Metal-Mediated Catalytic Polarization Transfer from para Hydrogen to 3,5-Dihalogenated Pyridines

The neutral catalysts [IrCl(H)2(NHC)(substrate)2] or [IrCl(H)2(NHC)(substrate)(sulfoxide)] are used to transfer polarization from para hydrogen (pH2) to 3,5-dichloropyridine and 3,5-dibromopyridine substrates. This is achieved in a rapid, reversible, and low-cost process that relies on ligand exchange within the active catalyst. Notably, the sulfoxide-containing catalyst systems produced NMR signal enhancements between 1 and 2 orders of magnitude larger than its unmodified counterpart. Consequently, this signal amplification by reversible exchange hyperpolarization method can boost the 1H, 13C, and 15N nuclear magnetic resonance (NMR) signal intensities by factors up to 4350, 1550, and 46,600, respectively (14.0, 1.3, and 15.4% polarization). In this paper, NMR and X-ray crystallography are used to map the evolution of catalytically important species and provide mechanistic rational for catalytic efficiency. Furthermore, applications in spontaneous radiofrequency amplification by stimulated emission and NMR reaction monitoring are also shown.


Table S2: 1 H NMR signal enhancements and T1 relaxation times for B at various loadings and in two solvents. The NMR signal enhancements are recorded by shaking a sample of [IrCl( 2 - 2 -COD)(IMes)] and A at the indicated loading and in the indicated solvent with 3-bar pH2 for 10 seconds in the fringe field of a 9.4 T magnet (ca 6.5 mT). The relaxation times are measured from hyperpolarised samples (see section S2).
Table S3: 13

S2: Measurement of 1 H T1 times
Immediately after shaking the solutions of the indicated SABRE catalyst with pH2 (3 bar for 10 seconds in the stray field of a 9.4 T magnet), the sample was rapidly inserted into the 9.4 T spectrometer to record a hypeprolarised T1.This involved the collection of a succession of single scan 1 H NMR spectra with a 5° pulse at 298 K that were separated by 7.5 s time intervals up to a 150-220 s time window.For the study investigating the effect of catalyst concentration on the T1 time of A (Figure S1), flip angles of 10.35 o and time spacings of 10 seconds were used, with a total measurement time window of 290 s.
The integral intensities of the 1 H NMR resonances for free A or B, and those bound to the SABRE catalyst (where applicable) were fitted to a model to extract a T1.In this model the hyperpolarised signals of species X, SX detected by the low flip angle pulse at time t is calculated according to Equation 1where MX is the magnetization of species X and is the flip angle.The magnetization of species X remaining after the pulse is given by Equation 2.
The magnetisation of species X changes during the time interval between pulses due to T1 relaxation according to Equation 3. Note that this model does not account for changes in magnetisation of species X due to either chemical reaction (i.e.binding or unbinding to the iridium centre) or due to rehyperpolarisation during the time window of the T1 measurement.
T1 times were calculated by fitting experimentally determined 1 H NMR integral intensities to values calculated using this model.Experimental integral intensities for each species were normalised to one at their highest intensity in the first spectra.T1 times were calculated using Microsoft Excel to give the smallest squared difference between experimental and modelled integral intensities.The SABRE-precatalyst, 1, in methanol-d4 was characterised, in the presence of A, accordingly.The 1 H NMR spectrum of the sample displays four resonances in the alkene region ( 3 -5) as shown in Figure S3.The signals at 3.10 and 4.02 are more intense than the signals at 3.36 and 3.79.They belong to the SABRE precatalyst 1. Integration of all four resonances shows that the 3.10 and 4.02 resonances come from groups in the same complex as they share the same relative integral, and 3.36 and 3.79 signals are attributed to the same, but a different, complex.A 2D NOESY experiment was used to identify signals from groups close in space to the protons in the alkene yielding resonances at 3.10 and 4.02.The signal at 3.10 showed an nOe to a signal at 2.31, which had a relative integral of ca 6, and an nOe interaction with another signal at 7.05, which had an integral of ca 2. The other alkene resonance ( 4.02) did not share any of these nOe interactions.Hence, 2.31 is clearly the equivalent ortho methyl protons of IMes, and investigation of the nOe interactions to the group yielding the 7.05 signal revealed that this signal was due to the meta aromatic protons of IMes, as they showed an nOe to a signal at 2.38, attributed to the para methyl protons of IMes.Therefore, it is clear that one set of COD alkene protons are bound trans to IMes ( 4.02), and the other cis to IMes ( 3.10).A 2D NOESY experiment identified the imidazole protons of the IMes ligand ( 7.25), which shared an nOe with the ortho methyl protons of IMes.This assignment was further confirmed by the singlet multiplicity, and a relative integral of ca 2, for the signal at 7.25.
Attributing the 1 H coupling partners of the COD alkene resonances involved 2D COSY experiments.The 3.10 signal coupled to 1.28 and 1.65, whilst the 4.02 signal coupled to 1.34 and 1.72, which reflects the alkyl resonances of COD.Analysis of the 2D NOESY spectrum for these alkyl signals identified their relative positions since the 1.65 resonance showed an nOe peak to 1.72 and 1.28, but not 1.34, whereas the 1.28 signal showed an nOe to 1.34 and 1.65 and not 1.72.
Signals for the substitution product, 2A were also visible: its alkene resonances at 3.79 and 3.36 are shifted from those of 1 ( 3.10 and 4.02), suggesting the formation of a new complex with COD bound.An nOe interaction was observed between the signal at 3.79 and a doublet resonance with integral ca 2 at 7.87.A signal of this multiplicity and relative integral could only belong to the ortho protons of bound A. The para proton of 3,5-dichloropyridine ( 8.19) was then assigned via its triplet multiplicity and mutual spin-spin coupling to the ortho protons of A. The other alkene signal in this complex ( 3.36) showed an nOe connection to a resonance at 2.20, which had a relative integral of ca 6, and another nOe to a resonance at 7.16, which had a relative integral of ca 2. 2.20 was assigned to the ortho methyl protons of IMes and 7.16 to the aromatic meta proton of IMes.The 2D NOESY data also revealed an nOe interaction between the aromatic meta protons of IMes ( 7.16) and an alkyl resonance with a relative integral of ca 3 at 2.46; this signal clearly belongs to the para methyl protons of IMes.The imidazole proton at 7.40 was identified via a nOe to only the ortho methyl protons at 2.20.COSY experiments revealed the alkyl coupling partners for the alkene signals of COD: 3.79 coupled to both 1.70 and 2.08, whilst 3.36 coupled to both 1.94 and 2.14.The 13 C chemical shifts reported in Table S7 were assigned through analysis of the cross-peaks in HMQC experiments.However, the cross-peaks for the COD alkyl protons overlapped significantly and were not be assigned unequivocally.The structure of 1, and its NMR resonances are shown in Table S7 and Figure S4.Resonances in dichloromethane-d2 at 243 K are also provided.The structure of the substitution product, 2A is shown in Figure S5, and its NMR resonances are given in Table S8.The ratio between 1 and 2 at equilibirum was measured on separate samples when A or B was present as an ~13 fold excess relative to 1 in methanol-d4 at a range of temperatures from 248 K to 303 K.The resulting equilbirium constant, K for these processes was determined by integration of the distinctive COD signals for each (appearing at 4.02 and 3.10 in 1, but moving to 3.36 and 3.79 in 2A) which were linked to their concentration.A Van't Hoff plot from these data (Figure S6) was used to determine H  and S  , from which G  could be calculated at a given temperature.

S3.2: Hydrogen addition to equilibrium mixtures of 1 and 2A
A solution of 1 (5 mM) and 2A (50 mM) was cooled to 243 K and the reaction with pH2 monitored by 1 H NMR spectroscopy.Some example NMR spectra are shown in Figures S7 and S8.

S3.4: Characterisation of 3
The low abundance of 3 necessitated characterisation in CD2Cl2 where higher solubility of 1 is beneficial.The first experiment used to characterise 3 was a 1D NOESY experiment which selectively saturated the hydride signal at -13.43 with a mixing time of 0.8 s.This revealed nOe interactions to a hydride resonance ( -18.04) and resonances in the alkene region ( 4.62 and 3.87) which suggests that COD is still bound.Another 1D NOESY was then used to selectively saturate the other hydride resonance ( -18.04), using the same acquisition parameters as above, which revealed an nOe to three resonances in the alkene region ( 4.62, 3.87 and 4.48) -further supporting a COD bound complex.A COSY spectrum revealed the fourth alkene resonance at 3.66 via coupling to that at 4.48.Each alkene resonance coupled to two different alkyl resonances in a COSY spectrum; used to assign all resonances within the bound COD ligand.The 1D NOESY spectra collected for both hydrides also revealed an nOe interaction to two alkyl signals ( 2.15 and 2.27).These alkyl resonances appear as singlets of relative integral 3 in the 1 H NMR spectrum, and so are attributed to inequivalent ortho methyl proton in IMes.A weaker nOe interaction from these hydrides to 6.97 was also seen, thus assigning the meta aromatic protons.Furthermore, the nOe of both hydrides to the alkene resonance at 4.62 and 3.87 demonstrated that these alkene proton sites are cis to both hydrides, whereas the alkenes sites at 3.66 and 4.48 are trans to the hydride at -13.43, but cis to the hydride at -18.04.

S3.5: Characterisation of 6A
A 1D NOESY experiment was used to selectively saturate the hydride resonance of 6A at -23.87 with a mixing time of 0.3 s.An nOe interaction was observed to the resonance at 8.82, a signal which has a doublet multiplicity and a relative integral of ca 2, suggesting that this resonance belonged to the ortho protons of a bound 3,5-dichloropyridine.A 1D NOESY experiment exciting the other hydride signal ( -24.10) revealed nOe interactions to 8.82 and 8.25, both with doublet multiplicity and a relative integral of ca 2, indicative of the ortho protons of bound 3,5-dichloropyridine.This suggests that both hydrides are cis to a 3,5-dichloropyridine ligand ( 8.82) and the hydride at -24.10 is cis to a second 3,5dichloropyridine ligand, with the hydride ( -23.87) trans to bound A. Both hydrides displayed an nOe to alkyl signals at 2.16 and 2.21, which both had a relative integral of ca 3 and were assigned to the ortho methyl protons of IMes.The relative positions of these signals were determined by the strength of the nOe interaction between the hydride and the methyl resonance.Furthermore, the hydride signals also showed an nOe to a resonance at 6.79, a singlet with a relative integral of ca 2 in a 1 H NMR spectrum, thus this was attributed to the meta aromatic protons of IMes.Now that the meta aromatic protons of IMes have been assigned ( 6.79), the resonance for the imidazole protons can now be assigned.A resonance at 6.82 is a singlet, with a relative integral of ca 2, with a HMQC revealing it directly bound to an alkene carbon.Thus, 6.82 can be assigned to the imidazole proton of IMes.15 N NMR spectra were recorded to further support two bound 3,5-dichloropyridine ligands ( 270.00 and 247.83) in the complex.S11.A series of 1 H NMR spectrea of the hydride ligands of 6A in dichloromethane-d2, recorded at different temperatures are shown in Figure S14, with the corresponding J coupling values in Table S12.S12.

S3.6: X-Ray Crystallography of 6A
6A was prepared in methanol-d4 (0.6 mL) by reaction of [IrCl( 2 -2 -COD)(IMes)] (1) (5 mM) and A (50 mM) with 3 bar H2 and left at room temperature for 6 hours.At this point it was cooled to 278 K in a fridge and left for several weeks.Single crystals formed.A suitable crystal was selected and mounted on an Oxford-Diffraction SuperNova dual-source X-ray diffractometer equipped with copper and molybdenum sources and a HyPix-6000HE detector.Cooling to 110 K was achieved using an Oxford Instruments Cryojet.Using Olex2, the structure was solved with the SHELXT structure solution program using Intrinsic Phasing and refined with the SHELXL refinement package using Least Squares minimisation.Crystallography details are given in Table S13.The structure of 6A is shown in the main paper, Figure 2a.

S3.8: Characterisation of 4B
The sample described in S3.7 was then cooled in a dry ice acetone bath and 3 bar of pH2 was added to the NMR tube before being reintroduced into the NMR spectrometer (at 243 K).The sample took several minutes to reach 243 K.During this period, PHIP-enhanced hydride signals were also detected at -12.02 and -17.92, and -13.33 and -18.50 in agreement with the detection of [Ir(H)2(B)( 2 -2 -COD)(IMes)]Cl, 4B and [Ir(Cl)(H)2( 2 -2 -COD)(IMes)], 3. 4B was then characterised using thermally polarised 2D NMR methods.Its structure is shown in Figure S16 and its NMR resonances are detailed in Table S15.The characterisation of 3 was detailed in Section S3.4 S15.

S3.9: Characterisation of 6B
Upon warming the solution described in S2.8, 6B forms and was characterised at low temperature.Its structure is shown in Figure S17 and its NMR resonances are given in Table S16.S16.

S3.10: X-Ray Crystallography of 6B
6B was prepared in methanol-d4 (0.6 mL) by reaction of [IrCl( 2 -2 -COD)(IMes)] (1) (5 mM) and A (50 mM) with 3 bar H2 and left at room temperature for 6 hours.At this point it was cooled to 278 K in a fridge and left for several weeks.Single crystals formed.A suitable crystal was selected and mounted on an Oxford-Diffraction SuperNova dual-source X-ray diffractometer equipped with copper and molybdenum sources and a HyPix-6000HE detector.Cooling to 110 K was achieved using an Oxford Instruments Cryojet.Using Olex2, the structure was solved with the SHELXT structure solution program using Intrinsic Phasing and refined with the SHELXL refinement package using Least Squares minimisation.
The crystal gave relatively streaked reflections with evidence of twinning.This is believed to be the cause of high residual density peaks centred around the iridium.These results represent the best absorption corrections we were able to perform.Attempts to solve the crystal as a multi-component twin gave R1 and wR2 which were significantly poorer (5.79 and 18.98%, respectively) than modelling using a single component, although the peak and trough residual densities were improved, 2.3 and -2.1.Ir-H bond lengths were constrained to be 1.77 angstroms as allowing them to refine led to unfeasibly short Ir-H bond lengths.Crystallography details are given in Table S17.The structure of 6B is shown in Figure S18.

Table S20: 13 C NMR signal enhancements for A and B with the indicated sulfoxide co-ligands. The NMR signal enhancements are recorded by shaking a sample of [IrCl( 2 - 2 -COD)(IMes)] and A or B (50 mM) with either DMSO or DPSO (25 mM) in the indicated solvent with 3-bar pH2 for 10 seconds in a mu metal shield at 1 mG. Table S21: 15 N NMR signal enhancements for A and B with the indicated sulfoxide co-ligands. The NMR signal enhancements are recorded by shaking a sample of [IrCl( 2 - 2 -COD)(IMes)] and A or B (50 mM) with either DMSO or DPSO (25 mM) in the indicated solvent with 3-bar pH2 for 10 seconds in a mu metal shield at 6 mG.
A N NMR signal enhancements with DMSO were optimized further by variation of the polarization transfer field, substrate loading and shaking time.The effect of these variables is shown in Figure S19.

S5: Characterisation of sulfoxide-containing metal complexes involved in SABRE S5.1: 2D NMR characterisation of 8A
8A was prepared by reaction of 1 (10 mM), A (25 mM) and DMSO (25 mM) in methanol-d4 (0.6 mL) with H2 (3 bar) for a few hours at room temperature before being cooled to 245 K and characterised using 2D NMR.NOE connections were found from both hydride signals to IMes resonances at 7.21, 6.85, 6.79, 2.27 and 2.16, and the bound sulfoxide resonances at 2.77 and 3.14.The relative integral intensities confirmed these were due to inequivalent CH3 sites of the bound DMSO, rather than from two different ligated DMSO.Observation of their signals nOe to both hydries confirms that the IMes and DMSO are trans as no nOe peaks were observed between them.NOE from the hydride at -23.11 to the bound ortho signal of A at 8.76 confirmed the orientation of A cis to this hydride. 13C NMR resonances were assigned using HMQC.S22.

S5.2: 2D NMR characterisation of 8B
8B was prepared by the reaction of 1 (10 mM), B (25 mM) and DMSO (25 mM) in methanol-d4 (0.6 mL) with H2 (3 bar) over a few hours at room temperature, before being cooled to 245 K and characterised using 2D NMR.NOE connections were found from both hydride signals to the IMes resonances at 7.20, 6.88, 6.82, 2.29 and 2.18 and the bound sulfoxide resonance at 2.78 and 3.13.The relative integral intensities confirmed these were due to inequivalent CH3 sites of the bound DMSO, rather than from two different ligated DMSO.Observation of nOe to both hydries confirms that the IMes and DMSO are trans and no nOe were observed between them.NOE from the hydride at -23.12 to the bound ortho signal of A at 8.83 confirmed the orientation of A cis to this hydride. 13C NMR resonances were assigned using HMQC.S23.S24.

S5.4: 2D NMR characterisation of 9B
9B was prepared by reaction of 1 (10 mM), B (50 mM) and DPSO (25 mM) in methanol-d4 (0.6 mL) with H2 (3 bar) over a few hours at room temperature before being cooled to 245 K and characterised using 2D NMR.Characterisaton of 9B was challenging as when it was cooled to 245 K single crystals were formed in the NMR tube which affected the quality of the NMR data.Nonetheless, some 2D NMR data could be collected and this is shown in Table S25.NOE connections were found from both hydride signals to the IMes resonances at 6.96, 6.87, 2.24, and 2.32.NOE from the hydride at -23.08 to the bound signals of B at 8.89 and 8.22 confirmed the orientation of A cis to this hydride. 13C NMR resonances were assigned using HMQC.Signals for the bound sulfoxide and the imidazole part of the IMes ligand could not be located due to peak overlap and poor quality NMR data likely caused by crystal formation.The structure of 9B was confirmed by X-ray crystallography (see section S5.5).
The crystal structures for 6A, 6B and 9B indicate iridium-nitrogen bond lengths for the dissociating ligands of 2.238(2), 2.148(5) and 2.2344(19) Å.Consequently, these bond lengths do not follow the trend in H ¹ and suggests that other factors contribute to their values.Accordingly, substrate exchange could take an associative character, with other coordinating ligands such as methanol or even water playing a role.Any conclusive deductions will require a rigorous DFT study which is beyond the scope of this work.S25.

S5.6: PHIP time-courses for formation of 8A
The formation of 8A was monitored at 245 K using PHIP by recording a series of single-scan 1 H NMR spectra at 245 K immediately after addition of pH2 (3 bar) to a solution of pre-cooled 1 (5 mM), A (50 mM) and DMSO (25 mM) in methanol-d4 (0.6 mL).Example spectra are shown in Figure S25.S6: Ligand loss rates in 6, 8 and 9 The dissociation of A or B from either 6, 8 or 9 was measured using exchange spectroscopy (EXSY).This involved the selective excitation the ortho protons of bound A or B trans to hydride, followed by a variable delay time, before a 1 H NMR spectrum is recorded.Peaks are observed for excited metal complex, and as the delay time is increased a signal for free ligand that was previously bound to the metal centre becomes visible.The proportion of bound and free ligand as a function of delay time is measured and fitted to a kinetic model to extract a kinetic dissociation rate.The rate is found by minimising the difference between experimentally determined bound and free ratios, and those predicted by a kinetic two site exchange model.Errors were calculated using the Jack Knife approach.This involves sequentiually removing one data point, calculating the rate constant, and then taking an error of all of the rate constants.The enthalpy ( ± ) and entropy ( ± ) were determined by recording the dissociation rate, k, at different temperatures, T, and fitting the data to a linearized Eyring equation (equation 1) where is Boltzmann's constant, is Planck's constant, and is the ideal gas constant.Accordingly, a plot of ln against yields a straight line with a gradient of -D ± and a y axis intercept of ln + D ± , allowing calculation of the entropy and enthalpy.As the experiments measure the forward rate constant crossing the activation barrier, a factor of 2 in included to account for the equal probability of the symmetric transition state reacting onwards to give an exchange product, or going backwards to reform the starting materials.Samples used to determine these exchange rates contained 1 (5 mM) and A or B (45 mM) with H2 (3 bar) in dichloromethane-d2 (for exchange rates of 6) and 1 (5 mM) and either 1 (5 mM), A or B (15 mM) and sulfoxide (100 mM) with H2 (3 bar) in methanol-d4 or 1 (5 mM), A or B (80 mM) and sulfoxide (100 mM) with H2 (3 bar) in dichloromethane-d2 (for exchange rates of 8 or 9).Exchange rates for 9A could not be determined due to the inability to selectively excite A within 9A due to peak overlap (with the DPSO resonances).
C NMR signal enhancements for A and B at various loadings and in two solvents.The NMR signal enhancements are recorded by shaking a sample of [IrCl( 2 - 2 -COD)(IMes)] and A or B at the indicated loading and in the indicated solvent with 3-bar pH2for 10 seconds in a mu metal shield at 1 mG.Enhancements are for the free ligand signal.

Figure S1: 1 H
Figure S1: 1 H NMR signal T1 times for A (50 mM) at the indicated loading relative to 1 in dichloromethane-d2 at 9.4 T. Samples of different composition were prepared by fixing the concentration of A at 50 mM and adding the appropriate amount of 1.

Figure
Figure S3: 1 H NMR spectrum of 1 (23 mM) in the presence of A (46 mM)) in methanol-d4 at 298 K. Signals are assigned according to the resonance labels shown in Figure S4, with those shown in Black corresponding to 1, and those in Red to 2A.

Figure S6 :
Figure S6: Van't Hoff plot allowing H  and S  for the equilbirum between 1 (5 mM) and 2A (65 mM) in methanol-d4 to be calculated.

Figure
Figure S7: a)-b) Hyperpolarised 1 H NMR spectra recorded after addition of pH2 to an equilibrium mixture of 1 (5 mM) and 2A (50 mM) in methanol-d4 at 243 K.The resonances are labelled according to Figure 2 of the main paper.c) A thermally polarised spectrum of the same sample (not to scale).

Figure S8 :
Figure S8: Hyperpolarised 1 H NMR spectra recorded after addition of pH2 to an equilibrium mixture of 1 (5 mM) and 2A (50 mM) in dichloromethane-d2 at 243 K.The resonances are labelled according to Figure 2 of the main paper.The inset shows an expansion of the region at ca -24 ppm.

Figure
Figure S11: 1 H NMR spectra showing the resonances for 3 recorded after addition of H2 to an equilibrium mixture of 1 (5 mM) and 2A (50 mM) in dichloromethane-d2 at 243 K.Note the resonances for 3, labelled according to Figure S10) are weak in comparison to those of 2A (not assigned).

Figure S12 :
Figure S12: Structure of 6A, its NMR resonances are given in TableS11.

Figure S13: 1 H
Figure S13: 1 H NMR spectra showing the resonances for 6A recorded after addition of H2 to an equilibrium mixture of 1 (5 mM) and 2A (50 mM) in dichloromethane-d2 at 243 K.Note the resonances for 6A labelled according to Figure S12, are weak in comparison to those of 2A (not assigned).

Figure
Figure S14: 1 H NMR spectra of 6A between 238 K and 268 K in dichloromethane-d2 which show the temperature dependence of the hydride chemical shifts, resulting in the AX spin-system changing to AB upon cooling.Associated J coupling values are presented in TableS12.

Figure S15 :
Figure S15: Structure of 2B, its NMR resonances are given in TableS14.

Figure S17 :
Figure S17: Structure of 6B, its NMR resonances are given in TableS16.

Figure S21 :
Figure S21: Structure of 8A, its NMR resonances are given in TableS22.

Figure S22 :
Figure S22: Structure of 8B, its NMR resonances are given in TableS23.

S5. 3 :
2D NMR characterisation of 9A9A was prepared by the reaction of 1 (10 mM), A (50 mM) and DPSO (25 mM) in methanol-d4 (0.6 mL) with H2 (3 bar) over a few hours at room temperature, before being cooled to 245 K and characterised using 2D NMR.NOE connections were found from both hydride signals to the IMes resonances at 6.93, 6.84, 2.22 and 2.19 and the bound sulfoxide resonances at 7.39 and 7.44.Observation of nOe to both hydries confirms that the IMes and DPSO are trans and no nOe were observed between them.NOE from the hydride at -23.08 to the bound signals of A at 8.77 and 8.09 confirmed the orientation of A cis to this hydride.13C NMR resonances were assigned using HMQC.

Figure S23 :
Figure S23: Structure of 9A, its NMR resonances are given in TableS24.

Figure S24 :
Figure S24: Structure of 9B, its NMR resonances are given in TableS25.

Figure S25 .
Figure S25.Series of single scan 1 H NMR spectra recorded with 45 o pulses a) immediately after pH2 (3 bar) to a solution of pre-cooled 1 (5 mM), A (50 mM) and DMSO (25 mM) in methanol-d4 (0.6 mL) and shaken for 10 seconds at 6.5 mT.b) is recorded 10 seconds later and c) 2.5 minutes later.d) is recorded after shaking with fresh pH2 and is recorded 10 minutes after a).e) is recorded 1.5 minutes after d).The species labelled S is likely an analogue of 5 where L is replaced with DMSO.

Figure S26 .
Figure S26.Linearised Eyring plot linking the rates of substrate dissociation as determined by EXSY at the specified temperature.The gradients and intercepts are used to calculate the entropy and enthalpy of ligand dissociation.