Mechanistic Insights into Molecular Crystalline Organometallic Heterogeneous Catalysis through Parahydrogen-Based Nuclear Magnetic Resonance Studies

The heterogeneous solid–gas reactions of crystals of [Rh(L2)(propene)][BArF4] (1, L2 = tBu2PCH2CH2PtBu2) with H2 and propene, 1-butene, propyne, or 1-butyne are explored by gas-phase nuclear magnetic resonance (NMR) spectroscopy under batch conditions at 25 °C. The temporal evolution of the resulting parahydrogen-induced polarization (PHIP) effects measures catalytic flux and thus interrogates the efficiency of catalytic pairwise para-H2 transfer, speciation changes in the crystalline catalyst at the molecular level, and allows for high-quality single-scan 1H, 13C NMR gas-phase spectra for the products to be obtained, as well as 2D-measurements. Complex 1 reacts with H2 to form dimeric [Rh(L2)(H)(μ-H)]2[BArF4]2 (4), as probed using EXAFS; meanwhile, a single-crystal of 1 equilibrates NMR silent para-H2 with its NMR active ortho isomer, contemporaneously converting into 4, and 1 and 4 each convert para-H2 into ortho-H2 at different rates. Hydrogenation of propene using 1 and para-H2 results in very high initial polarization levels in propane (>85%). Strong PHIP was also detected in the hydrogenation products of 1-butene, propyne, and 1-butyne. With propyne, a competing cyclotrimerization deactivation process occurs to afford [Rh(tBu2PCH2CH2PtBu2)(1,3,4-Me3C6H3)][BArF4], while with 1-butyne, rapid isomerization of 1-butyne occurs to give a butadiene complex, which then reacts with H2 more slowly to form catalytically active 4. Surprisingly, the high PHIP hydrogenation efficiencies allow hyperpolarization effects to be seen when H2 is taken directly from a regular cylinder at 25 °C. Finally, changing the chelating phosphine to Cy2PCH2CH2PCy2 results in initial high polarization efficiencies for propene hydrogenation, but rapid quenching of the catalyst competes to form the zwitterion [Rh(Cy2PCH2CH2PCy2){η6-(CF3)2(C6H3)}BArF3].


General Procedures
All manipulations, unless otherwise stated, were performed under an inert (argon or nitrogen, BOC, N4.8 purity) or dihydrogen (BOC, N4.0) atmosphere using standard Schlenk line and glovebox (<0.1 ppm H2O/O2) techniques. Glassware was oven-dried at 140 °C overnight and flame dried under vacuum prior to use. All solvents were degassed by three successive freeze-pump-thaw cycles and stored over activated 3 Å molecular sieves under inert gas in resealable glass ampoules fitted with PTFE high vacuum stopcocks (Kontes Hi-Vac, J. Young or Rotaflo HP). CH2Cl2, pentane and hexane were dried using a commercially available solvent system (Innovative Technologies or MBraun) by passage through stainless steel columns containing activated alumina. 1 Heptane was purchased anhydrous from Sigma-Aldrich and decanted by cannula into resealable glass ampoules and stored as above. CD2Cl2 and 1,2-C6H4F2 (pre-dried by stirring over activated alumina) were dried over CaH2, before vacuum transfer and storage as above. All solution phase NMR were prepared on a greaseless high vacuum line (<1 x 10 -5 mbar) by condensation of the solvent under static vacuum onto solid samples in 5 mm thin wall NMR tubes fitted with high vacuum PTFE (J.  12 MHz, 13 C = 100.61 MHz) at 298 K unless otherwise specified. Residual proteo solvent was used as reference for 1 H and 13 C{ 1 H} spectra in deuterated solvent samples. 5 31 P{ 1 H} NMR spectra were externally referenced to 85% H3PO4. 1 H assignments were aided by 1 H{ 31 P} experiments. All chemical shifts (δ) are quoted in ppm and coupling constants (J) in Hz. NMR assignments were aided by 2D experiments ( 1 H-1 H-COSY, 1 H-13 C-HSQC, 1 H-13 C-HMBC) where required.

C{ 1 H}
INEPT spectra are recorded in the gas phase with a single scan with a 0.6 s acquisition time (aq) and 0.01 s recycle delay (d1) with 28842 datapoints (td) over a 238.91 ppm spectroscopic window (sw). 1

H-1 H COSY
were recorded in the gas phase with two scans per increment (ns) over a spectroscopic window of 15.02 ppm (sw F1, sw F2) with 128 datapoints in the indirect dimension (td F1) and 2402 in the direct (td F2) with 0.2 s (aq F2) and 0.01 s (aq F1) acquisition times and 0.01 s recycle delay (d1). 1 H-13 C-HMQC spectra were recorded in the gas phase with two scans per increment (ns) over a spectroscopic window of 120 ppm (sw F1) and 10 ppm (sw F2) with 128 datapoints in the indirect dimension (td F1) and 1198 in the direct (td F2) with 0.15 s (aq F2) and 0.005 s (aq F1) acquisition times and 0.01 s recycle delay (d1). OPSY spectra in the gas phase were recorded using the OPSYdq pulse sequence with a 1 μsec recycle delay (d1),
Post-catalysis samples were prepared in 10 mm OD NMR tubes fitted with coaxial high vacuum PTFE (J. Young) valves. Rotors were packed and sealed with Kel-F caps or zirconia caps for variable temperature experiments in an argon filled glovebox. Spectra are referenced externally to Si(CH3)4 or H3PO4 using the secondary references adamantane ( 13 C δ = 29.5 for the shielded methylene resonance) 7
The red solution was stirred for 1 hour at ambient temperature before the crude product was precipitated by the addition of excess hexane (ca. 300 mL); the suspension was filtered and washed with further hexane (

Speciation and Post-Catalysis Experiments
Catalytic Scale Speciation

General Procedure
A 5 mm thin wall NMR tube fitted with a high vacuum PTFE (J. Young) valve containing the appropriate quantity of sieved catalyst was evacuated and backfilled with the target substrate gas (1.05-1.08 bar absolute) on a specially constructed stainless-steel vacuum/para-hydrogen/substrate gas NMR tube triple manifold as has been described previously. 10,11 The tube was sealed at this pressure and the headspace above the sealed tube evacuated and re-filled (4 bar absolute) with para-hydrogen five times on the stainless steel manifold. The manifold was isolated from the para-hydrogen source after a final refill to a static pressure (4 bar absolute). The valve was then opened at the NMR tube to equilibrate substrate gas and para-hydrogen pressures (final system pressure ~3.5 bar absolute) in the NMR tube, which was then rapidly sealed and vigorously shaken. For experiments requiring repeat cycles, the tube was returned to the gas manifold, evacuated and recharged with the same substrate gas and re-pressurised as above. When required the tube was evacuated on the high vacuum line and CD2Cl2 (ca. 0.6 mL) condensed into the tube and the tube left frozen until immediately before analysis.

Para-Hydrogen with a Single Crystal
Using the general procedure without any substrate gas and a single crystal of  . 31

Gas Phase Monitoring of Equilibration of p-H2 to o-H2
A 5 mm thin wall NMR tube fitted with a high vacuum PTFE (J. Young) valve containing a single crystal of pressurised with para-hydrogen to a total system pressure of 3.5 bar absolute and then agitated vigorously for 5 minutes to ensure gas mixing during catalysis. The vessel was evacuated, backfilled with propene (1 bar absolute), re-pressurised with para-hydrogen (total system pressure 3.5 bar absolute) and then vigorously mixed as previously. This process was repeated until a total of five hydrogenation cycles had been completed, whereupon the vessel was evacuated a final time, backfilled with argon and then transferred to an argon containing glovebox and packed into a solid-state rotor and 5 mm NMR tube for analysis.  Figure S30. 13  pressurised with para-hydrogen to a total system pressure of 3.5 bar absolute and then agitated vigorously for 5 minutes to ensure gas mixing during catalysis. The vessel was evacuated, backfilled with propene (1 bar absolute), re-pressurised with para-hydrogen (total system pressure 3.5 bar absolute) and then vigorously mixed as previously. This process was repeated until a total of five hydrogenation cycles had been completed, whereupon the vessel was evacuated a final time, backfilled with argon and then transferred to an argon containing glovebox and packed into a solid-state rotor for analysis.  Figure S35. 13   S30 Figure S37. 31 Figure S38. 13   Rh-P were assigned their known values from diffraction and an excellent fit was obtained (allowing the variation of the Rh-Rh CN also did not result in a better fit or a value significantly different to the expected CN of 1). In the fit of the sample of 1 that had mediated five cycles of propene hydrogenation, the Rh-Rh CN was permitted to vary, but was still found to be close to 1 (the Rh-P co-ordination was kept fixed to minimize the number of variables in the fit as there are two coordinating P atoms in both possible monomeric species and the dimer). In addition to obtaining a clear fit to the Rh dimer with a near identical bond length to the reference sample, attempts to fit data for this sample with nearby carbon atoms rather than a nearby Rh, as would be present in monomeric species, failed. This is expected given the contribution of the Rh-Rh at 2-3 Å, which cannot be replaced by contributions from Rh-P or Rh-C, which do not extend to these high R values. This can be seen in the plots of individual contribution paths for Rh-Rh and Rh-P that follow. All fits were performed using multiple k-weight fitting, although k 3 -weighted data is shown in figures. Improved fits of the reference metal foil were obtained by fitting to higher values of R using a model accounting for multiple scattering paths (those regarded significant in FEFF) up to 5.5 Å effective path length and using the further physically reasonable assumption that all Rh-Rh interactions grow or shrink by a constant scaling factor, α relative to the known bulk Rh fcc metal crystal. Assumed relations of coordination number and mean-squared relative deviation are tabulated in Table S2. The upper k-limit for the range employed in each sample fit was selected on the basis of where the data became too noisy to identify clear oscillations, although nearby values of k were also checked to ensure the selection did not dramatically change the fit outcome.

Figure S45. Normalized XANES spectra at the Rh K-edge for a sample of 1 that had mediated five cycles of propene hydrogenation and a reference sample of independently synthesized 4. The photon energy scale has been corrected by alignment of concurrently measured Rh foil data.
Scattering path Co-ordination number σ 2  No. of independent points 32.7 No. of fitted parameters 7 Table S3. EXAFS fitting parameters for the Rh K-Edge data on Rh foil, were α is the scaling factor applied to each path length relative to the reported crystallographic bond lengths in bulk fcc Rh. Table S3.

Gas-phase NMR of 1-Butyne to 1,3-Butadiene Isomerisation
A sample of [Rh( t Bu2PCH2CH2P t Bu2)(nbd)][BAr F 4] (8.8 mg, 6.4 μmol) in an NMR tube was hydrogenated (2 bara absolute) for 90 minutes before the tube was evacuated and the placed under an atmosphere of propene (2 bar absolute) for 120 minutes. The headspace was evacuated again and the tube backfilled with 1-butyne (1 bar absolute) and the headspace monitored by gas phase NMR spectroscopy for 30 hours, wherein a slow consumption of gaseous 1-butyne and appearance of signals attributed to 1,3-butadiene was observed.

Solid/Gas Alkyne Cyclotrimerisation General Procedure
[Rh( t Bu2PCH2CH2P t Bu2)(propene)][BAr F 4] 1 (ca. 6 μmol) in a valved NMR tube was hydrogenated (2 bar absolute) for 90 minutes before evacuation and re-pressurisation with propene (2 bar absolute). The sample was stored at room temperature for 120 minutes before the tube was re-evacuated and back-filled

Solution Alkyne Cyclotrimerisation Experimental
[Rh( t Bu2PCH2CH2P t Bu2)(norbornadiene)][BAr F 4] 1 (10 mg, 7.3 μmol) in a valved NMR tube was hydrogenated (2 bar absolute) for 90 minutes before evacuation and re-pressurisation with propene (2 bar absolute). The sample was stored at room temperature for 120 minutes before the tube was re-evacuated 1,2-C6H4F2 (ca. 0.6 mL) was condensed into the tube under vacuum and then back-filled with 1-butyne (1 bar absolute) and the contents monitored by 31 P NMR spectroscopy over 24 hours. F 4] (δ 114.0). packed in a nitrogen glovebox) was evacuated and backfilled with the target substrate gas (1.05-1.08 bar absolute) on a specially constructed stainless-steel vacuum/para-hydrogen/substrate gas NMR tube triple manifold as has been described previously. 10,11 Tubes for experiments involving gas mixtures (1-butene and propene, 1.5 bar absolute) were filled by condensation of the individual gas (0.75 bar absolute) into the tube with the aid of a mercury manometer on a glass high vacuum line. The tube was sealed at this pressure and the headspace above the sealed tube evacuated and re-filled (4 bar absolute) with parahydrogen five times on the stainless steel manifold. The manifold was isolated from the para-hydrogen source after a final refill to a static pressure (4 bar absolute). The valve was then opened at the NMR tube to equilibrate substrate gas and para-hydrogen pressures (final system pressure ~3.5 bar absolute) in the NMR tube, which was then rapidly sealed and vigorously shaken during transit to the NMR spectrometer (distance ~5 m) before insertion (without use of the spectrometer's lift) and immediate acquisition through means of a pre-loaded experiment ('autosuspend' entry before d1 in the Bruker pulse program). The interval between pressurisation and first acquisition is reliably 10 seconds. Subsequent acquisition of data was made through the Bruker AU program 'multizg.' After all the desired data was acquired, the tube was removed from the spectrometer, returned to the gas manifold, evacuated and recharged with the same substrate gas and re-pressurised as above. Data acquisition was completed similarly.

Single Crystal X-Ray Diffraction
Single-crystal X-ray diffraction data were collected (ω-scans) on a Rigaku SuperNova diffractometers with Cu-Kα (λ = 1.54184 Å) radiation equipped with an N2 gas Oxford Cryosystems Cryostream unit. Diffraction images from raw frame data were reduced using CrysAlis Pro. The structures were solved using SHELXT 15 and refined to convergence on F 2 and against all independent reflections by full-matrix least-squares using SHELXL 16 in combination with the Olex2 GUI. 17 All non-hydrogen atoms were refined anisotropically; hydrogen atoms were geometrically placed unless otherwise stated and allowed to ride on their parent atoms. The CF3 groups on the [BAr F 4] anions were disordered and modelled over two domains and restrained to maintain sensible geometries. Distances and angles were calculated using the full covariance matrix.  Table S4. Selected crystallographic data.