Laurie J.
Donnelly
,
Jian-Bin
Lin
,
Benjamin S.
Gelfand
,
Chia Yun
Chang
and
Warren E.
Piers
*
Department of Chemistry, University of Calgary, 2500 University Drive NW, Calgary, Alberta T2N 1N4, Canada. E-mail: wpiers@ucalgary.ca
First published on 26th December 2023
Two rigid β-elimination immune PCcarbeneP pincer ligands, differing in their electron donor properties by variation of the substitution pattern on the aromatic linker arms, were complexed to ruthenium to form the dichlorides LRRuCl2 (R = H or NMe2). These compounds were converted to hydrido chlorides by treatment with dihydrogen (H2) and a base. By converting to tert-butoxide derivatives in situ under an atmosphere of H2, the poly hydride PCalkylP complexes LHRRu(H)3 compounds were generated. In these complexes, H2 has added across the RuC bond in the PCcarbeneP starting materials. The polyhydrides are dynamic in solution and extensive NMR studies helped to elucidate the speciation and fluxional processes operative in this dynamic system. The polyhydride complexes react rapidly with CO2 to give the PCcarbeneP formato hydride complexes LRRu(H)-κ2-O2CH. For R = H, the 1,2-hydride shift from the anchoring alkyl of the PCalkylP carbon to the metal is reversible, but for R = NMe2 it is irreversible. The CO2 incorporated into the formato ligand of these compounds exchanges with free CO2via a bimolecular mechanism that is more rapid for R = NMe2 than for R = H; plausible explanations for this observation are proffered. Experiments designed to evaluate the efficacy of the R = NMe2 formato hydride complex as a catalyst precursor for CO2 hydrogenation to formate salts reveal poor performance in comparison to state-of-the-art ruthenium-based catalysts. This is due primarily to the precipitation of a dimeric μ-κ2-κ1-CO3 carbonate complex that is not an active catalyst for the reaction.
One specific family of PCcarbeneP ligands, which were first reported by Shaw16 (I, Chart 1) offer a unique form of MLC via reversible addition of H2 across the anchoring CM moiety of the pincer ligand framework. Because of the β-hydrogens present, the saturated hydrocarbon linker in these seminal ligands posed stability problems for polyhydride derivatives but in 2012,17 our group reported a “β-hydrogen elimination immune” variant that incorporated aryl groups as linkers (II) that allowed for generation of dynamic polyhydrides of rhodium18 and iridium.13,19–23 Activation of other E–H bonds in small molecules across this linkage is also possible, especially in nickel complexes.24,25 This family of PCcarbeneP ligands has also been studied by other research groups as summarized by a recent review article from Manzano and Young.26
Against this backdrop, we were interested in exploring the use of our PCcarbeneP ligands in ruthenium polyhydrides for use in catalytic small molecule conversions. Previously, Gusev had examined use of the saturated Shaw type ligand I for ruthenium complexes and found β-hydride elimination to be problematic in these systems as well.27 Ozerov devised the PCcarbeneP ligand shown in III and used it to develop some ruthenium chemistry,28 touting this ligand as being synthetically more accessible than the aryl-linked ligands II, which at that point had only been assembled within the coordination sphere of ruthenium, in early work by Roper et al.29 Since we have available a small library of ligands II,30 we report here the synthesis and characterization of some (PCcarbeneP)Ru(H)n polyhydrides and their reactivity towards CO2.
When the dichlorides LRRuCl2 were treated with an atmosphere of H2 in the presence of triethylamine at 70 °C (R = H) or room temperature (R = NMe2), smooth conversion to the analogous monohydrido chlorides LRRu(H)Cl was observed. These hydrido chlorides can also be prepared directly from the ligands and [Ru(p-cymene)Cl2]2 in one pot if triethylamine is included, albeit with some contamination (≈10%) by the dichlorides but, this can be driven forward by introducing additional dihydrogen. Conversion stops at the monohydrido derivatives, which are evidenced by upfield triplets at −11.75 ppm (R = H) and -14.57 ppm (R = NMe2) in the 1H NMR spectra. Both LRRu(H)Cl derivatives were also characterized by X-ray crystallography (Fig. S1†).
Removal of the second chloride ligand for conversion to a polyhydride species required treatment of the LRRu(H)Cl derivatives with NaOtBu under an atmosphere of dihydrogen (Scheme 1), but this smoothly afforded solutions of the PCalkylP ligated polyhydride species LHRRu(H)n. 31P{1H} NMR spectra indicate relatively clean crude reaction mixtures, but upon work up by exposing to vacuum, a pale yellow to dark green color change was observed, and assay of the washed solid product by 31P NMR spectroscopy seemed to indicate the presence of several phosphorus containing products (Fig. S2 and S3†). Remarkably, simply reintroducing one atmosphere of H2 reconstituted the polyhydride product LHRRu(H)n as the sole phosphorus containing species in solution.
The solution speciation of these compounds was probed by NMR spectroscopy; all spectra were obtained under 1–4 atmospheres of dihydrogen. In compounds LHRRu(H)n, hydrogenation of the CRu bond in the PCcarbeneP ligand has occurred, producing a PCalkylP donor, and the dibenzylic hydrogen appears in the 1H NMR spectra for LHRRu(H)n at 4.99 ppm (R = H) and 5.00 ppm (R = NMe2), respectively. The 13C{1H} spectra also support this conclusion as there are no carbene resonances but rather peaks for the HC–Ru groups at 39.7 ppm and 67.8 ppm, as confirmed by a 1H–13C HSQC NMR experiment. In the hydride region of the spectra, broad resonances at −7.80 ppm and −7.52 ppm that integrate for a total of about three hydrogens are observed, suggesting that another two equivalents of H2 have been added to the complex in the synthesis described above (one hydrogen atom is incorporated into the tert-butanol byproduct). The broadness of the signal in the hydride region of a room temperature acquired 1H NMR spectrum is suggestive of complex dynamic behavior, and although the benzylic peaks do not coalesce fully with the hydride/dihydrogen resonances, exposure of these compounds to 1 atmosphere of D2 results in rapid deuteration of the benzylic position on the ligand in addition to the hydrogens on the ruthenium. This indicates that the benzylic hydrogen is in exchange with the metal hydride/dihydrogen ligands on the chemical timescale, akin to what has been observed for iridium complexes with these types of ligands.13,17
The variable temperature NMR spectra for the two compounds are similar in pattern, and here we focus here on the data for the LHNMe2Ru(H)3 derivative. Samples generated as described above and cooled from room temperature to 198 K leads to a spectrum with three broad signals in the hydride region, in addition to another minor species with a hydride multiplet at about −10.5 ppm (Fig. S4†). The relative integrations of the three major peaks are about 1:
2
:
1 and their broadness and peak shape suggest there are further dynamic processes that cannot be frozen out at this limiting temperature. The benzylic region of the spectrum also shows multiple resonances, one of which is associated with the minor hydride resonance at −10.5 ppm, by EXSY spectroscopy (Fig. S5†). In the 31P NMR spectrum acquired at 198 K, two resonances at 97.1 and 90.0 ppm appear in a 6
:
1 ratio (Fig. S6†).
We interpret this (limited) spectral information as depicted in Scheme 2. The data suggest that the polyhydride compounds are best formulated as (PCalkylP)Ru(II) hydrido/η2-dihydrogen complexes which may exist as two isomers (present in an approximately 6:
1 ratio) delineated by the disposition of the terminal Ru hydride relative to the benzylic hydrogen across the Ru–C bond of the PCalkylP ligand.18T1 values at 298 K for the broad hydride resonances of LHRRu(H)3 were determined using inversion recovery experiments, giving values of 128 ms (R = H) and 61 ms (R = NMe2). We were unable to determine T1(min) values but the values obtained at room temperature are low for a purely classical hydride species (usually >350 ms) providing indirect evidence for some dihydrogen character in these systems.33,34 The hydrogen atoms in these isomers are in dynamic exchange, likely involving intermediates LH2RRu(H)2, LHRRu(H)3, LRRu(H2)2, or LRRu(H)2 as shown in Scheme 2. The nature of the minor species with a hydrido resonance at ≈−10.5 was elucidated by attempts to grow crystals of either trans-LHRRu(H2)H or cis-LHRRu(H2)H at low temperatures. Both isomers were highly soluble in aromatic and hydrocarbon solvents under H2 and could not be coaxed out of solution under any conditions. However, exposure of these solutions to partial vacuum did lead to deposition of X-ray quality crystals, which turned out to be the dimeric species LHHRu(H)(μ-H)2(H)RuLHH depicted in Scheme 2 and shown in Fig. 1. We propose that this compound forms upon loss of H2 from cis/trans-LHRRu(H2)H and subsequent dimerization of the resulting unsaturated LHRRuH species with (formally) intermediate LHRRu(H)3. Hydride locations were assigned based on residual electron density in open coordination sites suggested by the geometry of the ligand. For example, the P(1)–Ru(1)–P(2) angle of 119.12(6)° is consistent with a PCalkylP ligand and accommodates a coordination site for a terminal hydrido ligand. The Ru1–Ru2 distance of 2.5714(7) Å is consistent with other hydride-bridged ruthenium dimers.35,36 The 1H NMR spectra of these crystals exhibited a doublet of triplets at −10.45 ppm and an H–C–Ru proton at 5.20 ppm, confirming the nature of this species (which appears in varying amounts from sample to sample) in the spectra of cis/trans-LHRRu(H2)H. Finally, it should be noted that exposure to 4 atm of H2 gas results in loss of these signals as this dimer is converted back to cis/trans-LHRRu(H2)H.
Further support for the speciation map proposed in Scheme 2 arises from attempts to grow crystals of the green colored crude product mixture formed upon complete removal of H2 atmosphere and solvent from cis/trans-LHNMe2Ru(H2)H. Here, long term recrystallizations lead to crystals that turned out to be the dinitrogen bridged dimer LHNMe2Ru(H)(μ-N2)(H)RuLHNMe2 (see Fig. 2 for its molecular structure). This species formed in an argon filled box, picking up adventitious N2 over the course of several days and underscoring the high affinity of the LHNMe2RuH fragment for dinitrogen. This compound was confirmed by its intentional in situ synthesis by exposing cis/trans-LHNMe2Ru(H2)H to an atmosphere of dry N2. This dinitrogen complex has a characteristic signal in the 31P NMR spectrum at 74.6 ppm. The rather short N(3)–N(3′) distance of 1.0892(95) Å indicates that the dinitrogen is relatively unactivated and essentially retains an N–N triple bond. Consistent with this, it is easily displaced by dihydrogen to return to clean samples of cis/trans-LHNMe2Ru(H2)H.
Given that ruthenium polyhydride complexes are among the most active molecular catalysts for CO2 hydrogenation,4 we evaluated the reactivity of PCalkylP hydrides LHRRuH with CO2 (Scheme 3). In the products LRRu(H)-κ2-O2CH, the PCcarbeneP ligand is restored, as indicated by the absence of benzylic C–H groups in the 1H NMR spectra, and the appearance of resonances at 275.6 (R = H) and 274.5 (R = NMe2) ppm for the carbene carbons in the 13C NMR spectra. In addition, the formate carbons resonate at 170.9 and 170.4 ppm, and triplets for the terminal hydrido ligands are seen at −8.99 and −11.14 ppm. The nature of these products was further confirmed by their separate synthesis via treatment of the hydrido chlorides LRRu(H)Cl with sodium formate as depicted in Scheme 3. The compound LNMe2Ru(H)-κ2-O2CH was also characterized by X-ray crystallography (Fig. 3) and the more planar tricyclic ligand backbone along with a short Ru(1)–C(13) distance of 1.890(9) Å confirms the presence of a PCcarbeneP ligand in these complexes.
We presume that formation of these formato hydrides occurs via CO2 insertion into the terminal hydride of the complexes LHRRuH, followed by a 1,2-hydride shift from the anchoring alkyl group to the metal. Indeed, for R = H, the presumed PCalkylP Ru(II) formato intermediate is observed spectroscopically (C–H benzylic proton at 5.55 ppm, 31P resonance at 65.8 ppm); the rate limiting 1,2-hydride shift to produce LHRu(H)-κ2-O2CH, occurs over the course of about 30 minutes. For R = NMe2, however, the 1,2-shift is extremely rapid, and the PCalkylP intermediate could not be detected. We hypothesize that for this more electron rich ligand, the step is driven by the restoration of conjugation of the π-electrons of the donating dimethyl amino groups with the carbene moiety that forms as the 1,2-shift proceeds. This is further evidenced by the observation that dissolution of these two complexes in acetonitrile indicates that, for R = H, the ruthenium hydride shifts back to the carbene carbon (5.70 ppm in the 1H NMR spectrum, Fig. S7†) and forms an acetonitrile adduct (Scheme 3). For R = NMe2, however, no reaction with acetonitrile is observed, indicating that for this species the 1,2-hydride shift is essentially irreversible. This is supported by the observation that when the headspace of NMR samples of these compounds is removed via freeze–pump–thaw routines, LNMe2Ru(H)-κ2-O2CH is indefinitely stable, while LHRu(H)-κ2-O2CH loses CO2 over the course of about an hour, generating solutions of “LHHRuH” that convert quantitatively to LHHRu(H2)H upon exposure to dihydrogen. Thus, the ability to open a coordination site via this transfer of a Ru–H to the carbene carbon renders the CO2 insertion reversible in the less electron donating R = H ligand system, as is observed in other ruthenium-based CO2 hydrogenation catalysts.37
We also observed exchange of inserted CO2 in the formato ligands of LRRu(H)-κ2-O2CH with free CO2. Interestingly, the rate of this exchange of free CO2 with the inserted carbon dioxide follows the opposite trend in that the R = NMe2 system exchanges CO2 faster than the parent R = H system. This was probed by preparing 13C labeled derivatives (using 13CO2via the method shown in Scheme 3) and exposing solutions of the labelled LRRu(H)-κ2-O213CH materials to various partial pressures of unlabelled 12CO2. An observed pseudo first order rate constant (kobs) could be obtained by monitoring the loss of 13C label in the formato ligand via1H NMR spectroscopy. As shown in Fig. 4, the rate of label loss is about 3–4 times faster for the R = NMe2 derivative and increases for both compounds as the partial pressure of 12CO2 rises, tending towards saturation as the solution concentration of CO2 in benzene plateaus. We interpret these results as being consistent with either an “inner sphere” or an “outer sphere” mechanism38 as depicted in Scheme 4. The inner sphere process assumes that CO2 insertion into the remaining hydride requires an open coordination site and slippage of the κ2 formato to a κ1 variant would be required to form LRRu(H)-κ1-O213CH. The equilibrium concentration of this intermediate should be higher for R = NMe2 than for R = H, since the more electron donating former ligand should stabilize the unsaturated κ1 species, increasing the rate of bimolecular CO2 insertion to form the intermediate I (Scheme 4, bottom) from which exchange of label occurs. Loss of CO2 from this species is rapid and can be either labeled or unlabeled, but due to the excess of 12CO2 in the system, eventually the 13CO2 label is washed out of the compounds. In addition to enabling higher concentrations of the key LNMe2Ru(H)-κ1-O213CH intermediate, the ruthenium hydride is likely more hydridic in the R = NMe2 system possibly leading to further rate enhancements in the bimolecular step for this more electron rich ligand. An outer sphere process39,40 (Scheme 4, top) wherein direct bimolecular hydride transfer from formato hydrides LRRu(H)-κ2-O213CH to 12CO2 occurs cannot be ruled out but would also favor the more hydridic R = NMe2 derivative.
Given its observed facile reactivity with CO2, we evaluated the efficacy of polyhydride species LHNMe2Ru(H)3 as a carbon dioxide hydrogenation catalyst. Unsurprisingly, the formato hydride LNMe2Ru(H)-κ2-O2CH does not react with dihydrogen, and a basic additive was employed to generate a formate salt, providing a favorable thermodynamic impetus for reduction of CO2.4,5,37,41,42 Optimization of catalytic conditions led us to carry out catalytic reactions with 0.64% catalyst loading in dioxane/water mixtures (2:
1 ratio), at 70 °C using NaOH as the basic additive (K2CO3 or NEt3 did not lead to significant conversion of CO2 to formate). Catalytic reactions were conducted with a 1
:
1 mixture of CO2 and H2 at pressures ranging from 1 to 40 atmospheres. Turnover numbers (TON) were determined via NMR integration of the C–H formate signal in NaO2CH by diluting 0.20 mL aliquots of catalyst solution to 1.0 mL in D2O and adding 5.0 mg (0.035 mmol) of sodium benzoate as an internal standard.3 In all cases it was found that turnovers were much lower (<10) than expected for the catalyst loading employed and in comparison to related catalyst systems (Table S1†). Furthermore, in most catalyst runs, a red precipitate was observed to form under catalytic conditions. Single crystals that were retrieved from these precipitates were subjected to X-ray analysis, which showed them to be the μ-κ2-κ1 carbonato dinuclear compound [LNMe2Ru(H)]2(μ-κ2-κ1-CO3) (see Fig. 5). This species probably forms via insertion of CO2 into ruthenium hydroxo complexes present in the catalytic reaction, and while this doesn't necessarily preclude catalytic turnover if this insertion is reversible,37,43 here the apparent insolubility of this dinuclear species has the effect of suppressing catalytic turnover by removing catalyst from the system.
Footnote |
† Electronic supplementary information (ESI) available: General Experimental details, synthesis and characterization of all new compounds, dynamic NMR studies and crystallographic parameters. CCDC 2309264–2309271. For ESI and crystallographic data in CIF or other electronic format see DOI: https://doi.org/10.1039/d3dt04014c |
This journal is © The Royal Society of Chemistry 2024 |