the 1H NMR spectrum bound ethene (8 H relative integral) was observed at δ 4.15. Warming to 298 K resulted in a broadening of all these signals, but no significant chemical shift change. After only 20 minutes at 298 K in CD2Cl2 solution significant decomposition had occurred, even when placed under an ethene atmosphere, to give unidentified products. Dissolving [1-(ethene)2][BArF4]-Oct in 1,2-F2C6H4 solvent returned [1-F2C6H4][BArF4].
Remarkably, given that NBA is being expelled, this transformation is also a single-crystal to single-crystal one in the solid-state, as shown by an X-ray structure determination at 150 K. Starting from [1-NBD][BArF4] (Scheme 2) this represents a rare example of a sequential reaction sequence for such processes.50 We suggest that the CF3 groups on the anions result in some plasticity of the solid-state lattice, which allows for the movement of the NBA,65 given that there are no significant channels in the crystal lattice. There is a space group change from to P21/n (Z = 4) to C2/c (Z = 4) on substitution, and we,56,59 and others,52,66–68 have commented upon similar changes previously in solid/gas reactions. The final refined structural model (Fig. 1a) has a significant R-factor (10.7%) which we attribute to an increase in mosaicity on the single-crystal to single-crystal transformation and the loss of some high-angle data.
Nevertheless the refinement is unambiguous and shows a [Rh(Cy2PCH2CH2PCy2)(η2-C2H4)2]+ cation encapsulated by an almost perfect octahedron of [BArF4]− anions in the extended lattice (Fig. 1b). There is crystallographically imposed C2 symmetry. The ethene ligands on the Rh-center are disordered over two sites, with a CC distance of 1.36(1) Å, consistent with a double bond and are also canted slightly from the square plane by 14°. Similar distortions have been noted in trans-[Rh{PR2(alkene)}2]+ species and are thought to be driven by enhanced π-back donation from the Rh dz2 orbital.69[1-(ethene)2][BArF4]-Oct is stable to short periods of vacuum but satisfactory elemental analysis was not obtained as the NBA formed during the reaction was persistent and could not be removed.
[1-(Ethene)2][BArF4]-Oct is a rare example of a bis- or tris-ethene adduct of a simple {Rh(PR3)n}+ fragment, which in solution are generally sensitive to loss of ethene.70–72 Bis-ethene complexes with other supporting ligand sets are more common. This scarcity no doubt reflects the instability of species such as [1-(ethene)2][BArF4] in solution, and highlights the benefits of the solid/gas technique. This allows for [1-(ethene)2][BArF4]-Oct to be reliably prepared in ∼0.2 g batches (unoptimized).
Over time, in the solid-state under an ethene atmosphere (1 atm), the butadiene complex, [Rh(Cy2PCH2CH2PCy2)(η2η2-C4H6)][BArF4] [1-(butadiene)][BArF4], slowly forms (weeks), Scheme 5, as measured by 31P{1H} SSNMR. Interrogation of the head-space using gas-phase NMR spectroscopy after 1 week shows that approximately 1 equivalent of 2-butene is also formed, arising from initial ethene coupling and subsequent isomerization. [1-(Butadiene)][BArF4] is better made directly from addition of excess 1-butene to [1-F2C6H4][BArF4] in solution (see later). [1-(Butadiene)][BArF4] presumably forms in the solid-state via dehydrocoupling and loss of H2, as previously reported for [Rh(iBu2PCH2CH2PiBu2)(ethene)2][BArF4];63,73 possibly aided by sacrificial ethene, as ethane was also observed.
Porous materials made from metal–organic frameworks (MOFs) are well known and can be used for a wide range of applications including gas separation and catalysis, and can often incorporate reactive metal sites as part of the framework,52,78–83 or as an encapsulated cation in an anionic porous network.84,85 However porous organometallic materials that are principally constructed from non-covalent interactions are less common.51,86–89 As far as we are aware [1-(ethene)2][BArF4]-Hex represents a rare example where the likely site of any potential catalytic activity, that is labile ethene ligands, are focused directly into the pore, being similar to that reported by Brookhart for Ir(POCOP)(C2H4) [POCOP = κ3-C6H3-2,6-(OP(C6H2-2,4,6-(CF3)3)2)].51 The contrast between the extended structure of [1-(ethene)2][BArF4]-Oct and its polymorph [1-(ethene)2][BArF4]-Hex is dramatic. As demonstrated (Section 2.5) this leads to a significant difference in their ability to promote 1-butene isomerization catalysis when in single-crystalline form.
Scheme 6 Synthesis of [1-(propene)x][BArF4] by solution (x = 2) and solid/gas (x = 1) single-crystal to single-crystal reactivity. |
The solid-state structure of [1-(propene)][BArF4] is shown in Fig. 3. The C3-hydrocarbon is disordered over two positions, related by a non-crystallographic two-fold rotation, Fig. 3b; while the octahedral arrangement of anions is retained, Fig. 3c. As for [1-(ethene)2][BArF4]-Oct the solid/gas reaction led to loss in high-angle data and a reduction in the quality of the refinement (R = 12.7%). This, alongside the disordered organic fragment, means that a detailed discussion of the bond lengths and angles is not appropriate, and the hydrogen atoms were placed in calculated positions. Although the two C–C distances appear to show differentiation between C–C and CC bonds [e.g. C100–C200, 1.361(9); C200–C300, 1.239(9) Å], both measure shorter than might be expected (and calculated, vide infra)91,92 which likely is a consequence of the poor structure and rotational disorder. All three Rh–C distances reflect Rh–C bonding interactions, but within error are the same [e.g. 2.15(2)–2.29(3) Å]. Thus, although the gross structure is unambiguous in showing a single C3 fragment bound to the metal center, whether it is an η2-bound propene with a supporting γ-agostic93 interaction (I, Scheme 7) or the isomeric allyl-hydride, that arises from γ-C–H activation of propene (II),22,49,94–97 cannot be determined due to the quality of the data. We thus turned to variable temperature SSNMR and solution NMR spectroscopy, as well as periodic DFT calculations, to determine the precise structure. These studies show that at low temperatures the alkene/agostic tautomer is favored, which at higher temperatures accesses the allyl-hydride in both solid-state and solution.
The 158 K 31P{1H} SSNMR spectrum of [1-(propene)][BArF4], prepared in situ, shows two broad environments at δ 101.3 and 90.4, consistent with the two different phosphorus environments in the single-crystal X-ray structure. At this low temperature in the 13C{1H} SSNMR spectrum two, approximately equal intensity, signals are observed in the region associated with bound alkene ligands,98 at δ 94.2 and 78.8, alongside a high-field signal at δ 6.5 in the region indicative of an agostic M⋯H3C interaction.99 Warming to 298 K results in a broad, but asymmetric, 31P signal in the SSNMR spectrum at δ 95.6; while in the 13C{1H} SSNMR spectrum a broad signal was observed at δ 93.7, and the high-field signal present at 158 K was absent. These data suggest a fluxional process is occurring in the solid-state at room temperature,100 that is slowed at 158 K. Low temperature 1H/13C HETCOR experiments, that we56,62 and others101 have previously shown to be useful in determining 1H NMR chemical shifts for sigma interactions in the solid-state, were not successful. The variable temperature 31P{1H} NMR data have been modelled using rate-constants derived from a line-shaped analysis, and a resulting Eyring analysis gives ΔG‡ = 10(1) kcal mol−1 and ΔS‡ = −7(3) cal K−1 suggesting a slightly ordered transition-state.
Similar behavior is observed using solution NMR spectroscopy when [1-(propene)][BArF4], prepared by the solid/gas route, is dissolved in CD2Cl2. Although rapid decomposition (less than 30 minutes) occurs at 298 K to give unidentified products, immediate data collection led to reliable solution NMR data. At 298 K the 31P{1H} NMR spectrum shows a single environment δ 95.2 [J(RhP) = 181 Hz], while the 1H NMR spectrum shows a very broad signal at δ 5.07 of relative integral ∼1 H in addition to signals in the aliphatic and aryl regions. Such a chemical shift is characteristic of the methine proton in an η3-allyl ligand.98,102 The hydride region was featureless. These data suggest a fluxional process is also occurring in solution at 298 K. Cooling to 193 K in CD2Cl2 slows both decomposition and the fluxional process. The 31P{1H} NMR spectrum now shows two environments, δ 100.4 [J(RhP) = 200 Hz] and δ 89.9 [J(RhP) = 161 Hz], similar to those measured in the SSNMR spectrum at 158 K, with the larger coupling constant suggesting a weakly-bound trans ligand. The 1H NMR solution spectrum at 193 K shows three integral 1-H environments in the alkene region [δ 4.84, 4.54, 3.55] and high-field integral 3-H signal at δ −0.02 assigned to the methyl group that includes the agostic C–H⋯Rh interaction that is undergoing rapid rotation. The signals at δ 4.54, 3.55 and −0.02 become broad on warming, and disappear into the baseline at 253 K suggesting that they are mutually exchanging. In contrast the signal assigned to the methine proton remains essentially unchanged in chemical shift, and can be tracked to the broad signal observed at δ 5.07 at 298 K.
Insight into the detailed structure of the propene adduct [1-(propene)][BArF4] was obtained via periodic density functional theory (DFT) calculations at the PBE-D3 level, where this approach has previously been shown to reproduce the solid-state structures and fluxionality of related sigma-alkane complexes very effectively.54,58 Geometry optimization of [1-(propene)][BArF4] based on one component of the crystal structure (using propene carbon positions C100, C200 and C300, see Fig. 3) confirmed the presence of an η2-propene ligand that also engages in a γ-agostic interaction with the metal center (Rh⋯C3 = 2.40 Å; Rh⋯H3 = 1.90 Å; C3–H3 = 1.17 Å; see Fig. 4a for the labelling scheme used in the computational studies). The agostic interaction lies in the {P1RhP2} plane (the {P1RhP2}/{RhC3H3} interplane angle = 7.3°) whereas the C1C2 double bond is rotated by 52.3°. The extended solid-state structure is also well reproduced (see ESI† for an overlay of experimental and computed structures). The energy of this η2-propene cation within the extended lattice, I, was computed to lie 3.4 kcal mol−1 below an alternative η3-allyl hydride cation, II; however, the latter is computed to be kinetically accessible in the solid-state (see below). Addition of a second propene molecule to I to form a bis-η2-propene adduct in the solid-state was computed to be endergonic by 4.7 kcal mol−1. In contrast, in solution, molecular calculations indicate the formation of [Rh(Cy2PCH2CH2PCy2)(η2-C3H6)2]+ from free propene and [Rh(Cy2PCH2CH2PCy2)(η2-C3H6)]+ is exergonic by 9.5 kcal mol−1, and this bis-propene adduct is accessible experimentally in solution (see above).
Fig. 4b provides calculated 13C and 1H chemical shifts associated with the [1-(propene)]+ cation in the solid-state, based on GIPAW calculations on the extended [1-(propene)][BArF4] structure. Excellent agreement is found with the experimental low temperature 13C SSNMR data for the propene ligand, providing further support for the formulation of an η2-propene/γ-agostic complex. The calculations assign a high-field 1H resonance of δ −3.9 to the agostic proton in the 1H NMR spectrum of the static structure, while the average chemical shift computed for all three methyl protons is δ −1.0 that reflects a dynamic CH3 group. This is to the high field of the observed value of δ −0.02 in solution and may reflect environment effects in the solid-state. Thus when the model used in the calculation is changed to the isolated cation an average value of δ −0.1 is computed, with the geminal protons in particular shifting to lower field (δcalc(1H) +2.3, +0.9). In contrast, the agostic proton is less sensitive to the model employed, shifting by only 0.4 ppm to δcalc −3.5 in the isolated cation. Local ring current effects arising from proximal aryl groups of the [BArF4]− anion have previously been shown to be significant for [1-NBA][BArF4] in the solid-state.55
One possible mechanism for the fluxional process observed experimentally at room temperature in solution and the solid-state is a 1,3-hydrogen shift involving C–H activation of the bound propene in I to give the allyl-hydride II (Scheme 7), followed by reinsertion, either degenerate or onto the distal carbon atom. A similar process has been suggested for the double bond shift in Ir–pincer systems such as Ir(POCOPtBu)(η2-propene) [POCOPtBu = κ3-C6H3-2,6-(OPtBu2)2].22 The 1,3-hydrogen shift would result in a formal double bond isomerization in propene, but proceeds with no overall chemical change to the complex. If this was happening rapidly94 at room temperature, and such an equilibrium favored the propene tautomer, then a hydride signal would likely not be observed in the 1H NMR spectrum. In contrast, as the proton associated with the central carbon (C2 in Fig. 4) does not undergo rapid exchange, it should be observed, and we propose that this corresponds to the signal at δ 5.07 in the 298 K solution 1H NMR spectrum.
To probe this fluxional process further, solid/gas catalysis using [1-NBA][BArF4] and 3,3,3-d3-propene was performed using ∼6 equivalents of alkene at 298 K and the headspace gas interrogated using 2H NMR spectroscopy (Scheme 8). After 5 minutes deuterium was now observed in both the C1 alkene (cis and trans positions relative to the methyl) and the methyl positions, and after 1 hour D-incorporation approached that expected for a statistical distribution at the C1 and C3 positions [0.39:0.57 ratio]. A very small amount of D-incorporation into the C2 methine position (4%) was also measured at this time. After 16 hours all positions were deuterated to a level close to that predicted from a simple statistical distribution between all three positions. The rapid D-scrambling at the C1 and C3 positions is fully consistent with a mechanism for fluxionality that invokes facile C–H activation via an allyl-intermediate.24 This rapid catalytic solid/gas H/D scrambling in 3,3,3-d3-propene using [1-NBA][BArF4] can be compared to that measured in solution phase under stoichiometric conditions for Ir(POCOPtBu)(η2-d3-propene) that requires heating (∼40 h at 333 K),22 or the slow (greater than 16 h) solid/gas reactivity of [(Ph3P)3IrH2]3[PW12O40] with 3,3,3-d3-propene that results in intramolecular scrambling in the final allyl-hydride product.49 Interestingly, that this solid/gas catalysis is much faster (∼5 minutes) compared with bulk-scale synthesis of [1-(propene)][BArF4] (2 h) suggests that the most active sites are at, or near, the surface. We have previously drawn similar conclusions regarding the use of [Rh(iBu2PCH2CH2PiBu2)(η2η2-C4H6)][BArF4] as a solid-state ethene hydrogenation catalyst.63 As we discuss later, these observations are consistent with the relative rates of 1-butene isomerization by [1-NBA][BArF4] for different-sized crystalline samples.
Further evidence for both the agostic interaction and an exchange process occurring in [1-(propene)][BArF4] comes from interrogation of a number of samples prepared using 3,3,3-d3-propene after 16 hours, in which the D-label would be expected to be in all three C-positions (i.e.Scheme 8).103 The corresponding γ-agostic signal in the 193 K solution 1H NMR spectrum integrates to a total of 1.5 protons, as expected for the statistical distribution of deuterium, and comes from an ensemble combination of CH3 and CDH2 and CD2H groups (A, B and C, Scheme 9). A significant isotopic perturbation of equilibrium (IPE) would be thus expected to be observed in the 1H NMR spectrum for the three CH3, CH2D and CHD2 isotopomers, as, due to zero-point energy differences between C–H and C–D, agostic Rh⋯H–C interactions are favored.104
However four signals are observed at δ 0.08, −0.11, −0.28 and −0.58 in the 500 MHz 1H NMR spectrum at 193 K.105 We suggest that the extra signal comes from the diastereomeric pair in the CH2D isotopomer that arises from the relative orientation of the C–D, agostic and alkene bonds so that the two hydrogen atoms cannot become equivalent by a simple rotation (Ha and Hb in structure B). These data fully support the presence of an agostic interaction in [1-(propene)][BArF4]. Chemical shift calculations on the isolated cation of these isotopomers, taking into account the respective Boltzmann weighting factors recreate the observed relative chemical shifts well (δ −0.1, −0.24, −0.41 and −0.62, ESI†). The lowest-field signal for the agostic C–H, experimentally observed at δ 0.08, is assigned to isotopomer A and would be expected to have a very similar chemical shift to that observed in per-protio [1-(propene)][BArF4], δ −0.02, assuming any intrinsic chemical shift change is small.106 We speculate that this difference in chemical shift may be due to a small, but significant, equilibrium concentration of the (close in energy) allyl-hydride being present on isotopic substitution at low temperature, that is not observed in the 1H NMR spectrum at low temperature due to a combination of low abundance and broad signals. Three signals are observed for the alkene protons, that are slightly shifted from the per-protio complex: δ 4.87, 4.52 and 3.54 each integrating to 0.5 H. The isotopomers are not resolved in these signals.
Periodic DFT calculations have also been used to explore the fluxionality and related H/D exchange processes associated with the propene ligand in the solid-state (Scheme 10). Starting from cation I (0.0 kcal mol−1), oxidative cleavage of the agostic C3–H bond proceeds with a barrier of 9.8 kcal mol−1 to give allyl-hydride II at +3.4 kcal mol−1. C1–H reductive coupling then proceeds via a transition state at +10.9 kcal mol−1 to reform the propene complex as I′ (+1.2 kcal mol−1) in which the alkene and agostic moieties have swapped positions compared to I. This exchange process renders the two phosphorus centers near-equivalent with a modest overall barrier of 10.9 kcal mol−1, consistent with it being readily accessible at room temperature.107 The slightly different energies of I and I′ (and the transition states linking these structures to II) reflect the different orientations of the propene ligand within the crystal lattice.108 In addition to this net 1,3-H shift, rotation of the propene ligand is also readily accessible, with a barrier of 10.4 kcal mol−1 interconverting I and its rotated form Irot, while I′ and I′rot are linked via a transition state at 9.9 kcal mol−1. These rotated forms correspond to the alternative orientation of the propene ligand seen crystallographically (defined by positions C101, C201 and C301, Fig. 3) and their similar energies (in particular I and Irot are within 0.1 kcal mol−1) are consistent with the approximately 50:50 occupation of these two components in the solid-state structure. These calculated barriers to fluxionality compare very well with that derived experimentally (ΔG‡(exp) = 10(1) kcal mol−1). An alternative rearrangement via rotation of the η3-allyl ligand (II to IIrot) involves a transition state at +20.6 kcal mol−1 and so is not competitive.
The γ-agostic interaction observed in the ground state structure of [1-(propene)][BArF4] is directly related to C–H activation transition states calculated for the isomerization of η2-bound alkenes via allyl-hydride intermediates,22 and closely related to those calculated for β-methyl migration from alkyl groups – the microscope reverse of the chain propagation step in olefin polymerization.109
Scheme 11 Synthesis of [1-(butene)][BArF4] and [1-(butadiene)][BArF4] by solid/gas and solution routes. |
The single-crystal X-ray structure of [1-(butene)][BArF4] prepared by the solid/gas route shows a {Rh(Cy2PCH2CH2PCy2)}+ moiety on which a cis-2-butene ligand can be successfully modelled (150 K, space group C2/c, Z = 4, Fig. 5). Unfortunately, the alkene ligand bound with the metal is disordered over two sites (crystallographically imposed) which when coupled with the loss in high-angle data on the solid/gas transformation means that bond metrics have an associated significant error, and the hydrogen atoms associated with the butene fragment were not located. Nevertheless the structure is clear, and very closely related to that of [1-(propene)][BArF4]. More structural detail is provided by periodic DFT calculations on [1-(butene)][BArF4] that provide firm evidence for an η2-binding mode supported by a γ-agostic interaction from one methyl (see Fig. 6a). Although butene is introduced as the 1-isomer it is 2-butene that is predominately bound to the metal center, in its cis-form. This is verified by vacuum transfer of CD3CN onto [1-(butene)][BArF4] to form the CD3CN adduct, [Rh(Cy2PCH2CH2PCy2)(CD3CN)2][BArF4] [1-(CD3CN)2][BArF4] that is itself a poor isomerization catalyst, and free butene, followed by a further vacuum transfer of the condensable volatiles. Analysis by 1H NMR spectroscopy showed cis-2-butene to be dominant: δ 1.59 [CDCl3, d, J(HH) = 4.9 Hz, CH3].110 As we show next, this is fully consistent with the low temperature solution and SSNMR spectra, and DFT calculations that show 2-butene to be bound as the cis-isomer.
Fig. 6 Computational characterization of [1-(butene)][BArF4] in the solid-state: (a) structure of the molecular cation, III, with selected distances in Å (truncated Cy groups; input geometry based on the C1–C4 positions in the X-ray analysis, see Fig. 5); (b) computed 13C and 1H data for [1-(butene)][BArF4] (cation only) with selected 13C experimental data (SSNMR, 158 K) in parenthesis. |
The 31P{1H} SSNMR spectrum of [1-(butene)][BArF4] at 298 K shows two closely separated environments at δ 98.4 and 95.1, as well as a small amount of [1-(butadiene)][BArF4]δ 81.0. In the 13C{1H} SSNMR spectrum a single broad environment is observed at δ 91.8 in the region associated with the alkene ligand. A broad signal at δ 6.3 is also observed, which may point to an agostic interaction. Cooling to 158 K resolved the 31P{1H} SSNMR spectrum into two clear environments [δ 100.0, 93.3]; while in the 13C{1H} SSNMR spectrum two signals are now observed in the alkene region [δ 92.1, 89.3] alongside a high field signal at δ 3.4. Another relatively high field signal at δ 14.3 is also present. These data are consistent with the 150 K single-crystal X-ray structure (Fig. 5) and the structure computed in the solid-state by DFT [Fig. 6a], and point to a fluxional process in the solid-state at 298 K, that is slowed at lower temperatures, while retaining one agostic Rh⋯H3C interaction (i.e. δ 3.4) and one non-agostic methyl (i.e. δ 14.3). Computed NMR data for [1-(butene)][BArF4] featuring a cis-2-butene ligand also correspond well to this being the ground state structure [Fig. 6b]. We were not successful in obtaining a meaningful 1H/13C HETCOR spectrum, as for [1-(propene)][BArF4].
Dissolving [1-(butene)][BArF4] (prepared by solid/gas route) in CD2Cl2 at 193 K, allows more details to be revealed of the structure of this complex, which also point towards cis-2-butene being bound. The 31P{1H} NMR spectrum at this temperature shows two clearly resolved doublets of doublets: δ 97.5 [J(RhP) = 211, J(PP) 24 Hz] and δ 89.9 [J(RhP) = 159, J(PP) 24 Hz]. These, as for [1-(propene)][BArF4], indicate a weakly bound ligand trans to one phosphorus environment – likely the agostic interaction observed in the solid-state structure. The 13C{1H} NMR solution spectrum shows a single environment in the alkene region, δ 90.2, and a single high-field signal, δ 10.9, assigned to the methyl groups. Both these signals are at approximately the frequency average of the corresponding signals in the 158 K 13C{1H} SSNMR spectrum, which we suggest reflects the low-temperature limiting structure. The 1H NMR spectrum displays a single alkene environment (2 H relative integral) at δ 5.08, and an integral 6 H high field signal at δ 0.56. A 31P/1H HMBC experiment shows that this high field signal correlates strongly with the 31P environment that shows the large coupling with 103Rh; and DEPT experiments indicate it to be a CH3 group. These data suggest time-averaged Cs symmetry at 193 K in solution. On warming rapid decomposition starts that eventually forms [1-(butadiene)][BArF4] in ca. 50% yield alongside other uncharacterized products,111 that mean we have not been able to study this process at higher temperatures in solution.
DFT calculations have explored the behavior of the [1-(butene)]+ cation in the solid-state (see Scheme 12). Starting from the cis-2-butene isomer (III) oxidative cleavage of the agostic C1–H bond accesses an allyl-hydride species (IV) at +8.3 kcal mol−1via a transition state at 14.5 kcal mol−1. Both values are ca. 5 kcal mol−1 higher than the equivalent process with the propene analogue. Reductive coupling with the distal C3 carbon is no longer a near-degenerate process, but rather forms the 1-butene isomer (V) at +5.7 kcal mol−1. An adduct such as V is presumably initially formed in the reaction of [1-NBA][BArF4] with 1-butene, however, the calculations suggest this would readily isomerize to the more stable cis-2-butene form with an overall barrier of only 8.8 kcal mol−1. The calculations also indicate that V should be kinetically accessible at room temperature. Assessment of the energy of the [1-(trans-2-butene)]+ cation within the solid-state lattice indicates it would lie 11.4 kcal mol−1 above III. This large energy difference again reflects the environment imposed by the solid-state lattice, as calculations on the isolated cations indicate they lie within 0.3 kcal mol−1 of each other. The calculations also define a libration of the cis-2-butene ligand in III that serves to interchange the source of the agostic interaction trans to P2, from the C1–H1 bond in III to the C4–H4 bond in III′ (see Fig. 7). This process occurs with a computed barrier of 3.0 kcal mol−1 and would account for the fluxionality observed in the SSNMR spectra, and the 193 K solution NMR spectra. Further rotation of the butene moiety produces a structure equivalent to the second component in the X-ray structure (i.e. based on positions C1′–C4′, Fig. 5). This second form has a computed energy of −0.4 kcal mol−1 and is accessible via an overall barrier of 22.7 kcal mol−1.
These combined experimental and computational data suggest a low temperature limiting structure for [1-(butene)][BArF4] that has 2-butene bound in the cis-form with a supporting agostic interaction from the methyl group (Scheme 12). In solution at 193 K a low energy libration of the 2-butene ligand provides time-averaged Cs symmetry by exchanging the agostic methyl groups (i.e. C1 and C4, Fig. 7). This is slowed in the solid-state at 158 K. On warming in both solution and the solid-state there is evidence for further fluxional processes occurring. While the NMR data do not allow us to discriminate between a simple full rotation of the alkene fragment or a reversible C–H activation to give an allyl-hydride, the calculations suggest that the latter is more accessible with a barrier of 14.5 kcal mol−1 (Scheme 12) compared to 22.5 kcal mol−1 for C2 rotation.
Further evidence for the isomerization process in Scheme 12 being accessible in the solid-state comes from addition of D2 to [1-(butene)][BArF4], which is shown to have cis-2-butene bound, but forms 1,2-d2-butane as the condensable volatile product: the product of D2 addition to 1-butene (Scheme 13). This suggests that isomerization from 2-butene to 1-butene is fast (i.e.Scheme 12) and that hydrogenation of the terminal alkene is significantly faster than the internal, a well-known observation for cationic Rh-based catalysts in solution.112
Fig. 9 shows a time/conversion behavior for these three catalyst systems. [1-(Ethene)2][BArF4]-Hex is by far the fastest catalyst, the system essentially reaching equilibrium (∼97% conversion) after 6 minutes. 90% conversion is reached after 2.3 min, TOF(90%) = 1160 h−1[1-NBA][BArF4] and [1-(ethene)2][BArF4]-Oct, are slower, TOF(90%) = 29, 20 h−1 respectively taking 1.5 and 2 hours to reach 90% conversion. This demonstrates a significant structure/activity relationship, with the porous [1-(ethene)2][BArF4]-Hex operating as a much faster catalyst than its non-porous polymorph. To probe the influence of surface area finely crushed samples were prepared for which the surface area would be expected to be significantly greater.116 All of these crushed samples were significantly faster than for the larger crystalline samples, e.g. TOF(95%) = 3100 h−1 for [1-(ethene)2][BArF4]-Hex (ESI†). The effects of the porous structure are not evident with the finely crushed samples, and for practical purposes the three catalysts operate with the same efficiency.
The SMOM-Cat can all be recycled, and Fig. 10 shows time/conversion plots for 3 recharge events, when fresh 1-butene is added immediately after greater than 90% conversion has been achieved by brief exposure to vacuum (30 s, 10−3 mbar) and refilling. For [1-(ethene)2][BArF4]-Oct and [1-NBA][BArF4] very similar overall temporal profiles were observed compared to the first addition of 1-butene (cf.Fig. 9). For [1-(ethene)2][BArF4]-Hex some activity is lost, so that TOF is reduced to ∼450 h−1. We suggest this is due to the partial collapse of the lattice under vacuum during the recycling protocol. For both [1-(ethene)2][BArF4]-Oct and [1-(ethene)2][BArF4]-Hex ten charging cycles have been performed for 1-butene isomerization, with no appreciable drop in conversion between the first and last recharges. If samples are aged for 48 hours in the solid-state under 1-butene, conversion to [1-(butadiene)][BArF4] occurs, as described in Section 2.1. This results in a significantly attenuated catalytic activity and only very slow conversion is subsequently obtained (90% conversion, 5.5 hours, TOF = 8 h−1), Fig. 10.
Fig. 10 Comparison of recycling of the SMOM-cat in the isomerization of 1-butene to 2-butene as measured by gas phase NMR spectroscopy. Conditions as Fig. 9. |
We have previously shown that addition of CO(g) to crystalline samples of [Rh(iBu2PCH2CH2PiBu2)(η2η2-C4H6)][BArF4] is slow enough (days) to form a catalytically inactive, passivated, layer of [Rh(iBu2PCH2CH2PiBu2)(CO)2][BArF4] after a few hours in the resulting crystalline material.63 This allows for the activity of surface sites to be probed in the hydrogenation of ethene, which were suggested to be considerably more active compared to the bulk. This approach was inspired by the work of Brookhart on single-crystal solid/gas catalysis using Ir(POCOP)(C2H4).51 For [1-NBA][BArF4] reaction with CO (1 atm) is much faster, forming [(Cy2PCH2CH2PCy2)Rh(CO)2][BArF4] in 25% conversion after only 10 seconds as measured by 31P{1H} spectroscopy of the dissolved solid. At the same time considerable cracking of the crystals also occurred as indicated by optical microscopy, that likely exposes the interior of the crystals (see ESI† for full details).117 This means that passivation of just the surface sites is problematic and we have not pursued this approach further with these samples. However, that bulk crystalline samples show a significantly lower TOF compared to more finely-divided crushed samples, and that porous [1-(ethene)2][BArF4]-Hex is particularly active, suggests that the most active catalyst sites sit at, or near, the surface or an open pore. On the basis of the synthetic studies (Section 2.4) we propose that [1-(butene)][BArF4] is likely the resting state during catalysis.
Although catalyzed double bond isomerizations in alkenes are common, those involving 1-butene and well-defined transition metal catalysts are less well represented. A notable homogenous example is Ni(η6-C6H5CH3)(SiCl3)2 that rapidly isomerizes 1-butene to 2-butene at 0 °C in bromobenzene, at loadings as low as 0.1 mol% (TOF ∼8600 h−1).13 Although the long term stability and recyclability was not commented upon, in other solvents significant decomposition was noted. Other homogenous systems are known,14,114,118 as are heterogeneous systems that operate at room temperature.119,120 However, we believe that catalysts such as [1-(ethene)2][BArF4]-Hex are the first well-defined molecular systems that operate at 298 K under, industrially appealing, solid/gas conditions. In addition, they offer fine control of the spatial environment in the solid-state (i.e. show structure/activity relationships), show TOF(min) that are competitive with the fast homogenous systems, and, moreover are recyclable. Although the solid/gas catalysts Ir(PCPiPr)(C2H4)45,46 or [Rh(PPh3)2(CO)]3[PW12O40]47–49 promote the isomerization of alkenes, higher temperatures are reported for the former (125 °C) while the latter is ill-defined on the molecular level and also not particularly active. A MOF-supported {Ir(ethene)2} fragment has been reported to dimerize ethene to butene, for which a mixture of isomers is reported but no further details were given regarding the isomerization process.82
Scheme 14 Transfer dehydrogenation of butene using sacrificial ethene. Conditions: total gas pressure butane 1 atm. |
Footnote |
† Electronic supplementary information (ESI) available: Full details of experimental details, spectroscopic and other analytical data, X-ray crystallography, catalytic conditions, and computational studies. CCDC 1539832–1539836. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/c7sc01491k |
This journal is © The Royal Society of Chemistry 2017 |